# Strella Docs > Documentation for Strella, an AI-moderated user research platform. ## Developer & API reference The pages below are the product Guides (how to use Strella). For the REST API itself — endpoints, request/response schemas, auth — use these machine-readable sources: - [OpenAPI 3.1 specification](https://app.strella.io/api/public/v1/openapi.json): every endpoint, schema, and authentication requirement for the Strella public REST API (base path `/api/public/v1`, API-key auth). - [API reference](https://app.strella.io/docs/api): the interactive Scalar reference rendered from that spec. # API, MCP & AI assistants Source: https://app.strella.io/docs/product/account/api-mcp-assistants Strella offers three ways to connect from outside the app: - **REST API**: for automating Strella from your own code (creating studies, fetching results). - **MCP server**: lets AI assistants do real work with your research (build studies, run analysis, and pull quotes, themes, and results) by connecting over the Model Context Protocol. - **Pre-built connectors**: turn-key integrations for Claude, ChatGPT, and Figma Make so AI assistants can read your research. The REST API authenticates with an API key you generate here. The MCP server and the connectors authenticate with your own Strella sign-in, so they need no key. ## Steps The Developer page lists API keys and the MCP setup link. Only organization admins can see this page. Click **+ New API key**. Give it a name (so you can identify it later) and pick scopes: - **Read**: fetch studies, participants, interviews, themes. - **Write**: create studies, update settings, manage participants. Strella shows the key plaintext **once**. Copy it now: you can't see it again. Send requests with `Authorization: Bearer `. The full reference is at [https://app.strella.io/docs/api](https://app.strella.io/docs/api). It lists endpoints, request bodies, and response shapes. Strella shows no rate-limit usage in the app. The MCP server lets AI assistants work with your Strella data. [Connect an AI client](/mcp/connect) walks through each one: - [Claude](/mcp/connect#claude-web-desktop--cowork): web, desktop, and Cowork, which all share one Connectors flow. - [ChatGPT](/mcp/connect#chatgpt): from ChatGPT's plugin directory. - [Figma Make](/mcp/connect#figma-make): pull Strella research into Figma Make. - [Claude Code](/mcp/connect#claude-code): one command in your terminal. - [Cursor](/mcp/connect#cursor): a config file in your editor. - [Any other client](/mcp/connect#other-clients): add Strella as a custom connector. MCP setup does **not** use an API key. You sign in with your Strella account through a secure OAuth redirect, and the connection has only the permissions you already have. On the Developer page, click the trash icon next to any key to revoke it. Revocation is immediate, and calls with that key return 401. Your AI clients have access to your research. Ask Claude "what did participants in my last study say about pricing?" and it can answer from the actual data. ## Tips - **One key per use case.** If you're connecting both Claude and a CI pipeline, generate two separate keys so you can revoke one without breaking the other. - **API keys are organization-scoped.** Any teammate with admin access can create and revoke them. The keys themselves access every study in the organization. - **Use cases** for the connectors: [Use cases & example prompts](/mcp/use-cases) has worked examples: competitive analysis from past studies, pulling quotes into a doc, etc. ## Related Track API key activity in the org. Who has access to generate keys. --- # Activity logs Source: https://app.strella.io/docs/product/account/audit-logs Activity logs record actions across your organization: share links created, settings changed, participants imported, API keys generated. Available to organization admins. ## Steps Activity logs are for **organization admins** and are aimed at **compliance and access reviews**. Open **Settings → Advanced** and click **View activity logs**. (If you don't see it, your organization may need it enabled, so contact your account team.) The logs page lists events newest first, with filters for actor, project, and event type. Click **Filter** above the table and add one or more conditions (each can match several values at once). Type in a condition's list to search it down before picking: - **Actor**: show only actions taken by specific teammates. - **Project**: show only activity tied to one or more studies. - **Event type**: show only share-link creates, only setting changes, etc. Use the **⋯** menu on a row and choose **View log** to see the full event payload (the actor, the page or resource, and the raw properties) with a one-click copy for sharing or ticketing. Hover a row's avatar to see the acting teammate's name and email. ## What's logged Events span the actions teams care about: studies published, closed, or deleted; share links created and edited; settings changes; participants imported, invited, or rejected; API keys created or revoked; teammates added or removed; interviews excluded from analysis; and a long tail of page-view events. The **Event type** filter lists every event type. Open it and search to see what is available for your organization. ## Things to know - **The default view is filtered to page-view-only events.** Toggle other event types on to see them. The filter helps performance in busy organizations. - **Strella staff actions are excluded** from what customers see in the logs. If a Strella team member helped debug something in your org, that won't appear. - **Logs can be deleted org-wide** by an admin. This is for GDPR-style data deletion requests; once deleted, history is gone. - **Activity logs and public API request logs are separate records.** Activity logs cover the actions listed above. Requests made through the public API are logged separately, and the **Activity logs** switch does not turn that logging off. - **The activity log is reviewed here, not through the API.** Strella's public API does not expose an activity-log feed, so this record is read in the app rather than retrieved through the API. See the [API guide on logging and auditing API access](/docs/api#tag/Logging-and-Auditing-API-Access). ## Related Manage who has access to the things the logs track. API key creation and revocation appears in the logs. --- # Billing & credits Source: https://app.strella.io/docs/product/account/billing Strella has three ways to pay: pre-purchased **credits** (a wallet), **pay-as-you-go by card** at checkout, or an **invoice** (enterprise). Card payments go through a one-time checkout each time, and Strella doesn't store a card on file. ## How they work | Method | How it works | Who has it | | ----------- | ------------------------------------------------------------------------------------------------------- | ---------------------------------------------- | | **Wallet** | Pre-purchased credits. Strella deducts them when you place a panel order or pay out self-recruited participants. | Available to every organization. | | **Card** | Pay by card at a one-time checkout when you place an order. No prefunding, no card stored on file. | Available to every organization. | | **Invoice** | Choose invoice at checkout, and Strella bills that order on an invoice instead of charging you up front. | Enterprise customers. Contact sales to enable. | If your wallet has a balance, Strella draws from it first, and a card checkout covers anything the wallet can't. A wallet only holds funds after you purchase credits, so new organizations start at $0. ## Steps The billing page shows your current wallet balance and transaction history. Enter an amount and click **Purchase** to add credits to your wallet. Strella sends you to a one-time card checkout, where it adds a small processing fee. Strella adds the credit to your wallet as soon as the payment clears. Wallet credits don't expire (until you close the organization). The history table shows every charge and credit: date, amount, category (wallet spend, card payment, incentive, refund, etc.), and the reference order ID for panel orders. Each row carries its own details, so you don't need to click in. **Participants** shows how many participants that charge paid for, so you can see what an order bought next to what it cost. The column counts the participants you bought, not the participants who finished. An extension row shows only the participants it added, and a refund row shows only the unused participants returned, so the rows of one order may not sum to its total. Charges that don't buy participants, like a credit top-up, leave it blank. An order paid by invoice appears as soon as you place it, before you pay the invoice. If a charge you expect isn't listed, open the **⋮** menu above the table and choose **Resync all transaction history**. Strella rebuilds the list from your payment records and shows a progress bar until it finishes, which can take a few minutes. Anyone in the organization can resync. The option is unavailable while a resync is already running. Strella applies charges when you place a panel order or pay out self-recruited participants. It draws them from your wallet, runs them through a one-time card checkout, or bills them on an invoice. ## Things to know - **Participant incentives are billed in addition to interview credits.** A panel order's total includes both. - **Refunds appear as separate rows.** Strella does not reverse the original charge. It adds a refund row, shown in green with a **+** because the money comes back to you. - **Balance refreshes on page load.** If you've topped up in another tab, refresh to see the new balance. - **Invoiced billing isn't customer-self-serve.** Contact your account team to set it up. ## Related Where most billable charges originate. Each organization has its own billing. --- # Delete projects Source: https://app.strella.io/docs/product/account/delete-projects Deleting a project removes it from your dashboard along with its participants, interviews, transcripts, and share links. Every delete asks you to retype something exactly, so a stray click cannot remove a study that is still running. ## Delete one project On the projects list, hover the row and click the **...** menu at the end of it. The confirmation dialog lists what deleting affects. Type the project's name exactly as it appears on the row. The red **Delete** button stays greyed out until it matches, so you can always back out by closing the dialog. ## Delete several projects at once Tick the checkbox on each row you want to remove. A bar appears at the bottom of the list showing how many are selected. Open the **...** menu on that bar and choose **Delete**. You can also use the **...** menu on any selected row, which acts on the whole selection. Typing every name would be impractical, so a multi-project delete asks for the phrase **Delete N projects** instead, where N is the number selected. Deleting twelve projects asks you to type `Delete 12 projects`. The phrase states the count, so if the number in it is larger than you expected, your selection is bigger than you meant it to be. Close the dialog and check the selection before continuing. To clear a selection without deleting anything, press `Escape`, click the **X** on the selection bar, or click an empty part of the page. ## What deleting does - The project disappears from your dashboard, from search, and from every filter and workspace it belonged to. - Strella deletes its interviews, transcripts, and questionnaires too. - Its share links stop working. Anyone who opens one cannot join. - Teammates in your organization lose access to it at the same time you do. Deleting does not close an open Recruitment Order or settle participant incentives. If the project is still recruiting, close the order and pay out anything outstanding first. Otherwise recruitment keeps running against a project you can no longer see. ## Getting a project back There is no undo button, and a deleted project cannot be restored from the dashboard. It is hidden rather than erased, so contact support if you delete something you need back, and the sooner the better. Strella handles permanent erasure as a separate, deliberate process for data-removal requests. It is not what the **Delete** button does. ## Alternatives to deleting If you want a finished study out of the way but intend to keep its recordings and transcripts, archive it instead. Archiving hides a project from your working lists while keeping everything in it, and you can unarchive it later from the workspace's Archive view. See [Teammates, organizations & workspaces](/account/teammates-and-workspaces) for how workspaces and archiving fit together. --- # Connect Figma Source: https://app.strella.io/docs/product/account/figma Participants open your Figma prototype live during the interview and move through its screens. Connect Figma and Strella also records where they click, and turns that into a heatmap you can read in Results. A Prototype step works whether or not you connect Figma. Paste the prototype link and participants can explore it in the call either way. Connecting adds the screen list, start and goal screens, and click heatmaps. If you skip it, the prototype still shows, and no clicks are recorded. ## Before you start You'll need: - A Figma account with access to the prototypes you want to use. - The prototype **published** (not just shared) and set to **Anyone with the link can view**, so Strella can render it and participants can interact with it. ## Steps Click **Integrations** in the home sidebar to open your integration settings, which list each service you can connect. Strella opens Figma's authorization screen. Sign in if you aren't already. Figma asks you to grant Strella access to view your prototypes. Confirm. The connection belongs to you inside the organization you are in when you connect. Each teammate who wants click heatmaps connects their own, and a connection made in one organization does not carry over to another organization you belong to. **+ Add element → Prototype** in [Build](/build/build-screen) lets you paste any prototype URL you can access in Figma. With Figma connected, the import also includes the prototype's screen list. See [Add a stimulus](/build/stimulus) for the rest. Your studies in this organization capture clicks on Figma prototypes. The connection stays until you disconnect it, until your Figma session expires, or until you connect Figma in another organization. ## Tips - **Each teammate connects their own Figma.** If a coworker authors the prototype steps, they need their own connection, not yours. - **One organization at a time.** Connecting Figma in another organization disconnects this one. Prototype steps here keep working, but they stop recording clicks until you connect again. - **If a prototype that recorded clicks stops recording them**, reconnect Figma. The connection can expire without notice, and a study run in the meantime has no heatmap to show. - **For Figma Make connections** (the AI design assistant), see [API, MCP & AI assistants](/account/api-mcp-assistants). ## Related Drop a connected prototype into an interview. --- # Media links and who can open them Source: https://app.strella.io/docs/product/account/media-links A project holds two kinds of file, and Strella treats them differently. The materials you show a participant (stimulus images, consent forms, terms documents, your brand assets) open for anyone holding the link. The people who read them have no Strella account: someone sitting in an interview, a panelist reading a non-disclosure agreement on a recruiting vendor's site, a participant on a phone app. There is nothing for them to sign in to, so the link itself is the access. Everything else asks the person to be signed in to the workspace that owns the study. That covers a participant's uploaded photo or video answer, the session recording, and files you attach to a project chat. Anyone else gets a "not found" page, including a teammate signed in to a different workspace. A "not found" page is what a refused request looks like. It is deliberately the same page you would get for a file that does not exist, so nobody can use a link to work out whether a file is there. ## Deleting a project Deleting a project switches off the media that needs a sign-in, for everyone, straight away: participant answers, session recordings, and your chat attachments. That includes the links inside a results export somebody downloaded months ago. The materials you show participants work differently. A stimulus image, consent form, terms document, or brand asset keeps opening for anyone holding its link after the project is deleted. This is deliberate. Duplicating a project leaves the copy reading the original's files, so switching those files off would blank the stimuli of a study you are still running, mid-interview. If a document already circulating needs to stop opening, ask Strella to erase it. Erasure removes the stored file itself, which deleting the project does not. Restoring the project brings back the media that needs a sign-in. Deleting is reversible, and so is this. ## Moving a project to another workspace The project's media moves with it, all of it: participant answers, chat attachments, and session recordings alike. The new workspace can open everything, and the previous workspace can open none of it. ## Exported results A results export contains full links to these files. The links keep working, but only for someone signed in to the workspace that owns the study. Sending the spreadsheet to a stakeholder outside it does not give them the media. They also stop working if the project is deleted. ## The address in your browser bar Opening one of these files sends you on to a temporary address at the storage provider. You will not notice this in normal use, because opening the file again simply issues a fresh one. It only shows up if you copy the address out of your browser bar after the redirect and try to use that copy the next day. Copies of files that need a sign-in stop working after twelve hours. For interview materials, which have no sign-in check anyway, the copy lasts a week. Control how much participant information each teammate's AI tools and API keys can read. Invite teammates and organize projects into workspaces. --- # Participant data access Source: https://app.strella.io/docs/product/account/participant-data-access Participant records hold personal information: names, emails, profile photos, demographics, recordings, and survey and screener responses. Participant data access lets an admin choose, per teammate, how much of that information their AI tools and API integrations are allowed to read. This setting governs the **Strella API and AI assistants (the MCP connector)** only. It does not change what a teammate sees when they sign in to Strella directly. Anyone who can open a project in the dashboard sees its participants in full. ## What you can control Access is set field by field, so you can grant exactly what a teammate's tools need and nothing more: - **Name**: none, first name only, or full name - **Email** - **Profile photo** - **Demographics**: gender, age, race and ethnicity, education, household income, job title, seniority, company size, and location, each on its own - **Recordings** - **Survey responses, screener responses, and transcript text** When you turn a field off, Strella leaves it out of everything that teammate's AI assistant or API key returns. ## Data that would reveal a hidden field Some data reveals a field without containing it, so the same settings cover that data too: - **Filtering by age** needs **Age**. The participants a filter returns show who falls in the age range, so without Age, Strella rejects the request. - **Searching by name** follows **Name**. With first name only, a search matches first names only, so a match never confirms a last name. With name hidden, Strella rejects a name search. - **Quotas** need every demographic field. A quota usually names a group, such as "Women 25-34", so with any demographic field off, Strella hides a participant's quota and rejects a change to it through the API. - **A Saved Reel's public share link** needs **Recordings**. Anyone with the link can watch the reel, so without Recordings, Strella hides the link and rejects a request to create one. A link a teammate with access already shared keeps working. When Strella rejects a request for one of these reasons, the response says which field the teammate's access is missing. ## Who manages access Admins manage participant data access for everyone in the organization, including themselves. A teammate cannot raise their own access, and a non-admin cannot reach these settings at all. An admin can grant themselves or anyone else full access. To keep that accountable, **Strella writes every change to the audit log** with who made it, who it applied to, and what it changed. See [Audit logs](/account/audit-logs). ## The default Until an admin sets it, teammates can read participant information in full through their AI tools and the API. Restriction is something an admin turns on, not a default a teammate has to lift. An admin opens **MCP access** in the Developer section of the organization's account settings. Pick the member whose access you want to change. Turn fields on or off, or set name to first-name-only. Save. A change to a teammate's own access applies on their next request through the API or an AI assistant. A change to the organization default can take up to a minute to reach the teammates who use it. ## API keys An API key carries its own access, set from its scopes when you create it, independent of any person. A key can be limited to a single field, such as email only or one demographic, so an integration reads only what it needs. See [API, MCP & AI assistants](/account/api-mcp-assistants) to create and scope keys, and the [API guide on participant PII access](/docs/api#tag/Controlling-Participant-PII-Access) for the field-level scopes. Create API keys, connect AI assistants, and manage integrations. Review who changed access and when. --- # Project defaults Source: https://app.strella.io/docs/product/account/project-defaults Project defaults are the settings a new project starts with. Set them once and every new study your team creates begins already configured: the moderator voice, recording and device rules, language, appearance, and the rest. Available to everyone on your team. Only the settings that make sense for every study are covered. Per-study details (the project name, research goals, close date, and participant limits) are never part of the defaults, since those belong to one study, not all of them. ## Steps Open your organization **Settings** and select **Project defaults** under **Team**. Click **Configure** to open the settings. The Configure panel is the same settings editor you use inside a study, showing only the parts that apply to every project: - **Appearance**: logo, brand colors and font participants see. - **Interview setup**: language, device and recording rules, privacy, and accessibility. - **AI moderator**: the moderator voice, behavior, and pronunciation guide. Click **Save**. Every new project then starts with these settings. ## How defaults apply - **New projects only.** Defaults apply the moment you create a new study from your dashboard. Existing studies keep their own settings and are never changed. - **Only your changes are saved.** A setting you leave at its standard value isn't stored as a default: it simply follows the standard Strella default. Change a setting and it becomes your team's default for that field. - **Still editable per study.** Defaults are a starting point. Open any study's settings and change anything you like without affecting the defaults or other studies. ## Things to know - **Open to your whole team.** Everyone on your team can view and change project defaults. - **Changing a default doesn't change existing studies.** Updating project defaults affects studies you create afterward, not ones that already exist. - **One set of defaults per organization.** Everyone on your team creates new projects from the same defaults. ## Related Manage the teammates who share your organization's project defaults. Find organization settings next to project defaults. --- # Teammates, organizations & workspaces Source: https://app.strella.io/docs/product/account/teammates-and-workspaces Your top-level container in Strella is the **organization**. Inside an organization you can have **workspaces** (groups of studies) and **teammates** (people who can access studies). ## Steps Only an admin of your organization can invite new teammates. Check **Settings → Members**: if you can see an **Invite** button there, you're an admin and can go ahead. If you don't see one, ask an admin to add the person for you. From your organization sidebar, open the org switcher. Select **Manage organization → Members → Invite**. Enter their email. They get a sign-up link. Every teammate in your organization can see every study in the organization. There's no per-study or per-workspace access control, so inviting someone as a teammate is what gives them access to your studies. If you belong to more than one organization, use the org switcher in the top-left to switch between them. Each organization has its own studies, billing, and integrations. From the home sidebar, click **Workspaces** to see every workspace in your organization. Click **Create workspace** and give it a name (e.g. "Pricing research," "Onboarding studies"). You can assign new studies to the workspace. A study belongs to exactly one workspace or to no workspace (the default "no workspace" bucket). On a study, **Settings → Project overview → Workspace** picks which workspace it's in. Move freely: moving doesn't lose any data. Inside a workspace, create folders to group its studies further (e.g. by quarter or client). Open the workspace and click the **folder** icon next to **Create project**, then name the folder. The folder appears as a card above the study list. Click it to see only that folder's studies. Move a study into a folder from its **⋯** menu on the workspace page (**Move to folder**), or create a new study directly inside a folder. A study belongs to one folder or none, and folders don't nest (no folders inside folders). Moving a study to a different workspace clears its folder, since a folder lives in one workspace. A folder must be empty before you can delete it. Deleting a folder never deletes its studies, which stay in the workspace. Archive a study to remove it from the main list without deleting it. Use the study's **⋯** menu and choose **Archive**. Archived studies go to the workspace's **Archive**. Open it from the Archive card above the study list and choose **Unarchive** to bring a study back. Archiving keeps every recording, transcript, and setting. Hover a workspace in the sidebar and open its **⋯** menu for rename and delete options. You can only delete a workspace once it's empty (no studies in it). Move or archive its studies first. Deleting a workspace never deletes studies. ## Tips - **Use workspaces by program, not by team.** "Pricing research" stays accurate after a team reorganization. "Mei's studies" does not. - **Every workspace is listed in the home sidebar**, each with its study count. The list scrolls when it outgrows the space, and **Workspaces** shows them all as cards. - **There's no per-workspace or per-study permission**: every teammate in the org sees every workspace and every study. If you need isolation between teams, use separate organizations. - **Not sure if you're an admin?** Open **Settings → Members** and look for the **Invite** button, or check whether you have an admin-only "Manage organization" area. If you don't, ask an admin to invite people on your behalf. ## Related Each organization has its own billing. Track teammate actions within an organization. --- # AI chat best practices Source: https://app.strella.io/docs/product/analyze/ai-chat-best-practices The [AI chat](/analyze/ai-chat) reads every completed interview and answers in plain language. What you get back depends on how you ask. Use the practices below to get sharp, well-sourced answers, and to turn a good thread into a chart, a report, or a highlight reel. New to the chat? Start with [Use Strella's AI chat](/analyze/ai-chat) for where to find it and how it works. The **suggestion cards** on the chat home (identify key themes, create a chart, make a highlight reel, generate a report) are good starting points when you're not sure how to phrase a prompt. ## Prompt for better answers - **Say what you want to learn, not just a keyword.** "What made people hesitate at checkout?" gets you a synthesis; "checkout" makes the AI guess what you meant. A question gives it a target. - **Ask one thing at a time, then follow up.** Each turn builds on the last, so start narrow and drill in ("now just the SMB customers," "show me the negative ones," "why did they say that?") instead of packing five asks into one prompt. - **Name the segment when you have one.** The chat reads participant tags and screener answers. For guides with variants, it reads which variant each participant was assigned. For graded studies, it reads who passed or failed each task. Use those in the prompt: "How did first-time users differ from power users?", "How did people who saw the annual pricing variant react compared with the monthly one?", or "Why did people fail the sign-up task?" - **Check who reached the variant.** Strella assigns a participant a variant when their interview starts, so someone who left early may never have seen it. The chat confirms from the interview itself. Ask "did everyone on that variant actually get to it?" when the difference matters. - **Anchor to your research objectives.** Asking in the language of what you set out to learn ("Which of our objectives did the interviews actually answer?") keeps the analysis on the questions that matter to the study. - **Ask for the evidence.** Every answer cites quotes inline, and you can click one to jump to that moment in the source interview. If a claim feels thin, ask "which interviews is that from?" or "show me the quotes." - **Push back when something looks off.** The AI responds to challenges: "that seems high, double-check," "are there counterexamples?," or "you missed the Android users." Treat it like an analyst you can question. - **Set the scope with the study filter.** Narrow to one study for depth. Widen across studies to spot patterns that span your research. In very large organizations a broad query may sample interviews rather than read every one, so narrow the filter when you need thorough coverage. ## Make a chart - **Ask for the cut you want.** "Chart how participants rated the onboarding flow" or "break down mentions of price by customer size." You can also click the **Create a chart** suggestion card and refine from there. - **Charts work best on countable data** such as ratings, categories, coded open-ended answers, or task pass/fail. Open-ended sentiment is better explored as themes in [analysis](/analyze/browse-analysis). - **Iterate on the view.** Follow up with "make that a percentage," "split it by segment," or "only completed interviews" to reshape the chart. ## Generate a report When a thread is worth writing up, ask **"build a report from this."** Scoping it tightly ("a report on pricing feedback") produces a focused writeup instead of a sprawling one. The chat creates the report and opens it, and you refine the wording by asking the chat for changes. Full details: [Build & share a report](/share/reports). ## Create a highlight reel Ask **"make a highlight reel of the strongest moments about X."** The chat compiles the clips and opens the reel. **Save** it to pin it, so it stops changing as new interviews come in. Keep reels short (four to six clips) and mix in a dissenting reaction so they read as honest. Full details: [Create & share a highlight reel](/share/highlight-reels). **Say how long it runs, or how many clips it holds**, in the same sentence: "a two-minute reel of pricing objections," "keep it under 90 seconds," "five clips on the checkout flow," "no more than three moments." The chat treats a number of clips as exact. It aims at a length and may come out slightly over or under it. It picks whole moments and never cuts into the middle of an answer, so a reel asked for two minutes comes back near two minutes. Trim individual clips in the reel editor when you need an exact runtime. Twenty-five clips is the most any reel holds. **The chat tells you when it cannot hit what you asked for.** It picks only from moments already found in your conversation, so asking for five clips when three of them match gets you three clips and the reason why. The same goes for a length no combination of moments can reach. The chat never claims a reel met a length or a count it missed. ## Example prompts | If you want to... | Try asking | | ---------------------------- | ------------------------------------------------------------------------ | | Find a pattern | "What are the top reasons people abandoned sign-up?" | | Compare two segments | "How did mobile users react differently from desktop users?" | | Quantify and chart something | "What share of participants brought up price? Chart it." | | Chase down the evidence | "Show me the quotes behind that theme." | | Turn a thread into a writeup | "Build a report on the onboarding findings." | | Make a shareable reel | "Create a highlight reel of the strongest reactions to the new pricing." | | Cut a reel to a length | "Make a two-minute highlight reel of the checkout complaints." | ## Things to know - **Only completed, non-excluded interviews are in scope.** The chat leaves drafts, in-flight interviews, and interviews excluded from analysis out of every answer, chart, report, and reel. - **Answers cite their sources.** Click any quote to jump to that moment in the source interview's transcript. - **Reports and reels aren't edited inline.** Ask the chat to revise a report, and edit a reel's clips from the reel itself. ## Related Where the chat lives and how it works. The themes, sentiment, and charts the chat draws on. Turn a conversation into a polished writeup. Turn a conversation into a shareable video. --- # Use Strella's AI chat Source: https://app.strella.io/docs/product/analyze/ai-chat ## Two chats, same brain | Chat | Scope | Where to open | | ---------------------- | -------------------------------------- | --------------------------------------------------------------------- | | **Project chat** | One study's interviews | **Results → AI chat** inside the study | | **Multi-project chat** | Several studies in your organization | The **Chat** icon in your home sidebar | Either way, the **study filter** controls which studies are in scope, and you can change it anytime, even mid-conversation. ## Steps For one study, go to **Results → AI chat**. For several, click **Chat** in the home sidebar. Use the **study filter** to choose which studies are in scope. You can adjust it at any time. ![The AI chat home screen: a "What can I help with?" prompt with suggestion cards to identify key themes, create a chart, make a highlight reel, or generate a report.](/docs/analyze/ai-chat.png) Examples that work well: - "What did people say about onboarding?" - "Find quotes about price sensitivity." - "How did participants who use Tool X differ from those who use Tool Y?" - "Summarize what's blocking conversion." - "Who failed the checkout task, and why?" The AI returns a synthesis with inline quote citations. Strella shows a sentence's sources in one chip: the first participant's name, plus **+N** for the other participants quoted. Click the chip to page through each quote in full, and click a quote to jump to that moment in the source interview. ![Asking the AI chat a question and getting a synthesis with inline quote citations.](/docs/analyze/ai-chat-review.gif) Each turn builds on the last. "Now show me the negative ones" or "what about for SMB customers" picks up the context you've established. Ask for a question's results, and the chat draws the chart the Analysis page shows for that question: "show the answers to the pricing question", "what was the sentiment on checkout?" or "chart the screener question about bone broth". - **Answers** for a multiple choice, slider, rating, NPS, ranking, matrix, card sort, or tree test question, or for a screener question - **Sentiment**: how participants felt about the question - **Emotional responses**: the emotions read from the recordings of participants who consented, shown in place of sentiment, as on the Analysis page - **Pass rate** for a task - **Time on task**: how long participants took on a task, with the median and the Outliers toggle, as on the Analysis page The counts match the Analysis page with no filters applied, because the chat uses the same counts. To see the question with the rest of its results, click **Open on Analysis page** on the chart's card. When a thread of conversation is the basis for an artifact, ask: "build a report from this" or "make a highlight reel of the strongest examples." The chat creates the artifact and opens it. Refine a report by asking the chat for changes. Edit a reel's clips from the reel itself. A report can include charts of your study's AI codings and the highlight reels saved in it. Ask for them: "add a chart of the reasons people churned." If the study has no coding or saved reel to embed, the chat says so. For a reel, you can name a length or a number of clips in the same ask: "a two-minute reel of the checkout complaints," or "five clips on pricing." ## Things to know - **Only completed, non-excluded interviews are in scope.** Drafts, in-flight interviews, and interviews you've excluded from analysis aren't there. - **Quotes cite a source moment.** Click any quote to jump to the spot in the source interview's transcript. - **Ask for exact words.** Ask for an exact word, name, or phrase ("find every mention of 'free trial'") and the chat searches the transcripts for it word for word, rather than summarizing, and names who said each one. - **Ask for a report on one group.** Ask for "a report on just the renters", and the report counts and quotes only the participants who gave that screener or survey answer. Name the question if the chat can't find it. - **Ask how many.** "How many participants mentioned the free trial?" counts every match across the interviews, not only the quotes the chat shows you. When there are more than 100 matches, the chat shows a sample and says how many more exist. - **Ask how Strella works.** For a question about a Strella feature, such as quotas or screeners, the chat looks the answer up in these help docs. When the docs don't cover it, the chat says so rather than guessing. - **Task pass/fail is in scope.** For studies that grade tasks, the chat knows who passed or failed each task and the on-screen reason behind each verdict, so you can ask "who failed the signup task?" or "why did people fail checkout?" Studies without task grading simply have nothing here. - **A question's Analysis chart stays in the chat.** A report can't include it. Download a multiple choice, slider, rating, sentiment, or pass rate chart from the chart's menu as an image or as an .xlsx spreadsheet, the same spreadsheet the question's card on the Analysis page downloads, with every participant's answer, sentiment, or pass/fail result by name. A screener chart and the NPS, ranking, tree test, and time on task charts download as an image only. The Emotional responses, matrix, and card sort charts have no download. The chart's labels are in English, whatever the study's analysis language. - **Write-in answers share one Other bar.** To read each write-in, open the question on the Analysis page. - **Some questions have no chart in the chat.** A screener question answered in typed text gets none, and neither do studies you moderate yourself. Ask for a chart a question doesn't have, and the chat names the charts it does have. - **Adjust scope with the study filter.** Add or remove studies from a multi-project chat at any point. The next answer uses the updated scope. - **For very large scopes**, the chat reads every interview's summary but only a sample of quotes, and the answer says when it draws on a sample. To read more quotes from specific participants, ask about them in a follow-up message. When a question covers too much data to answer, the chat asks you to narrow it to fewer studies or start a new chat. ## Related How to prompt for sharper answers, charts, reports, and reels. Every place AI helps across a study, and how this chat fits in. What the chat can hand off into. The other artifact the chat can spawn. --- # Browse analysis Source: https://app.strella.io/docs/product/analyze/browse-analysis Analysis groups responses into themes, scores sentiment, and pulls representative quotes across every completed interview. Each quote links back to the source moment. ![Results → Analysis grouped by objective: the Objective/Question toggle, an objective summary, and highlight reels with participant counts.](/docs/analyze/browse-analysis.jpg) ## How sentiment, themes, and coded categories work - **Themes** are patterns the AI detected across multiple interviews, e.g. "users distrust the pricing page" or "screen share permissions confuse Mac users." Each theme lists the participants who exhibit it and pulls quotes. - **Sentiment** is shown per theme as **Positive**, **Neutral**, or **Negative**. - **Coded categories** are how Strella turns open-ended answers into something countable. The AI reads every participant's response to a question and assigns each one to a mutually-exclusive, collectively-exhaustive bucket, the same kind of work a researcher would do by hand when "coding" open-ended survey responses. ## Steps Two pills at the top let you switch between **By objective** and **By question**. - **By objective**: themes are organized under each of your research objectives. Best for storytelling around what you set out to learn. - **By question**: themes are organized under each interview question. Best for diving into a specific moment ("what did everyone say about X?"). At the top of each section, the AI writes a one-paragraph synthesis of what participants said. Below it, each theme appears with its participants, quotes, and an auto-generated highlight reel. Each theme shows: - The theme name and a one-line description. - How many participants exhibit it (as a count and percentage of total). - Aggregate sentiment. - Representative quotes: click any quote to jump to that moment in the source interview. The filter button (top right) narrows every number, theme, and quote on the page to the participants you pick. The same filter bar appears on the Interviews list. Stack as many rules as you need. Each option shows a live count of how many participants it matches. Strella leaves out options nobody in the study has, such as a share link no one used, and any filter with nothing to pick. Suggested filters mirror how you recruited: - **Audience**: the share link or panel order a participant came through. When two share a name, each shows its source in brackets, such as "Spring wave (Respondent)". - **Quota**: the quota cell they filled. To see which cells still need participants, check your quotas on the Recruit tab. - **Participant name**: find one person fast. More filters cover who participants are and how the interview went: demographics, screener answers, in-call survey answers, tags, duration, interview quality, variant assignment, task results, recruitment source, and **Device** (Desktop web, Mobile web, or Mobile app). Number answers, such as an NPS score or a rating, default to **is one of**, so you tick the scores you want the same way as a multiple choice answer. The list also offers **Greater than**, **less than**, and **between**. The page link stores your filters, so copying the URL shares the exact filtered view with a teammate. Option counts show how many participants each choice would select. A participant with no value for a field appears under no option (for example, an imported interview carries no device), so a field's counts can add up to less than the participant total. On the **By question** view, **Compare** (beside **Filter**) splits every chart into two or more groups of participants you define, so you can read them next to each other instead of flipping between filters. 1. Click **Compare**. Segment **A** is ready to set up. 2. Give each segment its rules, using the same fields as the filter. Strella labels each segment from its rules, for example **Mobile** or **Not Enterprise**. A participant must match all of a segment's rules. Click **And** between two rules to switch it to **Or**, so anyone matching any one rule is in the segment. Each segment shows how many participants it holds as you build it. A segment with no rules is everyone. 3. Click **Add segment** for more, or **Duplicate** in a segment's **⋮** menu to copy it. The same menu deletes the segment. The trash icon on a rule removes just that rule, so you can copy a segment and change only what differs. Then click **Compare**. Segments are lettered A to Z. Two shortcuts save setting segments up one at a time: - **Make each option a segment**, at the bottom of a question's answer list, gives every answer its own segment. The segment you're in keeps its answer, or takes the first one if none is picked. Strella adds a segment for each answer still missing and skips answers another segment already covers. The segment's other rules carry into each new one. - **Add everyone else** adds a segment holding everyone who isn't in your other segments, so comparing a group against the rest takes one rule instead of two. Beside a single segment, Strella labels it as the opposite, for example **Not Mobile**. Segments split whatever the page filter already shows, so the filter narrows everyone first and the segments divide what remains. A participant who matches two segments counts in both. While comparing, each chart draws one bar, tile, or box per segment: multiple choice, rating scale and slider answers, NPS, ranking, coded responses, themes, sentiment, time on task, and pass rate. Every share is out of that segment's own participants who answered. The Paths chart, Emotional responses, and the **By objective** view stay unsplit. Hover a segment's letter anywhere on the page to see its rules and size. The page link stores the comparison, so copying the URL shares it, and it stays in place as you move between questions. To go back to one view, open **Compare** and choose **Clear segments**. Downloads follow the comparison: each chart's spreadsheet includes a **By segment** sheet, and participant rows carry a **Segments** column. On the **By question** view, the sort button (the A to Z icon) beside a chart's **⋯** menu reorders that chart's bars. Each kind of chart offers its own orders: - **Multiple choice**: **Participants** (**Most participants** or **Fewest participants**) or **Display order**, the order the question lists its options. The chart opens on Most participants. A question with **Randomize** on offers Participants only, because each participant saw the options in a shuffled order. - **Rating scale**: **Scale order** or Participants. The chart opens on Scale order. - **Slider**: **Value** (**Lowest first** or **Highest first**) or Participants. The chart opens on Lowest first. Other answers and a multi-select question's "none of the above" choice stay at the bottom whatever their count. While comparing segments, the sort counts each participant once, even when they belong to two segments. The sort applies to that one chart and resets when you reload the page. **Download image** captures the chart as sorted; the spreadsheet downloads don't follow the sort. Below the charts, **Participant responses** lists everyone included, one row per participant. Type in **Search** to narrow it by name, summary, answer, coded category, or filter value. Downloads, **Expand all**, and selection follow the search. The table has these columns: - the participant's answer - each coded category for the question - the value behind each active filter - task result and time on task, for tasks - interview duration, sentiment, and date While comparing, a **Segment** column shows which segments each participant belongs to. The table scrolls sideways when there are many columns. Click a row to open that participant's summary and their transcript for this question right below it. From the row's **…** menu, **Play recording** starts their recording in a small player, and **View interview** opens the full interview. From the actions menu, choose **Refresh themes**. Three scopes: - **All**: re-run both objective and question analysis. - **By objective**: re-run only objective grouping. - **By question**: re-run only question grouping. Strella refreshes themes automatically as interviews come in. Refresh manually after big edits, such as changed objectives or excluded interviews. ## Things to know - **Statements** (the "I'm going to show you a concept" kind) aren't analyzed; only Ask-group steps produce themes. - **Excluded interviews don't appear** in the participant count, the themes, or the quotes. See [Review a single interview](/analyze/review-interview) for how to exclude. - **In-call survey responses** show as a section above the qualitative themes: distribution charts plus the raw counts. - **Questions you remove from the guide keep their data.** When a guide edit deletes a question (often when the AI assistant swaps one for a near-identical version), the answers already captured for it appear under a **Removed from guide** group in the **By question** dropdown, so that data stays reachable. A removed task keeps its **Passed** / **Failed** result too. ## Tips - **Start with By objective** for the synthesis story; switch to By question when you want to dig into a specific moment. - **Compare the groups you care about.** Putting "people who said 'yes' vs 'no'" side by side with **Compare** often reveals more than the overall analysis does. - **If a theme looks wrong**, check the quotes: sometimes the AI groups two distinct things under one label. Edit your objectives to be sharper and refresh. ## Related The objectives that drive theme grouping. Ask the AI analyst about what you see. Turn themes into a polished writeup. --- # See how participants sorted your cards Source: https://app.strella.io/docs/product/analyze/card-sort-analysis When participants complete a card sort, each of them drags your cards into groups and names those groups. In Results, on the card sort question, Strella shows the combined picture at the top, as one card with an **Overall sort** tab and a **Similarity matrix** tab, and each individual sort below, on the participant who made it. ## The Overall sort table The **Overall sort** table combines every participant's sort. Cards run down the left side, and every group name participants used runs across the top, both in alphabetical order. Each cell shows how many participants placed that card in that group, and the cell's shading deepens as more participants agree, so the strongest patterns stand out at a glance. - Groups with the same name are combined, even when participants typed them with different capitalization or extra spaces. The table shows the name the way the first participant wrote it. - **View as** switches every cell between **Percentage** (the share of participants who answered the question) and **Count** (the number of participants). - **Select a group's column header** to order the cards by that group, most-placed first. Select it again to flip the order, and a third time to return to alphabetical. - The **group filter** above the table narrows the view to the groups you choose. Cards that were never placed in a chosen group drop out of the table until you clear the filter. - Long card and group names are shortened to fit; hover over one to see the full name. ## The Similarity matrix The **Similarity matrix** tab shows how often participants placed each pair of cards in the same group, whatever they named it. Cards run along the diagonal, and the cell where two cards meet shows the percentage of participants who grouped them together. The deeper the shading, the stronger the agreement. Hover over a cell to see both card names and the underlying numbers, like "10 of 12". - Strella places cards that participants often group together next to each other, so blocks of dark cells along the diagonal show the natural clusters in your participants' mental model. - A **0%** cell is a finding too: no participant ever put those two cards together. - A blank cell means the pair can't be measured: no single participant sorted both cards, which can happen if you changed the card list while the study was running. - Percentages compare each pair against the participants who sorted both of its cards, so results stay accurate even when the card list changed mid-study. ## Each participant's sort Every participant's own sort appears in their entry in the **Participants** list at the bottom of the page. Expand a participant to see each group they made and the cards they placed in it. If they spoke during the sort, their transcript for the question appears alongside. ## The AI summary reads the sort itself Participants answer card sorts by dragging, not speaking, so Strella's AI writes the question's summary from the submitted sorts rather than from a transcript. The summary reads like a researcher's findings. It covers: - The summary describes where cards landed, never why a participant placed one; nothing was spoken, so motivations are not guessed at. - Judgments about where a card belongs come from which cards it was grouped with, so a stable cluster still appears even when every participant named their groups differently. - Sentiment and themes stay off for card sorts, and questions answered with a file upload have no summary at all; the sort table remains the detailed analysis. ## Tips - Sort by your most important group first to see which cards participants most agree belong there. - Disagreement is a finding: a card spread thinly across many groups is one participants could not place confidently. - If the table has near-duplicate groups (like "Pricing" and "Price stuff"), participants named the same idea differently. Read those groups together. ## Related The question-by-question view where card sort results live. Per-interview transcript and recording for any participant. --- # Code responses Source: https://app.strella.io/docs/product/analyze/code-responses Coding turns open-ended answers into something countable. For a given question, Strella reads every participant's response and places each one into a category, the same work a researcher does by hand when coding open-ended survey answers. The result is a chart of categories on that question's analysis page, and a column you can filter and group by. ## Automatic coding Every open-ended question, the kind participants answer in their own words, is coded for you. You don't have to write anything: Strella picks the most useful way to group the answers and sorts each participant into one category. Coding keeps up with your study as interviews arrive: - **Strella sorts new interviews into the existing categories.** As each interview finishes, Strella places that participant's answer into one of the categories already there, so the grouping stays stable while you collect data. - **Strella regenerates the categories as analysis refreshes.** As your interview count grows, Strella periodically rebuilds the categories from scratch over everyone so far. The grouping then re-forms around what the fuller set of answers shows. Automatic coding runs only on open-ended questions. Multiple-choice, rating, slider, and other survey-style questions already have fixed answers, so they aren't coded, and neither are tasks. ## Code a question yourself You can also code a question with your own prompt instead of the automatic grouping. Describe the angle you care about, for example "group by which pricing objection they raised", and Strella proposes the categories and assignments for you to review and adjust before saving. ## Refine a coding Open the 3-dot menu on a coding and choose Edit to refine it. You can: - **Edit the title and description,** rename the categories, and drag participants from one category to another. A coding carries a short title and a one-line description. They label its chart card, so a saved coding reads clearly on its own. - **Change the grouping prompt and regenerate.** Edit the prompt at the top of the editor and regenerate to re-group everyone from scratch on your revised instruction. The groups lock while Strella re-codes, then update in place for you to review before saving. ## Where coded categories appear - **On the question's analysis page** as a chart showing each category and how many participants fall into it. The chart card carries the coding's title and one-line description, so it stands on its own in an export or when a question has more than one coding. - **In the participants table** as a column. Automatically coded columns are hidden by default so the table stays readable when a study has many questions. Add any of them from the column menu when you want to filter or sort by it. - **As a way to order the responses below the chart.** Open the sort menu on the Participant responses list and pick the coding by name to read the answers grouped by their coded category, in the same order as the bars above. Participants Strella couldn't place in a category sit at the bottom either way you sort. ## Controls Automatic coding is always on for open-ended questions, so there's nothing to switch on. From the 3-dot menu on the analysis page: - **Re-code responses**: rebuilds the categories for every open-ended question right away, over all responses so far, instead of waiting for the next refresh. Use it after a burst of interviews when you want the grouping brought up to date immediately. Refreshing your analysis (All, or Questions, or a single open question's page) rebuilds the coding for you too. Re-coding regenerates the automatic categories from scratch. If you've hand-edited an automatically coded question and want to keep those edits, code that question yourself with your own prompt instead, so a refresh doesn't overwrite them. --- # See how participants use your Figma prototype Source: https://app.strella.io/docs/product/analyze/figma-prototypes When a participant interacts with a Figma prototype during an interview, Strella records the path they take and turns it into a click heatmap you can review afterward. It captures this for both free-explore and goal-based tasks. Use it to see where people clicked and where they got stuck. For a goal-based task you also see whether they reached the goal. ## Before you start Each teammate connects their own Figma account, and the prototype must be published and set to Anyone with the link can view. See [Connect Figma](/account/figma). Add a Prototype step to your interview and set up a free-explore or goal-based task. Strella captures the click heatmap for both task types once you connect and import the prototype from Figma. The import turns on the click tracking behind the heatmap. See [Add a stimulus to your interview](/build/stimulus). As the participant moves through the prototype, Strella captures each click and the path they take between screens. ## Read the results Open Results, go to Analysis, and select the prototype task. Across the top you see the average time on task and the misclick rate. A goal-based task also shows a completion rate. **Screen flow** combines everyone's routes into one chart of the screens participants moved through. Each screen is a bar, and the connections between them show how many participants went from one screen to the next. Hover a screen or a connection to preview it with the number of participants it covers. Select a screen to open its heatmap, the same as selecting its row in the Screens tab. Download the chart from the **…** menu beside the heading, as an image or as a spreadsheet. The spreadsheet holds two tabs, one counting the participants on each screen at each step and one counting the moves from each screen to the next. Its From node and To node columns name each screen with the step it sits at, so you can rebuild the flow in another charting tool. The chart appears once at least one participant has moved between two screens. When everyone stayed on a single screen, there is no flow to draw, so Strella leaves the section out. The click paths list the routes participants took through the prototype. A free-explore task shows them in a single Paths tab. A goal-based task splits them into Completed and Failed tabs by whether participants reached the goal, so you can see where people diverged. Select any path to open its click heatmap. The Screens tab lists each screen with the number of participants who reached it, the average time on it, and its misclick rate. It appears for both task types. Select a screen to open its heatmap. ## Tips - For a goal-based task, set a clear goal so the Completed and Failed split is meaningful. - Open a screen with a high misclick rate and read its heatmap to see where clicks land off target. - **Nothing here after an interview?** If you built the prototype step without a Figma connection, it showed the prototype and recorded no clicks. That is a supported way to run a study, and it produces no heatmap. Check that you are connected, connect if not, then re-import the prototype in Build before your next interview. Strella cannot add click data to interviews you already ran. ## Related --- # Get oriented in Results Source: https://app.strella.io/docs/product/analyze/overview Every finished interview appears in **Results**. From there you can review a single conversation, browse AI-generated themes, ask the AI analyst questions, and pull together a report. ![Results → Interviews: the grid of completed interviews, each a recording you can open, watch, and analyze.](/docs/analyze/results-overview.gif) ## When this matters Results is where you spend the most time once recruitment is moving. Knowing how the five tabs fit together matters more than knowing any single page. ## How it works Results has five sub-tabs: | Tab | What it's for | | --------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Interviews** | The list of recorded interviews. Click into any one to watch, read the transcript, and add highlights. The **⋮** menu on any session can export all session data to a spreadsheet. See [Exporting interview data](#exporting-interview-data) below. | | **Analysis** | AI-generated themes grouped by **objective** and **question**. Sentiment, quotes, jumps to source moments. | | **AI chat** | Ask the AI analyst anything about your interviews. It cites the source. Can generate reports and highlight reels. | | **Saved Reels** | The highlight reels you've saved. Saving pins a reel here, so it stops changing as new interviews come in. | | **Reports** | Generated reports you've made or are working on. | ## Exporting interview data From **Results → Interviews**, open the **⋮** (more) menu on any session and choose **Export interview data**. The export is a single Excel (.xlsx) file, not a plain CSV, with one row per session. Your question guide's sections and questions form a hierarchical header across the top, so each participant's transcript responses, in-call survey answers, and screener answers line up under the question they answer. The file also includes participant demographics as columns. The recording links in the file save each video under the participant's name when you click them, rather than playing it in a browser tab. The file downloads automatically once it's ready. You can keep working in Strella while Strella prepares it. This export excludes panel partials, which are incomplete interviews from vendor panels that never reached completion. ### Download every transcript From the same **⋮** menu, choose **Download all transcripts** to download every interview's transcript in a single text (.txt) file. The file uses the same layout as the transcript download on an individual interview. Each transcript uses the translated version when there is one, keeps speaker names, and adds any in-call survey answers alongside the spoken lines. Each interview opens with a header that names the participant (`Interview with `), numbered `(session N of M)` when one participant has more than one recording, then a status line marking the interview **Completed** or **Partial** (the participant left before the end), then the date. Human-moderated studies don't separate partials, so they carry no status line. Unlike the spreadsheet export, this file includes partial interviews and labels them, so you can still read what someone said even if they didn't finish. ### Download one interview's recording You don't need to open an interview to save its video. Hover over the interview's card, open its **⋮** menu, and choose **Download recording**. If the participant has one recording, it downloads as a video file. If they reconnected partway through and have several, Strella joins them into one video in the background. A message shows its progress on every page. When the video is ready, the message offers a **Download** button, and Strella emails you the link too. The option appears once the recording has finished uploading. ### Download all recordings From the same **⋮** menu, choose **Download all recordings** to get one zip file of every recording in the list. A dialog tells you how many interviews it covers. Choose **Download**, and Strella builds the zip in the background, usually within an hour for a large project. You can keep working in the meantime. The zip covers the interviews the list is showing, so search or filter first to download only some of them. One zip holds up to 500 interviews. For a larger project, filter the list and download it in parts. The option is greyed out when none of the listed interviews has a finished recording. While Strella builds the zip, a "Preparing recordings" message shows how far it has got. The message stays on screen as you move between pages in Strella, and comes back if you reload. When the zip is ready, the message offers a **Download** button, and Strella also emails you the link. Only you see the message and the link for a zip you asked for. A project prepares one zip at a time. While Strella builds a teammate's zip, everyone else on the project sees a "Preparing recordings" message on the Interviews page. They can request another zip once it finishes. The link goes only to the teammate who asked. A recording that makes no progress for 10 minutes is left out and listed in errors.txt. Strella reports a zip that hasn't finished after 7 hours as failed, and you can request it again. Inside the zip: - **One video per interview**, named after the participant and the interview date, the same name a single download gives. Dates are in your own time zone. When a participant reconnected partway through, their recordings are joined into one video. - **manifest.csv**, a list of every file with its participant, date and time, and length. The time column names your time zone. The list ends with any listed interview left out because its recording isn't finished yet. - **errors.txt**, only when a recording couldn't be included. The rest of the zip still arrives. The link works for 7 days. Opening it asks you to sign in to Strella with access to the workspace, so forwarding the email doesn't share the recordings. Everyone on the workspace can use this, viewers included. A zip of many long interviews can be tens of gigabytes. If your Mac's built-in unzip tool won't open a very large zip, open it with The Unarchiver or 7-Zip instead. ## How analysis updates Analysis runs in the background. As new interviews complete, themes update, usually within a few minutes after each interview finishes. If you've made big changes (added objectives, edited the guide), trigger a manual refresh: on the **Analysis** page, open the **⋯** (more) menu, choose **Refresh analysis**, and pick a scope: **All**, **Objective**, or **Question**. ## Things to know - **Results only shows completed interviews.** Drafts and abandoned interviews don't appear. - **You can exclude an interview from analysis** without deleting it (see [Review a single interview](/analyze/review-interview)). This is useful when a participant misunderstood the task or the recording is unusable. - **Themes are a snapshot at refresh time.** Editing objectives, adding highlights, or excluding interviews doesn't retroactively update the themes until the next refresh. ## Related Watch, read, and annotate one interview. Themes by objective or by question. Ask the analyst directly. --- # Review a single interview Source: https://app.strella.io/docs/product/analyze/review-interview When a participant finishes, their interview appears in **Results → Interviews**. Click in to see the recording paired with the transcript, plus everything else captured about the call. ![The interview detail view: the recording playing alongside the synced transcript.](/docs/analyze/video-transcript.gif) ## Steps From **Results → Interviews**, click any session. To move between sessions one by one, use the left and right arrows in the top-left of the session view. The video player is at the top. Use the timeline to jump around, or click any line in the transcript to jump to that moment. Keyboard shortcuts work whenever a recording is open: the left and right arrow keys skip back or forward 5 seconds, the space bar plays or pauses, and the up and down arrow keys adjust the volume. If the study recorded both screen share and webcam, they play side by side (**Split screen**). A view toggle switches between **Split screen**, **Webcam**, and **Screen share**. Recordings play at **1.2x** with subtitles on. The player's control bar shows the current speed. Click it to pick 0.5x, 1x, 1.2x, 1.5x, or 2x. Click **CC** to turn subtitles on or off. **CC** is underlined while subtitles are on. On a phone, or wherever the player is too narrow for the full bar, both controls sit in the **⋯** menu as **Playback speed** and a **Subtitles** switch. Strella remembers the speed and subtitles you last picked, so the next interview, clip, or AI chat reel you open in Strella plays the same way. Strella stores the choice in your browser, so another browser or device starts at 1.2x with subtitles on. Shared links, shared reports, and embeds ignore it. They play at the speed and subtitles chosen when the link was shared, or at 1.2x with subtitles on. The transcript runs beside the player. Strella generates it with speaker labels (the participant's name vs the moderator). The questions and steps from your guide appear inline (in gray) so you can see where each one starts. Search it with the search box at the top of the panel. As the recording plays, the transcript highlights the line being spoken. Turn on **Scroll with video**, at the top of the transcript, to keep that line centred as well, so you don't have to scroll to keep up. It starts off, because moving the transcript while you're selecting text for a highlight pulls the words out from under you. Turning it on takes you to where the recording is, whether or not anyone is speaking at that moment. In a pause it settles on the line that comes next, so you're in place before the speaker starts. With it on, the recording scrolls the pane while it plays, and you can't scroll by hand. Pause to read ahead or look back, or switch **Scroll with video** off to browse while it plays. Pressing play returns you to the line being spoken. Watching full screen leaves the transcript where it was, and leaving full screen moves you to the point the recording reached. The recording and the transcript scroll together in one pane. Following the transcript eventually scrolls the player out of view, and a small floating player takes over. On a phone or tablet there's no floating player, so the recording scrolls away. Close the floating player and it stays closed. A tab on the edge of the screen marks where it was. Click that tab to bring the player back without scrolling to find the video. Select any span of the transcript to save it as a **highlight**, a timestamped moment. The AI also auto-generates highlights on each interview. By default a highlight includes the preceding moderator question for context. You can toggle that per highlight if you only want the participant's reply in the clip later. Highlights are what get stitched into [highlight reels](/share/highlight-reels): take them generously. ![Adding a highlight to a reel from the interview's Highlights panel.](/docs/analyze/add-highlight-to-reel.gif) Alongside the transcript, the session view has three tabs: - **Summary**: an AI-written one-paragraph summary of the interview. - **Highlights**: the saved moments for this interview (with a count). - **Profile**: the participant's screener answers and metadata, plus the **audience** they came through: the share link or panel order, by the name it has on the Audiences page. Use it to tell segments apart without going back to the participants table. ![The interview view's right panel: Summary, Highlights, and Profile tabs, with the Highlights tab showing timestamped auto-generated highlights.](/docs/analyze/interview-highlights.png) From the actions menu (top-right), choose **Regenerate transcript**. Two modes: - **2 speakers**: assumes the AI and one participant. Best for solo interviews (the default). - **Multiple speakers**: for interviews with more than one voice. Useful when the auto-generated transcript mislabels speakers or misses names. The recording itself is unchanged. Highlights you've added to this interview may be reset by a regeneration, because they're anchored to transcript timestamps: re-take any you care about after regenerating. From the actions menu, toggle **Include in analysis** off. The interview stays in Results, but Strella leaves it out of theme generation, the AI chat, and reports. Use this when a participant misunderstood the task, the recording quality was bad, or the session went off-topic. Analysis shows the interview greyed out with an "excluded" badge. From the actions menu, choose **Delete interview** and confirm. Strella removes the interview from **Results → Interviews**, and it stops counting toward your completed-interview and recruitment totals. Deleting takes the interview out of the list, but a **direct link** to it still opens: if you bookmarked its URL or shared it (in a report, a reel, or a message), it loads with the recording, transcript, and highlights intact, with a **Deleted** badge next to the participant's name so you can tell. Deletion controls what appears in the list; it doesn't revoke links you've already shared. If a participant dropped out and rejoined, their interview has more than one recording, and the menu offers **Delete Partial Interviews** so you can remove just one. The interview stays in Results while any recording is left, and the recording you deleted disappears from the player and transcript immediately. ## Tips - **Highlights are your most useful artifact.** They're timestamped and the building blocks for highlight reels. Take them generously. - **Use the navigation arrows** to walk through interviews back-to-back. The Profile and Summary tabs make it fast to triage which interviews are worth watching in full. - **Set your speed once.** Pick the speed that suits how fast your participants talk; Strella keeps it for every interview you open next. - **Participants can't access their own interview.** After a participant submits, they have no access to their recording or answers. The recording and transcript live only here in Results, visible to your team. - **Quality-check badges** on each interview row flag fraud and quality risks. Each badge expands to show the full set of checks: completion percentage, response quality (rushing/one-word answers), response consistency, response timing, location consistency (browser timezone against the location the participant gave), and sensitive-content (MNPI) detection. Each check is Pass, Warning, or Fail with a short reason. The badges are guidance, not policy. Read the breakdown before you exclude an interview. ## Related Themes across many interviews. Stitch notes into a video. --- # See how participants complete on-screen tasks Source: https://app.strella.io/docs/product/analyze/task-analysis When you give participants a task on a live website or app, Strella watches the screen recording and reads what is actually on screen. From that it does two things for every participant: it grades whether they completed the task, and it traces the screens they moved through. You review both in Results, on the task. ## How Strella reads the screen Strella analyzes the screen recording visually, frame by frame, and judges only what is visible on screen. It does not rely on what the participant says out loud. - The verdict and the path come from on-screen actions and the screen's end state, never from a spoken claim like "okay, I think I found it." - If a task's screen never appears in the recording (for example the participant's camera is showing but not their screen), Strella leaves the task ungraded rather than guessing. Because Strella reads only the screen, the results show what participants did, not what they thought they did. ## Task pass / fail For each participant, Strella grades the task as **Passed** or **Failed** based on whether the screen clearly reached the end state the task asked for. - Opening, scrolling, or clicking toward the goal is not completion. The screen has to actually reach the goal for a Pass. - When the screen shows the task screen but the outcome is unclear before the recording ends, Strella leaves the task unresolved rather than marking it either way. - The verdict appears as a badge on each participant in the list, and the AI chat can answer questions about it, including the on-screen reason behind a verdict. See [Use Strella's AI chat](/analyze/ai-chat). ## Task pass rate and time on task Two charts at the top of the task summarize the whole group. - **Pass rate**: a donut of the share of participants who passed, colored by how commonly people failed: green at 80% or more passing, amber from 50 to 79%, and red below 50%. Participants the recording left ungraded are left out of both sides, so an unjudged task never reads as a failure. The donut appears once at least one participant has been graded. - **Time on task**: the median time participants spent on the task, with a box plot of the spread. An **Outliers** toggle includes or sets aside unusually long or short runs. Both export to PNG, and the pass rate also exports to a spreadsheet with a per-participant Passed / Failed breakdown. ## The screens each participant saw Expand any participant on the task to see a scrollable strip of screenshots, in order, of the screens they moved through. Each screenshot is labeled with the screen and the time it appeared. Select a screenshot to jump the recording to that moment. ## User paths across participants Above the participant list, the **Paths** chart combines everyone's routes into one shared flow of the screens participants moved through. - **Completion rate and median time on task** sit at the top of the chart. The median is the same number the question header and the Time on task chart show. - Each **screen** in the flow is a bar. The **connections between them** show how many participants moved from one screen to the next. - **Select a screen or a connection** to open a card next to it. The card shows a screenshot, how many participants it covers, the average time spent on that screen, and the faces of those participants. The card stays open until you click elsewhere. - **Select a participant's face** in that card to jump straight to their row in the list below, expanded to their screenshots and transcript. - **Download the chart as an image** from the **…** menu in the chart's header. The image holds the title, both stats, and the whole flow. That includes the part you must scroll to reach on screen. - **Download the chart as a spreadsheet** from the same menu. It holds three tabs. **Screens** lists how many participants stood on each screen at each step and the average time they spent there. **Transitions** lists how many moved from one screen to the next and how many stopped where they were. **Paths** has one row per participant with the route they took, whether they passed, and their time on task. - **Rebuild the flow in another tool** with the **From node** and **To node** columns on the Transitions tab and the participant counts beside them. Each column names a screen with its step. A screen a participant returned to appears as two separate points, as the chart draws it with two bars. The Paths chart and the per-participant screenshots are for tasks on a live website or app. Tasks that use a Figma prototype have their own click heatmap and path view instead. See [See how participants use your Figma prototype](/analyze/figma-prototypes). ## How participants looked while working If your workspace has facial emotion analysis turned on, Strella can also read how participants came across on camera while they work through a task, from what they said and how they said it. - An **Emotional responses** card above the participant list shows how many participants came across as delighted, interested, hesitant, skeptical, confused, frustrated, or indifferent. Hover a bar to see who is in it. - The reading is per sentence the participant spoke, so their transcript carries a small label beside the speaker name and time wherever something stood out. Most of what anyone says reads as neutral, and those lines are left unlabelled. This runs only when your project has facial emotion analysis enabled and is not blurring participant cameras. It also runs only for participants who saw the notice on the join screen. Interviews recorded in the mobile app have no reading. The card leaves out a participant with no reading instead of counting them as neutral. Treat this as a prompt to go and watch a moment, not as a measure. Concentration looks much like mild frustration, and someone can be pleased without showing it. Use it to find moments worth reviewing, then judge from the recording and what the participant said. Emotion analysis carries legal restrictions. It is prohibited in workplace and education settings in the EU, EEA, and UK, and restricted in some US states, so it must not be used to study employees, job candidates, or students. The person who enables the setting confirms the study covers none of those groups. ## Keeping results current Task grading and the Paths chart update automatically as interviews come in. Strella grades each finished interview and adds it to the shared flow on its own. The shared set of screens updates as more interviews arrive, so the chart reflects the full group without a manual step. ## Tips - Write the task with a clear end state ("reach the order confirmation page") so the Pass / Fail line is meaningful. - Use the Paths chart to spot where people diverged, then open that screen and select those participants to read their exact interviews. - If a participant's task looks ungraded, check that their screen was actually being shared during the task. ## Related Per-interview transcript, screen recording, and quality badges. The click heatmap and paths for Figma prototype tasks. Ask who passed or failed a task, and why. Add the website, app, or prototype participants act on. --- # Bring in interviews you ran elsewhere Source: https://app.strella.io/docs/product/analyze/upload-interviews Research you already ran does not have to stay outside Strella. Upload a recording or a transcript and it becomes a session in your project, with the same AI summary and the same searchable transcript as an interview Strella moderated itself. Notes are read the same way, so they can appear in themes, transcript search, and AI chat too. The difference is that a note isn't anyone's interview, so it never counts as a participant. ## When this matters You have a round of interviews from before you started using Strella, or from a study you moderated in another tool, and you want them in the same place as your Strella interviews. Themes drawn from thirty interviews are more useful than themes drawn from the eighteen that happened to run here. ## What you can upload | You have | Upload it as | Formats | Size limit | | --------------------------------- | -------------------- | ---------------------------------- | -------------------- | | A video recording of an interview | Video | MP4, MOV, WMV, WebM | 1 GB per file | | A transcript, in text | Transcript or notes | TXT, MD, DOCX, VTT, SRT, CSV, XLSX | 2 MB per file | | Notes about a conversation | Transcript or notes | TXT, MD, DOCX, VTT, SRT, CSV, XLSX | 2 MB per file | Transcripts and notes go through the same upload. Strella reads each file and classifies it as a transcript or as notes, so you do not label them yourself. Strella almost always reads subtitle files (VTT and SRT) as transcripts. ## Upload a file Go to your project's **Results** tab and stay on **Interviews**. Click the **⋮** menu at the top of the interviews list and open **Import**. **Upload video** for a recording. **Upload transcript or notes** for text. (**Import from projects** in the same menu is a different thing: it moves sessions between your own Strella projects.) Drop in one file, or up to 10 text files at once. Strella uploads each file on its own, so one that fails does not discard the rest, and you can retry just that file. The dialog closes once every file has uploaded. If some failed, the dialog stays open, lists those files, and shows a message with how many of your files uploaded. ## What happens to an uploaded interview A video or a transcript becomes a session in your project. Strella transcribes a video, then treats both the same way: - It gets an **AI summary** and **Highlights**, the same as an interview run here. - Strella includes it in the **themes** built across your research objectives. - It is searchable in **AI chat** and through the **Strella MCP connector**. - It **counts as a participant**, so it adds to your project total, your Completed count, and your participant limit. Because an uploaded interview fills a participant slot, uploading can fill your project and close it to new participants. The upload dialog warns you when a limit is set. Afterwards it shows how many participants you have against the limit, including when a batch went over it. ## What an upload does not get - **No per-question breakdown.** The **Questions** view uses what a moderator captured live while working through the question guide, so uploads do not appear there. This is true for uploaded recordings and uploaded transcripts alike. You get the summary, the highlights, and the searchable transcript instead. - **No highlight reel or click-to-seek, for a transcript.** A transcript has no video, so there is nothing to cut into a shareable reel and nothing to seek. Highlights still appear. Clicking one scrolls the transcript instead of jumping to a moment in a recording. - **Timing is only as good as the source.** A VTT subtitle file carries real timings. Other formats carry timings only when the file itself shows them (a `[0:03]` at the start of a line, for example). Without them, Strella spaces turns evenly and the timestamps are not meaningful. ## A note is analyzed too, but counts as nobody's interview When an uploaded file is your own notes rather than a record of a conversation, Strella saves it as a note. Notes appear in their own **Notes** section below the interviews grid, not as session cards. A note is read the same way an interview is: it can appear in themes, transcript search, and AI chat. What it never does is count as a participant, since a note isn't a conversation Strella can attribute to anyone. It also skips the Summary and Highlights tabs an interview gets, since those describe a person's session. Strella gives each note a short title drawn from its content, so you can tell notes apart at a glance. Open a note to read it in full, rename it from its title, or delete it. If a file you expected to become a session shows up as a note instead, Strella read it as more like notes than like a conversation. Re-upload it with the speaker turns intact, so Strella reads it as a transcript. ## Related Where uploaded interviews sit among your other results. Reading a transcript, adding highlights, and using the summary. Uploaded transcripts and notes both answer questions here, alongside everything else. Participant limits, and what happens when a study fills. --- # Best practices Source: https://app.strella.io/docs/product/build/best-practices A good study comes down to two things: screening for the right people, and asking questions that get honest, useful answers. This page collects the practices that make the difference. Some are **rules** worth following on every study, because they keep your responses genuine. Others are **rules of thumb**: good defaults that won't fit every case, called out as such. It's a solid self-check before you publish, by hand or after the [setup chat](/build/project-setup-chat) drafts your study. Each build page links back here, and this page links out to the page for the mechanics. Use it as a pre-publish checklist. ## Screener The [screener](/build/screener) has one job: it decides who gets in. These practices keep it doing that job. - **Mark every multiple-choice answer deliberately.** A screener with unmarked answers just collects data. It doesn't screen. Mark each option Qualify/Disqualify (single-select) or Must select/May select/Disqualify (multi-select), and confirm none were left unset. - **Use Disqualify only for hard requirements, not "less ideal."** Over-disqualifying is the costliest mistake: it slows recruitment to a crawl. Ask: would including this person _invalidate_ a finding? If yes, disqualify. If they're just not your dream participant, let them through. - **Keep interview questions out of the screener.** Anything beyond include/exclude dilutes the screener and slows recruitment. "What frustrates you about X?" is an interview question. The screener only needs "Do you use X?" - **Make options mutually exclusive and collectively exhaustive.** No two answers may both be true at once (watch off-by-one ranges like "1 to 5" and "5 to 10"), and anyone who shows up needs an option that fits. Turn on "Other" if they might not. - **Aim for three to five questions.** _(Rule of thumb.)_ A long screener slows recruitment, and if it takes eight questions to qualify, the criteria are probably tangled. Find the one or two filters that actually decide inclusion and trust the rest to manual review. Deviating is fine when eligibility is genuinely complex. - **On a panel, add a trap option marked Disqualify that looks real.** _(Rule of thumb.)_ Without one, inattentive panelists click through and pollute your data. Make it a plausible-sounding fake. ## Question guide ### Coverage - **Cover every research objective with at least one question.** If an [objective](/build/research-objectives) has no question, the study can't answer it. Read down your objectives and, for each, name the question that produces data for it. ### Question wording - **Don't lead or load.** A leading question plants the answer it wants ("Don't you think…", "What problems do you have with…") and participants are usually too polite to push back. Strip the assumption and let them supply it. _(A question inside a [conditional section](/build/conditional-sections) gated by a screener answer that already established the assumption isn't leading: the gate filtered to the audience for whom the framing is true.)_ - **Ask about what people did, not what they'd do.** "Tell me about the last time you…" is reliable; "Would you use a feature that…" is a coin flip: people are bad at predicting their own behavior. _(Reactions to something a participant is looking at right now, such as a concept, a prototype or a mockup, are fine; those are reactions to something concrete, not predictions.)_ - **Ask one thing per question.** "What do you like _and_ dislike about X?" is two questions, and you can't tell which half they answered. Split it, or ask one half and probe the other as a follow-up. - **Keep questions open-ended.** If a participant can answer in one word, the question isn't doing any work. "Walk me through…" beats "Did you…"; "How did you decide…" beats "Was it easy…". _(A closed question used as a checkpoint or filter is fine.)_ - **Don't ask the same thing twice.** Duplicate questions waste interview time and add noise, not richness. Consolidate them or sharpen the difference. _(Two questions on mutually-exclusive branches aren't duplicates: no participant sees both.)_ - **Match the question type to the data you want.** Open questions give qualitative texture; structured types ([survey questions](/build/survey-questions), such as multiple choice, ranking and slider) give data you can compare across participants. Don't stack several structured questions in a row. Mix in open questions, so the guide does not feel like a form. ### Structure & pacing - **Follow this order: warm-up → context → deep dive → exploration → wrap-up.** _(Rule of thumb.)_ One or two warm-up questions let participants settle before you ask them to recall specifics in detail. - **Five to seven sections is a comfortable range.** _(Rule of thumb.)_ One giant section feels monolithic; ten feels fragmented. The thing to watch isn't the count: it's one section that dwarfs the rest (say 80%+ of the questions), where fatigue and thin coverage creep in. - **Aim for roughly 15 to 30 minutes.** _(Rule of thumb.)_ Under 15 is usually too thin; past 45, participants fade and late answers get worse. The real question is whether each item earns its time, not whether the total clears some threshold. - **Put a short statement at each section transition.** _(Rule of thumb.)_ It signals a topic change and gives participants a beat to reorient. A guide that's only questions feels like an interrogation. ### Follow-ups - **If a question has follow-ups on, write follow-up guidance.** Without it, the moderator probes blind and you get generic "tell me more." Guidance tells it where to dig, e.g. "probe the specific pain points and the workarounds they use." - **Keep guidance on the current topic.** Follow-ups are a depth tool, not a coverage tool. If the guidance reads like "also find out about pricing" on a discovery question, a real question got buried. Promote it to its own item. - **Calibrate the depth.** Default to Low for most questions; reserve High for the one or two pillar questions per study. If the guide runs long, turn follow-ups down on questions that aren't adding nuance; if it runs short with good coverage, deepen them on the pillars; if it runs short with coverage holes, add questions instead. ### Sensitive topics - **Frame sensitive topics neutrally, and don't avoid them.** For AI, jobs, money, health: add a brief "there are no right or wrong answers, I'm just curious about your experience" statement before the section, and ask in a behavioral, non-judgmental form. ### Intro & close - **Keep the intro to project context in two friendly sentences.** Don't mention recording (already consented to; raising it is unsettling), time estimates (usually wrong; they make people rush), or future contact (don't promise what the researcher might not deliver). - **End with a simple thank-you.** No "we'll be in touch," no five ways of asking "anything else?", which bleeds the interview into "actually, one more thing…" Thank them and end clean. ### Media & tasks - **Keep questions with the [media](/build/stimulus), not before or after.** The visual stays on screen while the moderator asks about it. Questions placed next to the media instead of with it fire after it disappears, and the participant is answering about something they can no longer see. - **Use real URLs only.** No `example.com`, no placeholder links, no "this is roughly the link." If you don't have the real URL yet, leave the media empty and add it later: fake URLs ship to participants. - **Make tasks prescriptive, granular, and nested.** Prescriptive: no ambiguity about what to do. For usability tests, a guiding statement orients the participant (where to be, what to look at). Granular: several short tasks beat one long one. Nested: a task sits under the image, video, or [prototype](/build/stimulus) it's about. - **In front of a prototype, use a task, not a question.** Open questions about a static prototype get shallow answers: the participant is describing a screenshot. A task ("find where you'd update your billing info") forces interaction; for open exploration, frame that as a task too ("explore and tell me what stands out"). Any question comes _after_ the task. ### Choosing the right element - **Use a slider for ratings, scales, and magnitudes.** Do not render a 1 to 10 satisfaction question as ten radio buttons. Sliders show the answer is on a continuum and produce cleaner, comparable data. Use multiple choice only when the options are categorical and unordered. - **Mix element types.** _(Rule of thumb.)_ A guide built entirely from open questions misses structured signal. Mix in a statement at section transitions, a multiple-choice item or two for quick quantitative reads, and a ranking when the topic is about prioritization. Not every guide needs all four, but a guide with only questions is usually under-specified. ## Related Apply the screener rules above as you build. The steps that produce a well-structured guide. Every objective needs a question. Start here. The AI drafts a study against these practices. --- # Use the build assistant Source: https://app.strella.io/docs/product/build/build-assistant The Assistant panel is the third panel on the right side of [Build](/build/build-screen). It reads your current screener and question guide and edits them on request. Use it as a writing partner that works from your current draft. ![The Assistant panel in Build: typing a plain-language request ("Add a followup question about length of trial") and the question guide updating in response.](/docs/build/build-assistant.gif) ## Steps On desktop, the Assistant is visible by default. If it's collapsed, click the chat icon on the right edge. On mobile, tap the **Assistant** tab at the bottom. Examples that work well: - "Rewrite question 4 so it doesn't lead the answer." - "Add a follow-up to the pricing question that probes for willingness to pay." - "Group all the onboarding questions into a section." - "Make the screener shorter. Keep it to 5 questions." The assistant applies edits to your draft as it works. Review the changes, click **Undo** to reverse them, or ask for a revision. Each turn builds on the previous one. You can send a series of small edits without re-explaining your study. Strella saves any change the assistant makes to your draft, the same as an edit by hand. Participants see nothing until you publish. ## What the assistant deletes The assistant deletes a question, statement, section or screener question when your message asks for a removal, in any language, such as "remove the pricing question", "replace the guide with this document" or a yes to a removal it suggested. When your message asks for no removal, it keeps the item, and edits it or asks whether you want it removed. When the assistant moves a question, for example from the screener into the question guide, it deletes the original only after the copy is in place. ## Clear a section Ask to "Remove all questions from Comedy" to keep the section and its statements, or "Empty Comedy" to remove everything inside that section. If a name is missing or matches several sections, the assistant asks which one you mean before editing. Select a question or another item inside a section and say "Clear this section" to clear its containing section. If your reference is vague, such as "remove the questions from it", the assistant asks what you mean. If you edit a question while the assistant is clearing the section, Strella keeps your changed question and shows a notification asking you to retry. **Undo** restores the items that request removed and keeps your other edits. **Redo** removes those restored items again. If you add, move, or delete an item manually while the assistant is working, Undo follows the order of those actions. The assistant's changes before and after your manual action can take separate Undo steps. ## Copy from another project Ask the assistant to bring questions over from another project in your organization, by its name. - **Copy a whole guide or screener.** "Copy the screener from Clinic operations survey" replaces this project's screener with that project's. Ask for the question guide, the screener, or both. If several projects match the name, the assistant asks which one you mean. **Undo** restores what was here. - **See or copy one question.** "What does Clinic operations survey ask clinics that never take payment online?" shows you that question's wording and answer options and changes nothing. "Copy that question over" adds only that question here, with the same wording, answer options and qualifications. - **Its routing comes with it.** When you pick a screener question by where an answer sends people, the assistant sends that answer to the copy here and routes the copy's answers the way the other project does. - **Your routing stays.** The assistant doesn't turn off routing you set unless you ask it to. If copying a question means changing who qualifies, it asks first. ## Build from a website Paste a link and the assistant reads the page before it writes anything, so the tasks and questions it drafts are about what's actually there (the real sections, the real wording) rather than generic ones about the topic. This works for any public page: a live site, a hosted prototype, a product or pricing page, a competitor's site. It isn't limited to Figma. - **Paste the link with your request.** "Make a guide from this" or "what should we test here?" The link can go in the same message or an earlier one. - **Point at the exact pages you care about.** The assistant reads the links you paste, not the pages they lead to. If a specific flow matters, paste that page's link too. You can include several links in one message. - **Private addresses won't work.** A page only you can reach (`localhost`, an address on your office network, anything behind a login) can't be read. Use a publicly reachable link instead, like a preview deployment that doesn't require signing in. The assistant doesn't read links to your Strella projects either: to use another project's guide or screener, name the project. - **If a page won't load, the assistant says so.** Strella can't retrieve some public pages. Paste the parts that matter, or a screenshot, and the assistant works from those. It never searches the web for your product or your study. - **Check that it actually read the page.** The assistant doesn't always say when a page failed: it may write something plausible from general knowledge instead. If the guide feels generic or describes things you don't recognise, treat the page as unread and tell the assistant what's on it yourself. ## Attach a screenshot Copy an image and paste it into the message box, with no need to save it to a file first. It attaches the same way a dragged-in file does, and you can send it with your request. ## Tips - **It edits the draft, not the live interview.** To change how the AI moderator _delivers_ the interview, edit [moderator settings](/configure/voice-language-behavior). - **Ask for a critique, not just edits.** "What's weak about this screener?" is a useful prompt. - **Reference steps by their text**, not by position: "the question about pricing" is clearer than "step 4" if you've reordered. - **The assistant won't go off and answer your research question.** It's there to help you build the study, not to imagine what participants might say. ## Related Where the Assistant lives. Every place AI helps across a study, and how this assistant fits in. Drop a guide you already wrote into the Assistant and have it rebuilt question for question. --- # The Build screen Source: https://app.strella.io/docs/product/build/build-screen **Build** is where you design your study. It has three panels: - **Guide** (left): the tree of every step in your screener and question guide, in the order participants will encounter them. - **Properties** (center): the editor for whatever step you click. - **Assistant** (right): an AI build assistant that can read your draft and help you edit. ![The Build screen's three panels: the Guide tree (Screener and Interview) on the left, the Properties editor in the center, and the Assistant panel on the right.](/docs/build/build-screen.png) ## How it works The Guide panel has two regions: - **Screener** at the top: pre-interview questions that decide who qualifies. - **Question guide** below: the steps the AI moderator will run with anyone who passes the screener. Click **+ Add element** in either region to insert a new step. Click any existing step to open it in Properties. A small bar at the top of Build shows your study's status (draft, live, closed), a **Preview** button, and a **Publish** button. To leave Build, switch between **Build → Recruit → Results** at the very top of the page. ## Things to know - You're always editing a draft. Changes only go to participants when you click **Publish**. - The Assistant panel reads both the screener and the question guide. You can ask it about either. ## Related The pre-interview questions that decide who can enter the study. The steps the AI moderator runs through with each participant. Get the right-panel AI to draft and edit for you. Take the interview yourself before publishing. --- # Add a card sort Source: https://app.strella.io/docs/product/build/card-sort A card sort is an interactive exercise inside your question guide. Strella shows the participant a set of cards you wrote, and the participant drags each one into a group. It's the fastest way to learn how people naturally organize things: navigation labels, features, topics, or concepts. Card sort is a **Test** element. Like a survey question, the participant interacts with it and submits an answer. Like a section, it holds follow-up questions of its own, which stay available while the sort is on screen. See [Step types](/reference/step-types) for how it sits alongside the other elements. ## Steps In [Build](/build/build-screen), click **+ Add element** and choose **Card sort**. Strella adds it to the guide like any other element, and its properties open on the right. A card sort works one of three ways. Pick the one that matches what you want to learn: - **Open**: participants create, name, and delete their own groups. Use this when you want to discover their mental model from scratch, without steering it. This is the default. - **Closed**: participants sort into groups you defined, and only those. Use this to test whether a structure you already have (say, a proposed navigation) holds up. - **Hybrid**: participants start with your groups but can add their own. Use this when you have a structure in mind but want to catch the buckets you missed. Add the cards participants will sort. Each card is a short label: a feature name, a topic, a menu item. Add at least one card. The guide flags the element as incomplete until you do. Keep labels short and parallel, and avoid wording that hints at where a card "should" go. For a **closed** or **hybrid** sort, add the group titles participants sort into. At least one is required. An **open** sort has no researcher-defined groups, because participants make their own. Skip this step for an open sort. A card sort can hold its own follow-ups: open questions, tasks, statements, and survey questions. Drag them inside the card sort the way you'd nest steps in a section. The sort stays on screen while these run, so use them to ask *why* a participant grouped things the way they did, right after they finish. ## What the participant sees Strella shows the participant all the cards, and the participant drags each one into a group. In an open or hybrid sort they name their groups as they go and can rename or remove them. In a closed sort they drop cards into the groups you set. When they're done, they submit, and any follow-up questions you added run next. ## Reading the results Strella combines every sort into one card-by-group table and keeps each participant's own sort on their entry. See [How participants sorted your cards](/analyze/card-sort-analysis) for reading the results. ## Related How to add and arrange every element, and what can nest inside what. Read the combined sort table and each participant's sort. Every element you can add, grouped by purpose. --- # Conditional sections Source: https://app.strella.io/docs/product/build/conditional-sections Any Section in your question guide can be shown only to participants who match a rule. The rule, the section's **Section logic**, is built from one or more conditions on what you already know about the participant: their screener answers, their answers to earlier survey questions in the interview, or which recruitment quota they fall into. If the conditions match, the participant sees the section. If they don't, Strella skips the section. ## Set up Section logic ### Steps In [Build](/build/build-screen), select the Section, then open **Section logic**. The dialog opens with the section's label at the top (e.g. "Tracking habits"), so you know which section you're editing. Click **Add condition**. Each condition reads as a sentence (**If** _[field]_ **is / is not** _[answer]_) and can reference: - **Screener answers**: anything the participant answered on the screener. - **Earlier questions in this interview**: any survey question in a section _above_ this one. - **Quota assignment**: which recruitment quota the participant falls into (**is in** / **is not in**). For a multiple-choice, rating, or screener choice question, one condition can name **more than one answer**: open the answer menu and check each one you want. **is** matches a participant who chose _any_ of the answers you selected; **is not** matches one who chose _none_ of them. So a single condition can say "Product purchased **is** Frozen pizza or Frozen desserts." Disqualifying answers stay listed but can't be selected, because a participant who gave one is screened out before the interview. For Slider answers, the condition compares against a threshold instead: **above**, **below**, or **equal to** a value. For an NPS answer, pick groups (**Promoter**, **Passive**, **Detractor**), individual scores, or a mix of both, or compare the score against a threshold. Groups cover most cases, such as asking a follow-up only of detractors; a threshold handles a cut like "9 or higher". ![A Section's Section logic panel: a "Drivers-only section" shown only if the screener commute answer is "I drive my own car" and the quota assignment is Drivers.](/docs/build/conditional-section.png) When the gating field is a multiple-choice question, a single condition can name several answers at once: the participant matches if they picked **any** of the answers you choose, so you don't need a separate condition per answer. When you add more than one condition, choose how they combine. **and** means the participant must match every condition. **or** means matching any one of them is enough. The builder flags two common mistakes before you can publish: - **Contradictory conditions**: a combination that can never be true (e.g. "role is PM _and_ role is designer"). - **Conditions that exclude every participant**: based on the screener you've published, the section would never run for anyone. Fix or remove the conditions to clear the warning. The outcome is always the same: **Then: Show this section**. Section logic decides _who_ sees the section; it never changes the content. ### The ordering rule A condition can only reference a survey question that appears **earlier** in the guide than the section it gates, because the answer has to exist before the decision point. The builder enforces this; you won't see later survey questions in the field picker. ### The big limitation: structured answers only Conditions work on structured answers: multiple choice, sliders, ratings, NPS, and screener picks. Ranking, matrix, open text, and media upload answers can't be used. Conditions **can't** branch on what a participant _said_ in open conversation. "If they mention price, go to the pricing section" isn't supported. Workaround: ask it as a Multiple choice survey question first ("Which of these matters most to you?"), then set the Section logic to check that answer. (Within a single question, the moderator's follow-ups _do_ react to what was said. See [Follow-up guidance](/build/interview-guide). What you can't do is route the _path_ through sections based on free speech.) ## Things to know - **Conditions are sticky.** They're evaluated from data that doesn't change mid-interview (screener answers and quota membership), so a participant who leaves and comes back sees exactly the sections they saw before. - **Variants and conditional sections are different tools.** Conditional sections route specific participants to specific content based on what you know about them. [Randomization](/build/randomization) routes participants at random for A/B testing. Use the right one for the job. - **Multiple conditional sections compound.** A participant only sees a section if that section's own conditions match. Heavy condition use can quickly narrow who sees what, so test the real path by going through a [share link](/recruit/share-links) as a participant would. (Preview shows every section regardless of routing, so it won't reflect the conditional path. See [Preview your interview](/build/preview).) ## Related Where sections live in the guide. The random-and-sticky alternative for assignment. Quotas are one input for conditional sections. --- # Host a prototype and add it to your interview Source: https://app.strella.io/docs/product/build/host-a-prototype Strella shows a prototype to participants from a public web address. It does not host prototype files. A prototype you built as HTML, for example with Claude, v0, Figma Make, Lovable, or Bolt, needs a public URL first. Publish it with a free web host, then add that URL to your question guide as an embed, or as a website if the host does not let its pages load inside another page. ## What you need - Your prototype as one or more HTML files. Most AI builders export a single `index.html`, sometimes with a folder of assets. Keep the files together in one folder. - A free account with a web host. The steps below cover four common choices. ## Publish your prototype Each host takes your HTML and returns a public URL. Netlify Drop is the fastest for a single prototype because it needs no project setup. Go to https://app.netlify.com/drop. Drag the folder that contains your index.html onto the page. Netlify publishes the prototype and shows a public address on a netlify.app subdomain. Copy it. Sign in with a free account to keep the site and get a stable address. Sign in at https://vercel.com. Add a new project and upload your prototype folder, or connect a Git repository that contains it. Deploy the project. Vercel returns a public address on a vercel.app subdomain. Copy it. Create a repository at https://github.com and add your prototype files with index.html at the root. Open Settings, then Pages, and set the source to your default branch. GitHub publishes the site on a github.io address. Copy it. Sign in at https://pages.cloudflare.com. Add a project and upload your prototype folder, or connect a Git repository. Deploy the project. Cloudflare returns a public address on a pages.dev subdomain. Copy it. ## Check the URL Open the URL in a private browser window. The prototype loads without a sign-in prompt. If it asks you to log in, the site is private. Make it public in the host's settings, then check again. ## Add the prototype to your interview Once you have a working public URL, add it to your question guide as a media block. Use the Embed type so the prototype loads inside the call window and stays in front of the participant the whole time they work. The media block holds the prototype and keeps it on screen for the questions and tasks inside it. In your question guide, click Add element and choose Embed from the Usability group. Paste your public prototype URL into the media block. Static sites published to Netlify, Vercel, GitHub Pages, and Cloudflare Pages allow this by default, so an embed loads for the hosts above. Nest a task under the media block so participants work with the prototype while it stays in view. For the full media-block options, see [Add a stimulus to your interview](/build/stimulus). Embed loads the prototype inside the call window, so it works only when the host lets the page load inside another page. If the prototype does not appear as an embed, add it as a Website instead. A Website opens the prototype in a new browser tab: the participant screen-shares that tab, works through the task, and returns to the call when they are done. ## Related Show an image, video, website, embed, or Figma prototype during the interview. Review click paths and screen metrics after the interview. --- # Import from another study Source: https://app.strella.io/docs/product/build/import Import copies questions from another study in the same organization. The source study stays unchanged, and edits to it do not update the imported copy. ## Choose what to import 1. Open **Build**, open the screener or interview section's **More** menu, and choose **Import**. 2. Find the source study in **Recents**, or use **Browse** to open its workspace and folder. Search by project name to find studies across the organization. 3. Select a study to preview its screener and interview. Clear the checkboxes for anything to keep out of the import. A section checkbox selects or clears the section and its questions. 4. Click **Import**. An empty selection disables Import. Cancelling a confirmation returns to the preview with the selection intact. ## Keep conditional logic If selected sections or screener skip rules need questions left out of the selection, choose whether to include those questions: - **Keep conditional logic** includes the referenced questions so the copied rules can still work. - **Remove conditional logic** imports the selection and removes rules that cannot work without the omitted questions. Strella shows a warning when it cannot preserve some rules. Strella does not copy quotas and participant groups, so it removes rules that depend on them. Review the copied rules: importing repairs older answer references, which can change which sections participants see compared with the source study. ## Add or replace Choose how the selected items enter the current study: - **Add to the current guide** appends them and keeps the existing questions. This is the default. - **Replace the current guide** replaces the selected screener or interview content. If no questions are selected from one of those sections, that section stays unchanged. A screener can contain up to 40 questions. If adding would exceed the limit, choose Replace or cancel and select fewer questions. Replacing the screener clears its legal agreement setting. Review any existing quota or interview rules that referred to the questions being replaced. When keeping logic includes screener questions, choosing Replace also replaces the current screener with those questions. The **Replace** option names the content it overwrites. If an added screener question has the same wording as an existing question, its copy gets an `(imported N)` suffix. Participants see that wording. An import notification lists any renamed questions. The screener and interview import together in one step. **Undo** restores both. ## Human-moderated studies Human-moderated studies use a screener-only import. Choose a source study, review the screener, and click **Use this screener** to replace the current questions. ## Related - [Build your screener](/build/screener) - [Build your question guide](/build/interview-guide) --- # Build your question guide Source: https://app.strella.io/docs/product/build/interview-guide The question guide is the script the AI moderator follows. It's an ordered list of steps. The AI runs them in order, asks dynamic follow-ups where you've enabled them, and skips conditional sections that don't apply. ![Building a question guide in Build: the guide tree of sections and questions, with a section selected and its properties open.](/docs/build/build-guide.gif) ## Steps In [Build](/build/build-screen), the question guide appears below the screener divider in the Guide panel. Click **+ Add element** to open the step menu. Steps are grouped into five categories: - **Ask**: Open question, Task, Statement. - **Survey**: Multiple choice, Ranking, Open text, Slider, Rating scale, Media upload, Matrix, NPS. - **Structure**: Section, Randomization (shown as **Variant comparison** in the editor). - **Usability**: Prototype (Figma), Website, Image, Video, Embed, Text. - **Test**: [Card sort](/build/card-sort), [Tree test](/build/tree-test). The most common step is **Open question**, a question the AI moderator will ask in your own words. ![Adding a step to the question guide: the Add element control opening the step menu below a section.](/docs/build/add-interview-step.gif) In Properties, type the question text. Write it the way you'd say it aloud. By default the moderator paraphrases it conversationally rather than reading it word-for-word. Turn on **Ask question verbatim** (below) when the exact wording matters. To format what participants read on screen, use the buttons above the text box: bold, italic, underline, strikethrough, bulleted and numbered lists, a heading, a subheading, a quote, and links. Formatting works on questions, tasks, statements, and survey questions, but not on answer options, section titles, or media. The moderator still gets the words as one plain line, so a heading and its bullets are asked as part of the question. Optional fields depend on the step type: - **Follow-up depth** (on open questions): Off, Low, Medium, High (roughly 0, 1, 2, 3 follow-ups the moderator may ask). - **Follow-up guidance** (on open questions): tell the moderator what to probe ("dig into pricing objections"). - **Ask question verbatim**: force the moderator to read the question word-for-word instead of paraphrasing. - **Blur recording for privacy**: blur the screen recording while just this item is active. Useful for moments when the participant enters personal or sensitive information. A **Task** is for when the participant needs to do something: walk through a prototype, visit a website, open a file. Where an open question asks the participant to *talk*, a Task asks them to *do* something while thinking aloud. Tasks pair naturally with a stimulus: drop a [Figma prototype or other media](/build/stimulus) into the same Section, and it stays visible the whole time. Task completion behavior (the moderator decides, a button click, or a verbal "done") is set globally in [Task completion handling](/configure/voice-language-behavior). Tasks do not support dynamic follow-ups. Add an open question after the task to explore the participant's experience. A **Statement** is the AI saying something to the participant: context for the next question, a transition, a thank-you. No response is collected. Statements are read aloud exactly as written. Common use: put a Statement as the **first** item to add your own framing right after the moderator's greeting, or as the **last** item for a thank-you in your own words right before the moderator's built-in closing. Add a **Section** to group related steps. Sections can be: - **Conditional**: only shown to participants who match a rule. See [Conditional sections](/build/conditional-sections). - **Limited to a percentage** of participants: show this section to only N% of participants (sticky per participant). - **Randomized**: pick one of multiple paths at random per participant (see [Randomization](/build/randomization)). Drag steps into a section to nest them. Drag steps to reorder. Right-click for duplicate or delete. ## What can go inside what The builder enforces these containment rules and blocks invalid drops. | Building block | Can contain | Can live inside | | ----------------------- | ----------------------------------------------------------------------------------------- | -------------------------------------------------------- | | **Section** | Open questions, tasks, statements, survey questions, media blocks, Randomization blocks | Top level of the guide only | | **Randomization** block | Variant options only | Top level, or a Section | | **Variant option** | Open questions, tasks, statements, survey questions, media blocks | A Randomization block | | **Media block** | One piece of media, plus open questions, tasks, statements, and survey questions about it | Top level, a Section, or a Variant option | | **Card sort** (Test) | Open questions, tasks, statements, and survey questions as follow-ups | Top level, a Section, or a Variant option | | **Individual items** | (these are the content itself) | Top level, a Section, a media block, a card sort, or a Variant option | Common rules people hit: - **Sections can't nest.** For sub-topics, use separate consecutive Sections. Participants never see section boundaries anyway. - **Media blocks can't nest** and hold exactly one piece of media. For a second piece of media, add a second media block. - **Media blocks can't contain Sections or Randomization blocks**: just questions, tasks, statements, and survey questions. - **Only Variant options go inside a Randomization block.** - **A [card sort](/build/card-sort) holds follow-up questions but not media or another card sort.** Its follow-ups are open questions, tasks, statements, and survey questions. - **A [tree test](/build/tree-test) holds follow-up questions the same way.** Its follow-ups are open questions, tasks, statements, and survey questions, asked while the outline stays on screen. ## Best practices A few rules keep responses genuine and the conversation natural. See [Best practices](/build/best-practices) for the full checklist. - **Don't lead or load.** "What problems do you have with X?" plants the answer. Ask "How do you feel about X?" and let the participant supply it. - **Ask about what people did, not what they'd do.** "Tell me about the last time you…" beats "Would you use a feature that…" because people predict their own behavior poorly. (Reactions to something they're looking at right now are fine.) - **One thing per question.** "What do you like _and_ dislike about X?" is two questions. Split it. - **Keep questions open.** If a participant can answer in one word, the question isn't working: "Walk me through…" beats "Did you…". - **Cover every [research objective](/build/research-objectives)** with at least one question. - **Follow a sequence**: warm-up → context → deep dive → exploration → wrap-up, with a short statement at each section transition. - **Keep the intro to project context** (don't mention recording, time, or future contact), and **close with a simple thank-you.** ## Tips - **Write questions like you'd say them out loud.** The moderator's voice will deliver them. Long, jargon-heavy questions sound stilted when spoken. - **Use follow-ups generously on open questions.** They draw a fuller explanation from a participant who answers "yes". - **Use Statements between sections** to give participants context and warmth. The moderator will sound less robotic. - **Tasks need clear success criteria.** Tell the moderator what "done" looks like: completing a flow, finding a button, voicing a reaction. ## Related Structured response steps: multiple choice, ranking, ratings, sliders, open text. Images, video, websites, embeds, written concepts, and Figma prototypes. Show sections based on screener answers, earlier survey responses, or quotas. Show different content to different participants, randomly and stickily. Have participants drag cards into groups to reveal how they organize your ideas. Give participants a task and a menu of labels to learn whether they can find things in your navigation. --- # Preview your interview Source: https://app.strella.io/docs/product/build/preview Preview catches problems that reading the guide can't. Preview is for studies the AI moderator runs. A study you moderate yourself has no question guide for the AI to follow, so there is nothing to preview and Strella does not show the Preview button. Check your screener from the Screener tab instead. ## Steps Top-right of [Build](/build/build-screen). The preview starts from the beginning of the guide. Use the question list in the call to jump ahead. What happens next depends on the **Allowed devices** in [Interview setup](/configure/recording-modes). - **Screen recording off** (any devices): a new tab opens on the participant join page. Continue in this tab. - **Desktop and mobile, screen recording on**: a dialog asks where to take the interview, **Web** or **Mobile app**. Continue opens the new tab: Web runs the interview there, Mobile app shows a QR code. - **Mobile only, screen recording on**: no choice to make; the new tab shows a QR code. ![Clicking Preview in Build, with the participant join page opening in a new tab.](/docs/build/preview-interview.gif) On desktop, allow microphone access. When your study allows two or more [languages](/configure/voice-language-behavior), first pick the language to preview in from the menu above **Allow**. It lists only the allowed languages and starts on the default. The AI moderator greets you in that language and begins. For a Mobile app preview, scan the QR code with your phone's camera. It opens the interview in the Strella app, so install the app from the App Store or Google Play first if you haven't already. The app shows the same unpublished guide you see in Build. Preview plays the full guide, every section, in order, and skips screener qualification, so conditional routing doesn't apply. That's by design: you're checking wording and flow, not eligibility. To confirm that qualifying vs. disqualifying answers and [conditional sections](/build/conditional-sections) send people to the right place, test through a [share link](/recruit/share-links) as a participant would. A web preview skips camera and screen sharing. To try screen recording as a participant experiences it on a phone, preview in the Strella app. Answer like a participant would, and pay attention to: - **How the moderator's voice reads your questions.** Awkward phrasing is much more obvious out loud. - **Whether follow-ups help.** Are they useful, or do they repeat what you already asked? - **Whether tasks make sense.** Can you tell what success looks like from the instructions alone? In a web preview, the two arrows at the bottom left of the call move one question back or forward without answering. The count between them, such as **4 / 9**, opens a list of every item in the guide, grouped by section as in the question guide in Build. Type in its search box to find an item, and pick a row to jump straight to it. - **Going back** asks the question again from the start, and a new answer to a survey question replaces the earlier one. It steps over statements and images that move on by themselves. - **Jumping forward** skips the items in between without asking them. - **Previous** is off on the first question. **Next** is off on the last one, so end the call to finish. The Strella app has no arrows or list. Close the tab, go back to Build, and edit. Repeat preview as needed. Preview interviews show up in **Results → Interviews** alongside real ones (delete them if you want to keep the list clean), but Strella excludes them from analysis. ## Tips - **Preview both branches** of any randomization step (refresh until you get the other one). - **Preview in the Strella app** if your study allows mobile. Mobile participants take the interview in the app, and the layout is different from the desktop call. - **Opening the preview link on your phone** shows the app screen with a **Get Started** button that opens the interview in the app, or store links if it isn't installed. For a desktop-only study it shows the same "only available on desktop" notice a participant would see. - **Have someone else preview too.** They will catch confusing phrasing you no longer notice. ## Related What happens when you're ready for real participants. Confirm the recording settings before you preview. --- # Set up your study details Source: https://app.strella.io/docs/product/build/project-details Project details are what your study looks like to participants: its title, description, logo, and colors. Find them in one card, **Project details**, in Settings. ![The Project details settings card: study title, description, logo, and primary color.](/docs/build/project-details.png) ## Steps Click the gear icon in the top-right of any study page. Settings open as four groups: **General**, **Interview setup**, **AI moderator**, and **Advanced**. Each is a set of cards. **Project details** is the first card under **General**. The study title is the name participants see on the landing page and registration page. Use a clear descriptive name: "Pricing perception study" beats "Q3 research v2." Strella caps the title length, and the input shows the count. A short summary of what participants will be asked to do. It appears on the landing and registration pages. Capped at 300 characters. Don't promise outcomes ("change our product"). Describe the activity ("a 30-minute conversation about how you shop for X"). The same card holds the study **logo** and **primary color** participants see. See [Customize the participant experience](/configure/participant-experience) for how these appear. Title and description appear immediately to anyone who clicks your share link. ## Pick who gets notified Email notification recipients live on their own **Notification settings**, opened from the notification bell in the top-right of any study page. Open the bell, click **Notification settings**, and switch on each teammate you want emailed about activity on this study: interviews completed, orders shipping, problems. The list searches your organization members, and you can pick none, some, or all. ## Tips - **Email notifications are per-study**, not per-organization. Set them on each study you want to follow. - **Don't use placeholder titles in production.** Some panel vendors show your study title to candidates in their invite, so a placeholder like "Test 7" hurts your sign-up rate. ## Related The other half of study setup. Logo, colors, and the rest of what participants see. --- # Start with the study setup chat Source: https://app.strella.io/docs/product/build/project-setup-chat When you create a new study, Strella opens a setup chat. Tell it what you want to learn. It asks clarifying questions, then drafts both a screener and a question guide you can edit in Build. ![The study setup chat: describing a research question in plain language while Strella drafts the study's objectives and audience alongside.](/docs/build/setup-chat.gif) ## Before you start You'll need: - A clear sense of your research question, even if rough. "I want to understand why X" is enough. - An idea of who your participants are (B2B or B2C, and a description of the audience). You can leave the rest blank, and the chat asks you for it. ## Steps From your **Projects** list, click **New project**. Strella creates the study and opens the setup chat right away. Describe what you want to learn and who from. Give it as much context as you want up front: the more it has, the sharper the first draft. Already have a study outline? Upload it (see the next step). The chat will ask about objectives, the audience, what participants do during the interview, and how you want to recruit. Each answer makes the draft sharper. Attach supporting material such as a research brief, a discussion guide, or images (PDF, DOCX, TXT, MD, or images, up to five files). Anything you give it becomes context for the draft, so an existing outline can produce a near-complete study in one pass. A screenshot can go straight in: copy it and paste into the message box, with no need to save it as a file first. The plan beside the chat fills in as you talk: **Background**, **Research Objectives**, and **Target Audience**. To rewrite a section yourself, click **Edit** next to it, change the text, and click **Save**. The assistant builds on your wording from its next reply. **Edit** is unavailable while the assistant is still writing the plan. If you are mid-edit when it starts, **Save** waits until it finishes and keeps what you typed. When the chat says it's ready, click **Create project**. Strella opens the [Build screen](/build/build-screen) with a complete screener and question guide loaded. You have a draft. Nothing is published yet, so review it, edit anything you don't like, and publish when you're ready. ## Tips - **Be specific about your audience.** "B2B" generates generic questions; "VPs of revenue ops at companies with 200 to 2,000 employees who currently use HubSpot" generates targeted ones. - **Fix one section by editing it.** If a single section is off, click **Edit** and rewrite it. If the whole draft is heading the wrong way, **start a new chat** with a sharper brief. ## Related What you'll do after the chat hands off the draft. The in-Build AI for iterating after you have a first draft. --- # Randomize sections (A/B variants) Source: https://app.strella.io/docs/product/build/randomization Randomization is how you compare options. You set up two or more **Variant options** under one **Randomization** block (shown as **Variant comparison** in the editor), and Strella sends each participant down one variant, or a subset, or all of them, depending on how you configure it. ## When this is useful - **A/B concept tests.** Show half your participants design A and the other half design B, then compare reactions. - **Order effects.** Show three concepts in different orders to different people so order (and recency bias) doesn't skew the result. - **Stimulus rotation.** Test which of five tagline options resonates without showing every option to every participant. ## Steps In [Build](/build/build-screen), click **+ Add element → Randomization** (under the Structure group). It can live at the top level of the guide or inside a Section. Each variant is a small group of steps. Add as many as you want to compare. Drag steps from elsewhere in your guide into a variant, or add new ones inside. A Randomization block needs at least two Variant options. ![A Randomization block in Build labeled "Messaging Variant Test," expanded to show Variant A — Urgency-focused and Variant B — Discovery-focused, each with its own question.](/docs/build/variants.png) Put the variant-specific content in each option. The shared lead-in and follow-up stay outside the Randomization block. Example layout: - **Statement:** "I want to show you a few concepts." - **Randomization block:** - Variant A: Image of design A + Open question - Variant B: Image of design B + Open question - **Open question:** "Anything else you want to add?" If you have more variants than you want each participant to see, set the subset size. With five variants and a subset of two, each participant sees two of the five (still picked at random). By default each participant sees one variant per Randomization block, though you can show a subset or all of them (see the step above). In Results, you can filter by variant to compare reactions across them. ## How assignment works - **Random.** Variant choice is automatic per participant. You can't hand-pick who sees which variant. If specific participants need specific content, use [conditional sections](/build/conditional-sections) keyed to screener answers or quotas instead. - **Sticky.** If a participant leaves and comes back, they get the same variant they started with. Their place in the interview and their conditional-section eligibility stay the same too. - **Approximately balanced.** Assignment is random per participant, not strictly balanced. You'll get close-to-equal counts at scale, but with small samples one variant can run ahead. ## Tips - **Branches don't have to be the same length**, but keeping them roughly equal helps. A 5-step branch next to a 2-step one means some participants get a much longer interview than others, which can frustrate the long-branch group. - **For more than 4 variants**, use a subset: exposing every participant to all of them quickly hits diminishing returns and stretches the interview. ## Related Sections, conditions, and ordering: the structure around your variants. Assigns by conditions instead, for when assignment must be deterministic. Filter by variant in Results to compare reactions. --- # Write research objectives Source: https://app.strella.io/docs/product/build/research-objectives The AI analyst uses your objectives to group themes after interviews come in. Specific objectives produce specific themes. ## Steps Click the gear icon in the top-right of any study page, then choose **Research objectives**. One per line. Write each as something you want to be able to answer at the end of the study. Specific: - *Understand which pricing tier B2B buyers seriously consider first.* - *Find the friction points in onboarding for users who churned within 14 days.* Too vague: - *Learn about pricing.* - *See how onboarding feels.* ![Settings → Research objectives: a list of objective fields with an Add objective button, above the Research background and Company fields.](/docs/build/research-objectives.png) Strella saves all your changes together when you close the dialog or click **Save**. Once you have interviews, **Results → Analysis** groups themes under each objective. You can revisit and edit objectives anytime, and the analysis will regroup the next time it runs. ## Tips - **Every objective needs a question.** Map each objective to at least one question in your [question guide](/build/interview-guide). Otherwise the study can't answer it. See [Best practices](/build/best-practices). - **Write objectives in the voice of "what I want to learn,"** not "what I want to ask." Questions belong in the question guide; objectives belong in Settings. - **Edit objectives mid-study** if you realize you're learning something more useful than you planned. The themes update. - **The setup chat** can draft objectives for you if you describe the study. It's a good starting point if you're staring at a blank field. ## Related How objectives show up in Results after interviews come in. The steps that produce the data your objectives draw from. --- # Build your screener Source: https://app.strella.io/docs/product/build/screener The screener runs before the interview. Participants who pass continue, and those who don't are disqualified. Use it to filter for the right audience and to collect the structured answers that quotas and section logic reference later. ## Steps In [Build](/build/build-screen), the screener appears at the top of the Guide panel, above the divider. Click **+ Add element** at the top of the panel. Pick one of the five screener types: - **Multiple choice**: for a single-select question. Supports qualify/disqualify logic per option. - **Short text**: one-line written answer (e.g. job title). - **Long text**: a paragraph answer. - **Number slider**: a numeric value on a sliding scale. - **Legal agreement**: an NDA or consent the participant must check to continue. See [screener question types](/reference/screener-question-types) for the full reference. ![Building a screener question: choosing multiple choice, adding options, and marking qualify/disqualify.](/docs/build/build-screener.gif) Type the question in the editor. For multiple choice, add each option as its own row. The order is the order participants see it (unless you randomize options on the question). To format the question, use the buttons above the text box: bold, italic, underline, strikethrough, bulleted and numbered lists, a heading, a subheading, a quote, and links. See [Format screener questions](#format-screener-questions) for what participants see. On a **single-select** multiple choice, mark each option **Qualify** or **Disqualify**: - **Qualify** means the participant continues. - **Disqualify** means Strella sends the participant to the goodbye page at the end of the screener. On a **multi-select** multiple choice, each option is **Must select**, **May select**, or **Disqualify**. Participants must tick every "Must select" option to qualify, "May select" options don't affect qualification, and ticking a "Disqualify" option screens them out. By design, disqualified participants still finish the rest of the screener questions before being sent to the goodbye page, so they can't guess which question they "failed" and retry. To override this for a specific question, use skip logic with destination **End of screener** (see next step). On multiple-choice questions, you can route specific answers to a different question or to the end of the screener. Click the logic icon next to an option to set its destination: - **Next question** (default): continues to the next question in order. - **Jump to question**: skip forward to a specific later question. - **End of screener**: send the participant straight to the goodbye page without finishing the rest of the questions. Use **End of screener** sparingly. The default "finish all questions" behavior protects against fraud. Jumping to the end tells the participant which question disqualified them. If your study requires an NDA, add a **Legal agreement** step and upload the document participants accept. It can be a PDF, DOC, or DOCX, up to 6MB. Participants who don't check the box can't continue. Participants see the document called "Terms and Conditions" by default. To use the document's real name, fill in **What participants see this called**, for example "Terms of use, consent and release form". Leave it blank to keep the default. Either way participants reading in another language see it in theirs. Drag questions to reorder. Right-click a question for duplicate or delete. ## Format screener questions Formatting lets a longer question read clearly, for example a list of the apps you're asking about or a link to a definition. Participants who take the screener on Strella see the question formatted, just as you wrote it. To add a link, select the words and press the link button or Cmd/Ctrl+K, or paste a web address over the selected words. With nothing selected, the address itself becomes the link text. A link shows blue and underlined, so it reads differently from underlined text. A few things stay plain: - **Answer options.** Only the question text can be formatted, because quotas, skip logic, and recruitment panels match on the exact option wording. - **Questions a quota or section condition uses.** The question field is locked for these, the same way it can't be reworded. - **Translated screeners.** A participant reading in another language sees the question as plain text, with each link's address in brackets after its words. - **Panel screeners.** When a recruitment panel runs your screener on its own site, panelists see the question as one plain line: list items become a comma-separated list, and each link's address appears in brackets. The same plain line shows in the participant table and exports. A formatted question doesn't turn a typed web address into a link on its own. Add links as described above. The 500-character limit counts that plain line, including any link addresses in brackets, not the formatting. ## Best practices These keep the screener doing its one job (include or exclude) without slowing recruitment. See [Best practices](/build/best-practices) for the full checklist. - **Mark every multiple-choice answer deliberately**: Qualify/Disqualify on single-select, or Must select/May select/Disqualify on multi-select. An unmarked answer screens no one. - **Disqualify only for hard requirements.** Over-disqualifying is the costliest mistake; if someone is just not ideal, let them through. Would including them _invalidate_ a finding? Only then disqualify. - **Keep interview questions out**: "What frustrates you about X?" belongs in the [question guide](/build/interview-guide); the screener only needs "Do you use X?" - **Make options mutually exclusive and cover everyone**: no overlapping ranges, and turn on "Other" if someone might not find a fit. - **Aim for 3 to 5 questions.** Keep the screener tight; add more only if eligibility is genuinely complex. - **On a panel, add a realistic trap option** marked Disqualify, so inattentive panelists don't pollute your data. ## Tips - **Avoid free-text screener questions** unless you really need them: free-text answers can't be used as quota criteria. - **Preview the participant view** with the Preview button in Build. Bad questions are obvious when you take the screener yourself. - **A time estimate** appears at the top of the screener section (and at the top of the interview section), as a quick gut-check before you publish. ## Changing a question with skip logic Remove the question's skip logic before changing its type or turning **Multiple selection** on or off. After changing the question, configure skip logic again for the answers participants can select. A saved rule displays **Needs attention** in any of these cases: it does not match the question's type, it lacks its conditions, it refers to a removed answer or matrix cell, or it points to a missing or earlier question. Select the question, remove its existing skip logic, and configure the intended rule again. Strella blocks publishing until you repair or remove the rule. ## Related Use screener answers to cap how many of each segment can complete the study. The five types and what each can do. What participants do after they pass the screener. Copy a screener you've already built. --- # Add a stimulus to your interview Source: https://app.strella.io/docs/product/build/stimulus A **stimulus** is something you show a participant mid-interview: an image, a video, a website, an embed, a written concept, or an interactive Figma prototype. Adding one creates a **media block**: a container holding that one piece of media plus any questions you want asked about it. The media appears on the participant's screen and stays visible for every question inside the block, then comes down when the block ends. Questions inside appear as a card overlaid on the media. If you turn on **Split view** in project settings, they instead take a column of their own beside the media, so nothing is covered. ## The media types | Type | Source | Where it opens | | ----------- | --------------------------------------------------------------------------- | -------------------------- | | **Image** | Upload only (PNG, JPG, GIF, WebP; max 4 MB) | In the call window | | **Video** | Upload (MP4, MOV, AVI, WebM; max 1 GB) **or** paste a YouTube or Vimeo link | In the call window | | **Website** | Paste the URL | Opens in a new browser tab | | **Embed** | Paste a URL or the site's embed code | In the call window | | **Text** | Type the statement yourself | In the call window | For an interactive Figma design, with optional click tracking and goal detection, use a **Figma prototype** instead (covered in [Figma prototypes](#figma-prototypes) below). **Text vs Statement.** A **Statement** is read aloud by the moderator and never appears on screen, and nothing can be nested under it. A **Text** block puts the words on screen for the participant to read, and holds the questions you ask about them. Reach for a Text block whenever the wording itself is what you're testing: a product concept, a positioning line, a scenario to imagine, instructions to follow. ## Add a media block In [Build](/build/build-screen), click **+ Add element** and pick **Image**, **Video**, **Website**, **Embed**, or **Text** from the Usability group. Each creates a media block. ![Adding a media block to the question guide: inserting an image and a website via Add element.](/docs/build/stimulus-upload.gif) Per type: - **Image**: upload from your computer. There's no link option for images. - **Video**: upload, or paste a YouTube/Vimeo link. Other video hosts aren't supported as links; upload the file instead. - **Website**: paste the URL. The site opens in a new tab during the interview, not inside the call window. - **Embed**: paste a URL or the embed code (the builder pulls the link out of the code for you). Only works for sites that allow themselves to be embedded; if a site refuses to load, use **Website** instead. - **Text**: type the words participants read. There's nothing to upload or paste a link to; the text you write *is* the stimulus. Use the formatting buttons above the box for bold, italics, underline, a heading, a bulleted or numbered list, a quote, and links. You can also paste straight from a doc or a slide and the formatting comes with it. A link opens in a new tab, with the interview still running behind it, so use one only for a reference a participant may want; when the page itself is what you are testing, use **Website**. Properties shows a preview, so you can confirm the right media loads. ![A Video media block in Build: the Video element selected in the guide, with its label, source URL, and an embedded video preview in Properties.](/docs/build/add-stimulus.png) Drop open questions, tasks, statements, or survey questions into the same media block. They get asked while the media is visible. A media block can't contain Sections, Randomization blocks, or other media blocks; for those, use a separate section. Every media block needs its media, and image, website, embed, text, and prototype blocks also need at least one question or task inside. Video is the exception: a video can stand alone: participants watch it, say they're done, and the interview moves on, so questions inside a video block are optional. Two settings affect video playback: - The participant's **microphone is muted while the video plays**, so they finish or pause before responding. - **Pause video to talk** is a project-wide study setting (it applies to every video, not just this block). With it on, the AI stays silent during playback and the participant pauses the video when they want to talk. When the participant reaches the block, the media loads automatically. The moderator introduces it and asks the participant to confirm they can see it. By default, the moderator can't see the media itself, so it relies on what the participant says. [Agent vision](/configure/voice-language-behavior) (beta) lets it look periodically. ## Hold interaction until they answer For a **Figma prototype** or an **Embed**, you can stop participants from touching the media until they've answered the question in front of it. On any question or task inside the block, open **Advanced** and turn on **Block interaction until answered**. While the moderator asks that item, Strella dims the media and shows a short banner that tells the participant they can't interact yet, with the question on top. Strella removes the dimming and lets the participant interact again when the moderator moves to the next item. Use it to capture a genuine first reaction ("what's your first impression of this screen?") before the participant starts clicking around. Don't turn it on for a task that asks the participant to *do* something in the prototype, since it would leave them looking at a screen they're meant to act on but can't touch. The setting applies only inside a Figma or Embed block. Moving the item to another block, or changing the block to a Website, Image, or Video, turns it off. Turn it back on if the item returns to a Figma or Embed block. ## Send participants to a specific page For a **Figma prototype** with no Figma connection, or an **Embed**, you can send participants to a specific page the moment the interview reaches a question. On any question or task inside the block, open **Advanced** and enter the page's link in **Redirect to URL for this question**. When the interview reaches that question, the media reloads at that page. Moving on to a question with nothing set leaves the media where it is, and coming back to the question reloads it again. Use it to reload the media at a known starting page before a new task, or to open the screen a question is about. It works the same on web and in the mobile app. It is independent of **Block interaction until answered**: set both on the same question to open a fresh page and keep participants from touching it until they answer. Inside a Figma prototype, the link must be one Figma can embed, such as a prototype share link. Strella saves it in the form Figma uses for embedding, so the field shows a longer version of the link when you come back to the question. The setting applies only inside a Figma or Embed block, and inside a Figma block only before you connect it. Once you connect Figma and import the prototype, the block manages its own screens and the setting does not apply. Moving the item out of a Figma or Embed block, or changing the block to a Website, Image, or Video, clears the link, the same way it turns off **Block interaction until answered**. Changing an Embed block to a Figma prototype keeps the link. If Figma can't embed it, the field shows an error until you replace it. ## Watermark sensitive media A watermark marks unreleased or confidential media with a code unique to each participant, so a screenshot shared outside the study can be attributed to the participant who took it. It discourages sharing and makes a shared screenshot attributable; it doesn't stop a participant from capturing the screen. To turn it on, select an **Image**, **Video**, **Embed**, or **Prototype** block, open **Advanced**, and turn on **Watermark**. Website and Text blocks don't have the option: a website opens in its own tab, outside the call window. ### What participants see A repeating diagonal mark covers the media. It reads the participant's code, "Strella", and the date the participant joined the study, for example `K7QM9WR4 - Strella - 2026-09-28`. The mark is semi-transparent and never blocks clicks, scrolling, or playback. The code is an eight-character identifier, not the participant's name or email. - **Images** carry the mark inside the image file, so a participant who saves the image gets the watermarked copy. - **Videos, embeds, and prototypes** have the mark drawn over them in the call window. Watermarked embeds and prototypes can't be switched to fullscreen. Watermarked videos can't go fullscreen in Chrome or Edge. Firefox and Safari let a participant take a watermarked video fullscreen, and the mark does not cover a fullscreen video. - **In the mobile app**, only images carry the watermark. ### When someone tries to remove it If a participant hides or deletes the watermark in their browser, it comes straight back, and a banner warns them that tampering conflicts with the terms and conditions. Strella records the attempt on the interview: - The participant's quality checks show **Fraud detected**, with the time of each attempt in the recording, so you can jump to it. A mobile app session records screenshot attempts the same way. - Strella emails the teammate who created the project after the first attempt in an interview, with a link to that interview. Strella flags the check for your review. The check alone doesn't change whether Strella recommends accepting the participant. ### What it doesn't prevent - A screenshot, a screen recording, or a phone photo of the screen. The watermark makes the copy traceable. It doesn't block the copy. - A participant who is determined to strip the mark from a video, embed, or prototype, since it's drawn over those in the browser rather than into the media. Images are the most robust, because the mark is part of the file. ## Figma prototypes A **Prototype** step embeds a Figma prototype in the call. The participant explores it live, and the moderator asks about their reactions. There are two modes: - **Default (no Figma connection).** Paste the prototype share link. That's it. Participants explore the prototype inside the call window, and the task ends the normal way (the participant says they're done, or uses the end-task control). No screen picker, no goal detection, no click tracking. Pressing **Import** without a connection is fine: the prototype is embedded and Strella records no clicks. You can still control where a participant lands with [Send participants to a specific page](#send-participants-to-a-specific-page). - **Connected to Figma (per researcher).** Connect your Figma account from the builder. This imports the prototype's screen list and unlocks goal-based tasks, start-screen pickers, and click tracking. Use the prototype share link from Figma (a `figma.com` link to a prototype, design, or file). A website published from **Figma Make** or **Figma Sites** is _not_ a prototype. Add it as a Website or Embed instead. ### Steps In [Build](/build/build-screen), click **+ Add element → Prototype** (under the Usability group). Use the published prototype URL, the one Figma gives you when you share. If the prototype loads in the preview, you're good for default mode. Click **Connect** next to the prototype card and authorize Figma. The connection belongs to you inside the organization you are working in. Teammates who author prototype steps each connect their own. Connecting in another organization disconnects this one. See [Connect Figma](/account/figma). Once connected, choose **Import Figma prototype** to pull in the prototype's screen list. The prototype's share permission in Figma must be set to **Anyone** with **can view**. Otherwise the import can't read it. If you change the design later, use the refresh control to re-import; newly added or renamed screens may require re-picking your selections. Set the exploration mode on a Task. The Task options below appear only with Figma connected: - **Free explore**: participant wanders the prototype. Set a **Start screen** where they begin. The task ends the normal way (participant says they're done, or via the end-task control depending on your [task completion handling](/configure/voice-language-behavior)). - **Goal based**: set both a **Start screen** and a **Goal screen**. When the participant reaches the goal screen, the task ends automatically and the moderator moves on, with no confirmation needed. With **Goal based** mode, the click path the participant takes is recorded and appears as a heatmap after the interview. **Scale prototype** fits the prototype to the call window. It suits desktop designs. Mobile designs usually look better at native scale. Tell the moderator what success looks like: "find where you'd update your payment method" or "explore for 2 minutes and tell me what feels off." The moderator uses this both to brief the participant and to know when the task is done. When the participant reaches this step, the prototype loads in the call window. The moderator introduces the task and watches them work. After the interview, Strella generates a click heatmap (for goal-based tasks on imported prototypes). ## Best practices - **Keep questions inside the block, with the visual**: questions placed before or after fire when the media isn't on screen, so participants answer about something they can't see. See [Best practices](/build/best-practices). - **Use real URLs only**: no `example.com` or placeholder links; if you don't have the real one yet, leave the media empty and add it later. - **Nest a task under the media,** and keep tasks prescriptive and granular: tell the participant exactly what to do. - **In front of a prototype, use a task, not a question**: open questions about a static prototype get shallow answers; a task ("find where you'd update billing") forces real interaction. For open exploration, frame that as a task too. ## Tips - **Websites open in a new tab**: pair website tasks with [screen-share recording](/configure/recording-modes) so the moderator can see what the participant does. - **Keep videos short** (under 2 minutes if possible): long videos use up interview time and the participant's attention. - **For mobile-app reactions**, you'll need screen-share. The participant shares their phone screen during the interview. - **Goal-based prototype tasks need a real goal screen.** "Settings page" is a goal; "the right page" is not. - **Test the prototype yourself** with [Preview](/build/preview). Broken Figma links don't show up until they break for a participant. - **The participant doesn't see your prototype screen list**: they see the prototype as you'd ship it. If you want them to hop between flows, design the prototype that way. - **If a prototype or embed fails to load**, the participant can tell the moderator and it'll trigger a reload. ## Related Set up the integration. Required for the screen-aware prototype extras. Pick how the call captures what the participant sees and does. Where stimulus blocks fit among questions, tasks, and sections. The media and task rules above, in the full pre-publish checklist. --- # Add survey questions to your interview Source: https://app.strella.io/docs/product/build/survey-questions A survey step collects a structured answer mid-interview. The AI moderator pauses, the participant taps their answer in the call interface, and the AI continues. Survey responses always show on the participant's screen, and the interview advances only when the participant answers. The moderator can't answer for them or skip past on its own. ## Steps In [Build](/build/build-screen), click **+ Add element** and choose one from the **Survey** group: - **Multiple choice**: pick one option (or several with multi-select). - **Ranking**: drag options into a preferred order. - **Open text**: a written answer (typed, not spoken). - **Slider**: pick a number on a sliding scale. - **Rating scale**: Likert, Faces, or Stars. - **Media upload**: upload a photo, video, or document as the answer. - **Matrix**: pick one option for each row of a grid. - **NPS**: rate how likely they are to recommend something, from 0 to 10. Type the question. Keep it short: survey prompts work better as one clean sentence than as a paragraph. Each survey type has its own rules and settings: see below. Most survey steps have an **Ask follow-up questions** toggle. Turn it on if you want the moderator to ask why they chose what they did: "why a 3 and not a 5?" produces the richest data. Survey responses show up in two places: - Inline in each interview's transcript, alongside the spoken responses. - In **Results → Analysis**, grouped under each question (distribution charts sit above the qualitative themes). ## Per-type rules ### Multiple choice - Add each option as its own row. - **Multi-select** lets participants pick more than one option, with optional **minimum** and **maximum** selections. - **Other** option adds a free-text "Other" choice. - **Exclusive options** (multi-select only) adds an Exclusive section under the options. Drag an option such as "None of the above" into it: when a participant picks it, their other picks are cleared, and picking any other option clears it. An exclusive pick on its own completes the question, even with a minimum set. With a minimum of 2 or more, the hint participants see names the exclusive options too, for example "Choose at least 2, or 'None of the above'". Turning off Exclusive options or Multiple selection moves those options back into the main list. - **Randomize** shuffles option order per participant. Exclusive options always stay at the bottom, in the order you set. ### Ranking - Needs at least **2 options**. - **Randomize** shuffles the starting order. The participant drags options into their preferred order. ### Open text - Participants type their answer into a text box, and **Continue** becomes available once they type something. - Optional **Character limit**. With a limit set, the box stops accepting text at the limit and shows a running count, such as 42/200. Leave it empty for no limit. - In analysis, the **Responses** card lists every answer with the participant's name. Search the answers or participant names, sort by participant, sentiment, date, or length, and see the response count and average word count at the top. ### Slider A slider starts with a range of 1 to 10 in steps of 1. Pick a **Scale** to set a common range in one click, or type your own minimum and maximum. - **Scale presets**: Percentage (0 to 100), 0 to 10, 1 to 10, and 1 to 5. Each fills in the minimum, maximum, and a step of 1. - **Custom range**: edit **minimum** or **maximum** directly. The Scale shows the preset your range matches, or **Custom** when it matches none. - **Step**: choose from steps that divide the range evenly, so the participant can always reach the maximum. For steps of 10 on a 0 to 100 scale, keep the minimum at 0, since a 1 to 100 range doesn't divide by 10. - The minimum must be below the maximum. ### Rating scale - Exactly **4, 5, or 7 points**. No other point counts are supported. - Three styles: - **Likert**: labeled agreement scale. - **Faces**: emoji faces. - **Stars**: star icons. - Every point starts with a label filled in for you: **Strongly disagree** to **Strongly agree** for Likert, **Very bad** to **Very good** for Faces, and **1 star** to **5 stars** for Stars. - On **Likert**, you can rewrite any point's label. If you clear a label, the original comes back when you click away, so no point is ever left blank. - Changing the number of points updates the labels you haven't touched and keeps the ones you rewrote. Changing the style resets every label to that style's set. - Faces and Stars labels can't be changed. If you need a fully custom set of options, use **Multiple choice** instead. ![A Rating scale survey step in Build: the question with Style set to Likert, a 7-point scale size, and labels from Very dissatisfied to Very satisfied.](/docs/build/rating-scale.png) ### Media upload Participants answer by uploading one file from their device. Use it to collect a photo of a product in use, a short video of a routine, or a document such as a receipt. - **Accepted file types** has a toggle for each kind of file. All three are on by default, and at least one must stay on. - The upload box tells participants which file formats they can pick. After the upload finishes, they see a preview and can **Replace** the file before pressing **Continue**. - Media upload questions have no **Ask follow-up questions** toggle, because the moderator can't see the uploaded file. | Kind | Formats | Maximum size | | ------------ | ---------------------------------------------- | ------------ | | **Image** | JPG, PNG, WebP | 25 MB | | **Video** | MP4, MOV, WebM | 200 MB | | **Document** | PDF, DOC, DOCX, XLS, XLSX, PPT, PPTX, TXT, CSV | 25 MB | Uploads show in two places. In **Results → Analysis**, the question's **Uploaded files** card lists each participant's file: expand a row to preview an image or video, or open or download any file. In the interview's transcript, a **View media** link sits under the question and opens the file in a new tab. ### Matrix A matrix asks the same question about several items at once. Rows are the items (for example, store names) and columns are the answer options every row shares (for example, "This week", "This month", "Longer ago"). Participants pick exactly one column in each row. - Add **2 to 10 rows** and **2 to 7 columns**. - **Randomize rows** shuffles the row order for each participant. Columns always keep the order you set. - Participants must answer every row before they can continue. - On a wide screen the matrix shows as a table with a radio button under each column. On a phone or beside a stimulus, each row becomes its own box with the options listed underneath. - In that table, the row text gets twice the width of each answer column, and the table widens on its own when a row is long, so sentence-length rows stay a few lines tall. - Participants see the question labeled as multiple choice, not as a matrix. - A matrix answer can't decide which [conditional section](/build/conditional-sections) a participant sees, and can't be used as an analysis filter. In analysis, the **Responses** card shows a grid of rows against columns, shaded by how many participants picked each option for each row. Switch between percentages and counts, sort by a column, or filter the columns shown. Each participant's own answers appear as a table of the rows they placed under each option. Renaming a row or column keeps its earlier answers. Deleting one removes its answers from the grid. ### NPS An NPS (Net Promoter Score) question asks participants how likely they are to recommend something on the standard 0 to 10 scale, anchored "Not at all likely" and "Extremely likely". The scale and its labels are fixed, and participants reading in another language see the labels in theirs. For a numeric scale with your own range, use **Slider**. - **Style** sets how the scale looks. **Buttons** (the default) shows eleven buttons from 0 to 10 that participants tap once. **Slider** shows a handle they drag to a whole number. Both record the same score, so changing the style never affects answers you already have. - Participants see the question labeled as multiple choice. - A later [conditional section](/build/conditional-sections) can depend on the answer, either by group (Promoter, Passive, Detractor), by individual scores, or by comparing the score against a threshold, such as 9 or higher. In analysis, answers split into three groups: **Detractors** (0 to 6), **Passives** (7 and 8), and **Promoters** (9 and 10). The **Responses** card shows the NPS score on a gauge from -100 to +100. The score is the percentage of promoters minus the percentage of detractors. Next to the gauge, a 0 to 10 bar chart is colored by group, with each group's share below. In each transcript, Strella tints the score with its group's color. ## Best practices - **Use a slider for ratings, scales, and magnitudes** (satisfaction, 1 to 7 agreement), and **NPS** for likelihood to recommend. Reach for multiple choice only when the options are categorical and unordered. See [Best practices](/build/best-practices). - **Don't stack several structured questions in a row**: mix [open questions](/build/interview-guide) in so the interview doesn't feel like a form. - **Match the type to the data**: a slider for "how much," multiple choice for "which one," ranking for "in what order." ## Tips - **Survey moments interrupt the flow.** Use them where structured data matters (NPS, top-3 ranking, satisfaction), not for everything. - **Rating scales with follow-ups** are the best of both worlds: a clean number plus a real reason. - **Open text in a survey step** is for short typed answers (an email, a one-line description). For longer responses, ask it as an Open question instead: the moderator will get a verbal answer with follow-ups. ## Related The full picture of how survey steps fit in among questions, tasks, and sections. Where survey answers and themes are grouped after interviews come in. --- # Add a tree test Source: https://app.strella.io/docs/product/build/tree-test A tree test is an interactive exercise inside your question guide. The participant reads a task, such as "Your order arrived damaged. Where would you go to report it?", then clicks through an outline of your labels until they reach the place they would expect to find it. It tells you whether your structure and wording lead people to the right place, with no visual design to help or distract them. Tree test is a **Test** element. Like a survey question, the participant answers it. Like a section, it holds follow-up questions of its own, which are asked while the outline stays on screen. See [Step types](/reference/step-types) for how it sits alongside the other elements. ## Steps In [Build](/build/build-screen), click **+ Add element** and choose **Tree test** under **Test**. The element joins the guide like any other, and its properties open on the right. The element's question text is the task. Describe a situation and ask where the participant would go, without using the words of the label you want them to find. Click **Add item** to add a top-level label, and use **Add child** on a row to nest labels under it. Click a row to rename it, and drag rows to reorder them or move them under another label. Every item needs a label. Choose the right answer under **Correct destination**, or use **Set as correct answer** from a row's menu. Each tree test has one task and one correct destination. To test a second task against the same tree, duplicate the element and change the task and destination. **Tree style** sets how the outline reveals your labels: - **Focused** opens like a menu. Opening a section hides the sections beside it, so only the path the participant took stays on screen. This is the default. - **Full** keeps every top-level section visible, and sections expand in place under their parent. The style changes only what the participant sees. Results are measured the same way in both. With **Branches selectable** off, participants answer with a label that has nothing under it, and sections only open and close. Turn it on when a section itself is a valid destination, such as a landing page. A tree test can hold its own follow-ups: open questions, tasks, statements, and survey questions. Drag them inside the tree test the way you'd nest steps in a section. The outline stays on screen while they run, so use them to ask why the participant went where they did. ## What the participant sees The participant first sees a card with the task, and the outline hidden behind it. They click **Click to view** to reveal the outline, then open sections and select the label where they would expect to find what the task describes. They confirm their answer in either of two ways: by clicking **I'd find it here** beside the selected label, or by telling the AI moderator, for example "I'd find it here." Either way, Strella saves the answer, including when the tree test is the last step of the interview. For a participant interviewed in another language, Strella translates the labels before the outline appears, and the confirm button and selection bar use the translated labels. The AI moderator names the buttons in that language too. Your results keep the labels you wrote. ## Reading the results The question's results show how many participants chose the correct destination, how many got there without backing up, the median time on task, and the most common first click. Below them, Strella lists every distinct path participants took, with how many took it and whether it ended at the correct destination. Each participant's transcript also shows their own path step by step. ## Related How to add and arrange every element, and what can nest inside what. The other Test element, for learning how people group your ideas. Every element you can add, grouped by purpose. --- # Upload an existing discussion guide Source: https://app.strella.io/docs/product/build/upload-a-guide If your guide already exists as a document, you don't have to retype it. Drag the file into the [Assistant](/build/build-assistant) and ask the Assistant to rebuild it. Strella reads the document question by question instead of summarizing it. ## Steps In [Build](/build/build-screen), the Assistant is the panel on the right. On mobile, tap the **Assistant** tab at the bottom. Drop the file into the message box, or click the attach icon and pick it. Say what you want, in your own words: "Rebuild this as my question guide" or "Turn this into the guide, keep my wording." The Assistant works through the document and the guide fills in beside it. Edit the guide as you like. Check the places where your document did something a plain question can't say on its own, listed below. ## What Strella keeps A guide written for a human moderator holds more than a list of questions. Strella places each part in your question guide as follows: | In your document | In your question guide | | ------------------------------------------------------------------- | ----------------------------------------------------------------------------- | | Each question | Its own step, in your wording and in your order | | Sections with timings | Sections, in the same order | | Probes listed under a question ("Potential prompts", "Probe on") | Follow-up guidance on that question, so the AI moderator probes the same way | | Notes to the moderator ("Listen for: sweetness") | Kept out of what the participant is asked | | A list of things to rate | One rating question per thing, each on a 1 to 10 scale | | Answer options written beside a question | The options on that question, worded as you wrote them | | A multiple-choice question with no options written down | An open question, so you're not left with an empty list to fill in | Two questions on one line stay two questions. "Which felt appealing? Which felt unique?" becomes two steps, not one. The Assistant reads Word documents (`.docx`), plain text, and Markdown this way. It receives a PDF as-is, so it reads the PDF well but summarizes it more freely than it converts it. For the closest match to your original, upload the Word file. ## Long guides Strella rebuilds a guide of a hundred questions or more one section at a time. The guide takes longer to appear, and it arrives complete instead of thinning out toward the end. Strella truncates very long documents before the Assistant reads them. If the end of your guide is missing, ask the Assistant for the sections that didn't come through, or split the document and upload it in two parts. ## What to check afterwards Read the finished guide top to bottom before you [preview](/build/preview) it. Check these three things: - **Question count.** If your document had forty questions, the guide has about forty. A little more is normal, since one line asking two things becomes two steps. - **Rating scales.** Confirm each one covers what you meant, and change the range if 1 to 10 isn't right for you. - **Anything your document implied rather than said.** Instructions written for a human moderator ("show them the concept board here") may need you to add the [stimulus](/build/stimulus) yourself. --- # Version history Source: https://app.strella.io/docs/product/build/version-history Build keeps a version history of your study. Every time you publish, Build saves that version, and while you work, it takes periodic backups on its own. You can look back at any earlier version, preview it, and restore it if you want it back. ## Open the history In [Build](/build/build-screen), the history control sits in the header above the guide, and hovering it shows when the guide was last edited. Open it and switch to the **History** tab. (The same control lists any **Issues** that need fixing before you can publish.) The tab lists your versions newest first, grouped by month. Each entry shows: - **Published** or **Backup**: whether you published this version or Build saved it in the background. - When it was saved, and a short summary of its size (screener and interview length). - **Currently live** on the version participants are taking right now. A publish made through the API or a connected AI assistant appears here as a **Published** entry too, credited to whoever made it. You can recover a change an assistant published the same way as one you published in Build. ## Backups happen automatically Build saves a backup on its own about 30 seconds after your last edit, so history fills in as you work without you doing anything. To save one right away instead of waiting, use **Create backup** at the bottom of the History tab. ## Editing from more than one place You can edit only one copy of the guide at a time. A second browser tab, a teammate, or a connected assistant can save a newer version of the guide while your page is open. The open page detects this the next time you return to it or make a change, and it stops saving. Build shows a message titled **This guide changed since you opened it** and pauses editing on that page until you choose **Reload latest version**. Reloading loads the newest guide so you can keep working from it. Nothing on the older page overwrites the newer guide. Build saves anything you had changed on that page to the History tab as a **Backup**, so you can open it, compare, and restore it if it held work you want back. The same message can appear after using the browser's Back button to return to Build, since the browser may show an earlier copy of the page. Reload and continue. ## Preview a past version Open an earlier version to preview it. While previewing, a banner across the top tells you which version you're viewing and gives you **Restore** and **Exit preview**. Previewing is read-only: it shows you the older content without changing your draft, so you can compare before deciding. ## Restore a version Restoring brings an earlier version back into the builder as your working draft. Two things keep it safe: - **Your current work is kept.** If you have unpublished changes, Build saves them as a backup first, so restoring never discards your unpublished changes. - **Your live study doesn't change until you publish.** Restoring updates your draft only. Participants keep taking the live version until you [publish](/run/publish) the restored one. ## Related What's editable once a study is live, and how edits reach participants. What "live" means, and how publishing a restored version takes effect. --- # Changelog Source: https://app.strella.io/docs/product/changelog Strella ships continuously. This page captures release highlights, newest first. ## August 25, 2026 This release adds a similarity matrix to card sort analysis, lets you show an in-call question beside its stimulus, and pulls every interview's transcript into one download. **See which cards participants grouped together** Card sort results gain a **Similarity matrix** tab that shows, for every pair of cards, the share of participants who placed both in the same group. The rows and columns are ordered so cards people agree belong together form dark blocks along the diagonal, surfacing the natural clusters in your participants' mental model at a glance. **Show a question beside its stimulus in split view** A per-project **Split view** setting gives an in-call question its own column next to the image, video, prototype, or website it asks about, instead of a card floating on top of it. Participants can read the question and study the stimulus at the same time, which helps on dense screens where a card would cover the very thing you're asking about. **Download every transcript in one file** The Results **⋮** menu gains **Download all transcripts**, which gathers every interview's transcript into a single text file. Each interview is labeled with the participant, numbered when they have more than one session, and marked Completed or Partial, so a whole study reads in one document. ## August 18, 2026 This release lets you download a report as a Word document, refreshes the report viewer, and records question-guide publishes made through the API or a connected assistant in version history. **Download a report as a Word document** The report's Download menu now offers Word (.docx) alongside PDF. Citations come across as endnotes and your brand logo is carried over, so you can open the report in Word and finish the last bit of polish by hand before it goes to stakeholders. **A refreshed report viewer** Reports now read as long-form documents rather than a long chat message, with document-scale headings and spacing. The cover image sits inset in the text column with an Update cover control on it: upload your own image, or regenerate a new one. **Publishes from the API or an assistant show up in version history** A question-guide publish made through the developer API or a connected AI assistant, like Claude, now lands a restorable entry in Build's version history, attributed to whoever made it, so an agent-made change is recoverable just like one you published yourself. ## August 11, 2026 This release adds device and audience filters to analysis, lets you hold a prototype still until participants answer, and lists the Strella connector in the Claude and ChatGPT directories. **Filter analysis by device and how participants were recruited** Analysis and the Interviews list gain filters for the device a participant used (desktop web, mobile web, or mobile app) and for the exact share link or panel order they came through. Stack as many rules as you need; each option shows a live count of how many participants it matches, and the filtered view travels in the page link, so a teammate opens exactly what you shared. **Hold a prototype still until participants answer** On any question or task inside a Figma prototype or Embed block, an Advanced setting, Block interaction until answered, dims the media and holds it while that question is on screen. Participants give their first impression before they start clicking around, and the hold lifts the moment the interview moves on. **Find the Strella connector in the Claude and ChatGPT directories** Strella is listed in the Claude and ChatGPT connector directories, so you can add it by finding it and clicking Connect instead of pasting in a server URL. The custom-connector setup still works for workspaces that block directory connectors. ## August 5, 2026 This release requires a sign-in to open the files a study holds about its participants, and stops a deleted project's media resolving anywhere. **Participant uploads and chat attachments need a sign-in** A photo or video a participant uploads as an answer, and any file you attach to a project chat, now open only for someone signed in to the workspace that owns the study. A link that reaches anyone else shows a "not found" page. The materials you show participants during an interview (stimulus images, consent forms, terms documents) still open for anyone with the link, because the people who read them have no account to sign in with. See [Media links and who can open them](/account/media-links). **Deleting a project switches off its media links** Deleting a project stops every file it holds from opening, immediately and for everyone. That includes the links that need no sign-in, so a consent form already sent to a recruiting vendor stops working too. Restoring the project restores its media. **Moving a project takes all of its media with it** A project moved to another workspace hands over its session recordings along with everything else. The workspace the study ran in can no longer open them. ## August 4, 2026 This release redesigns the account settings pages and the home projects list, adds pass-rate and time-on-task charts to task questions, and lets a highlight reel pull clips from more than one project. **Account settings, redesigned as searchable tables** Activity logs, Billing, Developer, and participant data access each open as a searchable, filterable table on one shared settings layout, and Integrations moves onto its own settings page instead of a pop-up. The Developer table now shows who created each API key, so you can tell keys apart at a glance. **Pass rate and time on task on every task question** A task question in Results leads with two charts that summarize the whole group: a pass-rate donut showing the share of participants the AI judged complete, banded green, amber, or red by how often people fell short, and a time-on-task chart with the median duration and its spread. Both export to PNG, and the pass rate exports to a spreadsheet with a per-participant breakdown. **A refreshed home for your projects** The home page and each workspace show your projects as one card list. Every project card carries its owner's avatar, and hovering the completed count breaks it down into qualified, invited, and not-qualified participants. **Highlight reels can span multiple projects** A highlight reel filed in one project can include clips recorded in others. Foreign clips play alongside the rest and carry a chip naming the project they came from, and the reel's counts and its shared link account for every clip. ## July 30, 2026 This release brings outside interviews in up to ten files at a time, lets a video stimulus stand on its own, and teaches the Strella connector to review your question guide. **Upload up to 10 transcripts or notes at once** When you bring externally-recorded research into Strella, you can drop up to ten transcripts or notes into the upload dialog in one go instead of adding them one file per trip through the menu. Supported formats are .txt, .md, .docx, .vtt, and .srt. If a batch would push you past your participant limit, the dialog now tells you instead of failing quietly. **A video stimulus can stand on its own** A video you show participants no longer needs a question attached to publish. Add a video on its own and participants watch it, say when they're done, and the interview moves on. Image, website, embed, and prototype blocks still need at least one question or task, since there's nothing else to hold the participant's attention on them. **The Strella connector can review your question guide** Ask an AI assistant connected through the Strella connector to check a project's question guide against its research objectives. It returns per-objective coverage and prioritized suggestions for closing the gaps, and leaves the guide itself unchanged. ## July 28, 2026 This release makes the codings you build on open-ended questions read clearly on their own. **Coded responses carry their own titles** When you code an open-ended question, the coding now has a plain-language title and a one-line description that label its chart, so it stands on its own in an export or when a question has more than one coding. Edit both from the coding's dialog, alongside renaming its categories and dragging participants between them. ## July 15, 2026 This release moves your recruitment settings onto individual share links and makes filtered dashboard views shareable by link. **Recruitment settings now live on each share link** Each share link carries its own settings: whether to run the screener, collect an email or name, auto-invite qualifying participants, send reminders, and cap how many participants it accepts, so you can give different audiences different rules from a single study. A link follows the study default until you change it, and you can close one link to new responses without closing the whole study. Find them under Settings → Access & scheduling → Share links. **Shareable, hand-editable dashboard views** Filters in the projects dashboard URL now take precedence over your saved preferences, so a link with filters opens exactly the view you shared instead of being overwritten by the viewer's own filters. Simply opening a shared link never rewrites your saved preferences. ## July 8, 2026 This release renames Sessions to Interviews across Strella, reorganizes project settings into grouped cards, and lets returning participants pick up where they left off. **"Sessions" are now "Interviews"** Every recorded conversation is called an Interview throughout Strella. The rename runs across the Results tab, the sort and filter menus, empty states, data exports, and the emails you receive, so the product matches the way you already talk about your studies. **Project settings, reorganized into grouped cards** Project settings open as four groups of cards (General, Interview setup, AI moderator, and Advanced) so related options sit together and you can find a setting without scanning every tab. The teammates who receive email notifications for a study now live in their own Notification settings modal, opened from the notification bell. **Returning participants resume where they left off** When someone re-enters a screener or interview with the same email, Strella recognizes them and offers to continue where they stopped or start over, instead of beginning a brand-new attempt each time. A qualified participant who was already invited goes straight back into their interview. ## July 3, 2026 This week the redesigned question guide builder became the default for everyone, with more flexible section logic and interview time estimates that reflect your content. **The redesigned question guide builder is now the default** Every project opens in the redesigned question guide builder, where your interview guide and screener live together in one place. **Match several answers in one section logic condition** A single section logic condition on a multiple-choice question can name several answers at once, showing the section when a participant picked any of them, instead of adding a separate condition for each answer. **Interview time estimates that reflect your content** The estimated interview length in the builder accounts for how much you're asking, including prompt length, number of choices, follow-up depth, and attached media, so the estimate tracks the real interview instead of a fixed per-question guess. ## June 12, 2026 This week we shipped bulk highlight reel downloads, filters for your AI Chat history, exact participant counts and bulk invites in the Recruit table, and audit logs for organization admins. **Download all your highlight reels at once** You can now download all of a project's saved highlight reels together as a single ZIP file from Project → Results → Saved Reels. **Filter your AI Chat history by project and favorites** The AI Chat sidebar now lets you filter your chat history by project or show only your favorited chats, and it remembers your filter as you move between chats. **Exact participant counts and bulk invites in the Recruit table** The Recruit table now shows exact participant counts past 99 on every tab, including All and Not Qualified, and you can bulk-invite participants directly from the Not Qualified tab. **Audit logs for organization admins** Organization admins can now open an Audit logs tab in organization settings to review account activity, set the organization's logging behavior, and request deletion of logs. ## June 5, 2026 This week we shipped a full highlight reel editor in AI Chat, new filters and saved views across Analysis and Sessions, full-page Excel exports from Analysis, and automatic tool selection in AI Chat. **Trim and reorder highlight reel clips in AI Chat** Highlight reels built in AI Chat now have a full editor in the sidebar: reorder clips, trim each clip's start and end, remove or restore clips, and save the result back to your project, all without leaving the chat. **New filters and saved views across Analysis and Sessions** Analysis and Sessions add three new filters (Author, Session Quality, and Sentiment), and every filter dropdown now shows how many participants match each option before you apply it. Your filters and sort also save per project and travel in the URL, so you can reopen or share the exact view you're looking at. **Full-page Excel export from Analysis** You can now download an entire Analysis page as a multi-sheet Excel workbook with your current filters applied, so the file's structure matches what you see on the page. **AI Chat picks the best tool automatically** AI Chat now automatically selects the best tool for your prompt (a chart, a highlight reel, or a report), so you don't have to choose one yourself. ## May 29, 2026 This week we shipped a new Rating scale question type, the ability to hide sessions from analysis, AI Chat project filtering by workspace, more reliable call recordings and study entry, and accurate order completion counts. **A new Rating scale question type** You can now add a Rating scale question to your Question Guide to capture sentiment, with three styles to fit the use case: Likert (strongly disagree to strongly agree), faces, or stars. Multiple-choice questions also gain selection limits, so you can require a minimum number of selections or cap the maximum. **Hide sessions from analysis** You can now exclude a low-quality session from analysis without deleting it, straight from the session menu. Excluded sessions stay in your session list but are filtered out of analysis views, theme summaries, AI charts, and AI Chat quotes, and the toggle is reversible at any time. **Filter AI Chat projects by workspace** The project picker in AI Chat now lets you filter projects by workspace and flags projects that have no sessions yet, so you don't pick an empty one. Helpful for organizations that spread their work across many workspaces. **More reliable call recordings and study entry** Calls now recover a recording when the start signal is dropped during connection, so fewer interviews finish with a missing recording. Separately, an error that could block participants from entering a study or joining a call has been resolved, so participants reach their interview reliably. **Accurate order completion counts** Recruitment orders now count only fully completed sessions toward an order's total, so partial sessions no longer make an audience look full and close it before you've reached your target. ## May 22, 2026 This week we shipped automatic flagging of sensitive (MNPI) disclosures in Analysis, more consistent filtering and sorting across Strella, and a unified popover for managing table columns and custom fields. **Flag sensitive (MNPI) disclosures in Analysis** Analysis now scans interview transcripts for material non-public information (MNPI) and shows an informational warning when a participant may have disclosed something sensitive, so teams in regulated industries can catch it during review. **Consistent filtering and sorting across Strella** Filtering and sorting now work the same way across notes, responses, sessions, and the participant table, with the same controls and behavior wherever you work with your data. **Manage table columns and custom fields in one place** A single Manage table popover now controls which columns are visible and lets you add or edit custom fields directly, with the same controls across your participant and responses tables. ## May 15, 2026 This week we shipped copyable participant call links, default captions with a redesigned overlay, mid-call screen switching, and improved call reliability on slower networks. **Copy a participant's unique call link** A new Copy call link option in each participant's Share menu copies their unique call URL with all custom fields pre-filled, ready to paste into an email or message without manually constructing it. **Default captions and redesigned caption overlay** Project settings now include a Default captions on toggle that turns captions on automatically when participants join a call. The caption overlay has been redesigned too: it's draggable, shows each speaker's words in a persistent card that changes color by speaker, and no longer flickers during transitions. **Change shared screen during a call** Participants can now switch the screen or window they're sharing mid-call without stopping and restarting their share. A new Change shared screen option appears in call settings while screen sharing is active. **More reliable call setup on slow connections** Calls now wait for a fully established connection before recording begins, preventing the "We had trouble setting up your call" error that would redirect participants on slower networks. Calls that fail to connect now stay on the connecting screen instead of ejecting. **Sort the participants table** A new Sort button lets you sort the participants table by any column (including session length), with a searchable field picker and configurable direction. ## May 8, 2026 This week we shipped time-on-task charts in Analysis, multiple per-project share links, a combined session export, and pinnable favorite projects on the dashboard. **Time-on-task charts in Analysis** Task questions in Analysis now show a box plot of participant completion times (median, quartile spread, and outliers) with moderator time automatically excluded from the calculation. An Outliers toggle lets you show or hide extreme values to focus on the core distribution or flag participants who ran long. **Multiple share links per project** Projects can now have multiple share links, each with its own redirect URL for when participants finish, get screened out, or hit a full quota. Useful for routing different audiences to different destinations, or for running the same study with multiple entry points without duplicating the project. **Combined session export** The Sessions page now exports all session data (demographics, transcripts, and survey responses) into a single Excel workbook instead of requiring multiple separate downloads. **Favorite projects on the dashboard** Pin any project to the top of your dashboard by marking it as a favorite, and filter the project list to show only your favorites. Favorites are saved per user, so each team member keeps their own pinned set independently. **Background blur for participants** Participants can now blur their background during calls, with a slider to control the intensity. ## May 1, 2026 This week we rebuilt Analysis around filter-based segmentation, added a faster way to build highlight reels directly from the Analysis section, shipped Excel exports for participant tables, and started sending daily participant payment reminders. **Filter-based segmentation in Analysis** Analysis segmentation now works through filters. Stack multiple filters across demographics, screener answers, quotas, and variants *(segments no longer need to be mutually exclusive)*, and the participant list updates as you go, with the active filters applied to downloads. Filter-based segmentation in Analysis **Build highlight reels from Analysis** From any Question in the Analysis section, you can now create highlight reels by selecting words in a participant's transcript and saving the quote to a new or existing reel. Building highlight reels directly from Analysis **Excel exports for participant tables** Participant tables can now be downloaded as XLSX in addition to CSV, with a separate sheet per participant tab so the structure of the export matches what you see in the app. **Daily payment reminder emails** Researchers now receive a daily email listing sessions scheduled for auto-payment the next business day, giving you a window to flag concerns before credits go out. This pairs with the recently shortened payout window, where panel-recruited participants are now paid three business days after a successful session. ## April 24, 2026 This week we shipped two new question types (Slider and Free Response), AI-generated report covers, chart exports for Analysis, automatic thumbnails for uploaded recordings, and a full preview mode for conditional Question Guides. **Slider and Free Response questions** Question Guides now support two new question types for everyone: **Slider** (numeric scale with configurable min/max, step size, and endpoint labels) and **Free Response** (open-ended text input). Both work in the builder, in live interviews on web and mobile, and show up in Analysis alongside your other survey responses, with translation support across 40+ languages. AI generation and paste flows now produce these types automatically. **AI-generated report covers** Reports now ship with AI-generated cover images, and you can regenerate a fresh cover or upload your own (up to 4MB) directly from the report or from AI Chat. Each "Re-roll" asks for a distinctly different style, so you don't end up with near-duplicate variations. **Export Analysis charts as images or Excel** Charts in Analysis can now be downloaded as PNG images or as Excel files. The Excel export includes the underlying formulas and a summary, so you can drop charts into a deck or keep working with the data in your own spreadsheet. **Thumbnails for uploaded session recordings** Uploaded MP4 recordings now generate thumbnails automatically and show them on session cards in Results, just like sessions recorded inside Strella. No more grey placeholders for imported content. **Full preview for conditional Question Guides** Previewing a guide with conditional logic now walks through every section, even ones that would normally be skipped based on screener or in-call answers. A new pre-preview dialog tells you when conditional logic is in play and points you to the share link if you want to test the real participant routing. ## April 17, 2026 This week we shipped project-aware AI Chat suggestions, dashboard filters, participant custom identifiers for deduplication, and streamlined participant cards in the Analysis section. **AI Chat suggestions for your selected project** AI Chat now shows context-aware prompt suggestions when the chat is empty, tailored to whichever project you have selected. When no project is selected, default suggestion tiles appear so you still have a starting point. **Project filters on the dashboard** The projects dashboard now supports filtering by status, creation date range, and participant count. Filters persist across sessions, and sorting and pagination stay consistent with infinite scroll. **Participant custom identifiers for deduplication** You can now mark a custom column (employee ID, CUID, etc.) as the unique identifier for a project. Strella auto-matches incoming participants to the existing imported record instead of creating a duplicate Anonymous participant. **Streamlined participant cards in Analysis** Participant cards on the Objectives and Questions tabs no longer include inline transcripts and mini video players. Clicking anywhere on a card now opens the full session modal, cued to the exact timestamp for that question or objective, so you jump straight to the right moment. ## April 10, 2026 This week we shipped a redesigned participant table, automatic participant payments, and AI-powered participant ratings for new projects. **Redesigned participant table for new projects** New projects now use an upgraded participant management table with custom columns, inline field editing, and the ability to do CSV imports. **Automatic participant payments** Participants in new projects are paid automatically 7 days after their session. If something went wrong, you can challenge the session within that window to get a refund, so there are no more manual approvals to release payments. **AI-powered participant ratings and auto-invite** Panel participants now receive a 1 to 5 star AI rating directly in the participant table, scoring research fit, consistency, and engagement so you can spot strong candidates at a glance. Turn on **Auto invite** to automatically invite top-rated panelists and eligible self-recruited participants. ## April 3, 2026 This week we shipped richer Question Guide exports, faster Question Guide previews, keyboard shortcuts for video playback, and custom redirect URLs for participant routing. **Richer Question Guide exports** Question Guide exports (DOCX, TXT, and clipboard) now include screener questions, quotas with criteria summaries, and skip logic, making it easier to share your full study setup with anyone outside of Strella. **Preview Question Guide from any question** You can now start a Question Guide preview from whichever question you have selected, instead of always starting from the beginning. This makes it faster to test specific sections of your guide. **Custom redirect URLs for participant routing** Participants who are screened out, drop out, or hit a full quota can now be redirected to custom URLs. Each redirect forwards the participant's custom fields so your downstream tools stay in sync. **Keyboard shortcuts for video playback** The video player now supports spacebar to play/pause and arrow keys to adjust volume, so you can review sessions without reaching for on-screen controls. ## March 27, 2026 This week we shipped a Figma Screens tab for prototype analysis, translated survey options for multilingual interviews, and explicit task confirmation as a new default. **Figma Screens tab for prototype analysis** Individual screens from your Figma prototypes now have a dedicated Screens tab alongside the existing Paths, Completed, and Failed views. Each screen shows how many participants visited it, their average time on screen, and misclick rate, so you can quickly spot which screens need attention without scrubbing through full session replays. **Translated survey options in multilingual interviews** Survey answer choices now display in the participant's selected language during live interviews. If a translation is unavailable, the original text is shown as a fallback, so multilingual studies run more smoothly without manual workarounds. **Explicit task confirmation as default** New Question Guides now default to requiring spoken confirmation from the participant before the AI moderator moves on to the next task. This prevents the moderator from advancing too early and gives participants a clear moment to add anything before continuing. ## March 20, 2026 This week we shipped a unified AI Chat experience, stronger participant anonymization, more precise AI moderator controls, and more shareable reports. **Unified AI Chat** AI Chat is now one unified experience instead of separate Analyze, Create Charts, and Create Reports modes. You can create charts, highlight reels, and reports from the same place, and citations now open inline so you can preview the underlying user recordings more easily. **Stronger participant anonymization** You can now turn on name anonymization in Project Settings to replace participant names with random aliases throughout Strella. Anonymization carries through transcripts, analysis, reports, CSV exports, and session imports, and also hides emails and social links to better protect participant identity. **Per-question verbatim control in Adaptive mode** You can now keep the AI moderator in Adaptive mode overall while marking specific questions to be asked exactly as written. This gives you more control over critical wording without making the whole interview fully strict. **More shareable reports** Shared report pages and PDF reports now support shareable video links, so it’s easier to send evidence-backed findings with direct access to the underlying clips. ## March 13, 2026 This week we shipped in-app notifications, smoother participant data imports and exports, and better citation highlighting for non-English transcripts. **In-app notifications** You can now see project notifications directly inside your Strella study with a new notifications bell, making it easier to stay on top of things like interviews waiting for review or studies that are still in draft mode and haven't been launched. **Participant data import and export improvements** You can now manage participant data more reliably across import, participant table review, and export. Custom fields like UTM parameters are detected more consistently, appearing automatically as columns in the participant table, and are included in CSV downloads, while imports better map demographic fields and prevent duplicates. **Better citation highlighting for non-English transcripts** Citations now highlight the exact matching text more accurately in non-English transcripts, making it easier to review sessions. ## March 6, 2026 This week we shipped faster highlight reel downloads, new privacy blur controls for sensitive interviews, configurable AI moderator reaction length, and more accurate Figma heatmaps. **Faster highlight reel downloads** You can now choose between a fast highlight reel download for quick turnaround and a more precise export when exact clip boundaries matter more than speed. Fast mode is now the default, while precise exports continue through the slower email-based flow. **Privacy blur controls for sensitive interviews** You can now blur participant camera or screen share for privacy, either across the whole interview or only on specific questions. This makes it easier to protect sensitive content without changing the rest of the session experience. **AI moderator reaction length controls** You can now control how concise or detailed the AI moderator sounds, so interview responses better match the tone and pacing you want. **More accurate Figma heatmaps** Figma heatmaps now scroll more smoothly and register clicks on fixed elements like headers and navigation more accurately. This makes interaction patterns easier to trust when reviewing embedded prototypes. ## February 27, 2026 This week we shipped new AI question behavior controls so you can choose between Strict (reads your guide exactly) and Adaptive (allows minor phrasing changes for a more natural flow), plus more reliable AI Chat on long transcripts and variant-based analysis segmentation. **AI question behavior controls** You can now choose whether the AI moderator follows your question guide in Strict or Adaptive mode. Strict reads questions exactly as written in your guide, while Adaptive can make minor phrasing changes to keep the participant experience more natural. **More reliable AI Chat for long transcripts** AI Chat now handles very large transcripts more reliably, reducing errors and mismatches when you explore long sessions. **Analysis segmentation by variant** You can now segment analysis results by variant groups with a dedicated Variant filter on the Analysis page. It appears when interviewees are assigned mutually exclusive, collectively exhaustive variants, and helps you compare outcomes across variants in one view. **Clearer language labels in settings** Language selectors now show full readable names like "German" instead of short codes, making language settings easier to scan and less ambiguous. ## February 20, 2026 This week we shipped more reliable uploaded-session analysis, clearer screener language controls, and better recruiting workflow feedback. **More reliable uploaded-session analysis** Uploaded participant sessions now complete analysis reliably even without a Question Guide in the study, so your Objectives section and Highlight Reels update as expected. **Screener language controls** You can now explicitly choose screener translation language when ordering panels, and Strella better detects the actual language used in Question Guide and screener content for more accurate translation and matching. **Recruiting workflow updates** - Participants are now auto-disqualified when quotas are full to prevent over-recruitment. - Bulk invites now show real-time progress and update participant invite status incrementally as sends complete. **Safer Question Guide editing** You’ll now see a warning if you try to close a Strella tab with unsaved changes. **Mirrored webcam self-preview** Webcam self-preview is now mirrored, so participants seeing themselves on camera feels more natural. ## February 13, 2026 This week we shipped research document import for faster project setup, task timers for User Interviews-recruited participants, refreshed voice and audio defaults, and brought session tags back. **Research document import with live preview** You can now upload a research document and have Strella auto-populate your interview guide steps. As the AI generates outputs, the preview panel updates in real time so you can watch your guide take shape as it’s created. **Task timers in interviews (User Interviews panel)** Participants recruited through the User Interviews panel now see a task timer during interview tasks, making it clearer how long they have to complete each activity. **Voice and audio settings improvements** We made several improvements to make voice configuration clearer and more consistent across projects: - **Curated voice options** with clearer choices, plus gender-based moderator voice selection. - **Updated defaults for new projects**: new projects now start with English preselected by default. **Session review improvements** The session modal now supports session tags again in the Profile tab, with tag styling that matches other badges in the UI. We also added the ability to download individual clips which you can do so directly from a note's menu. **Quality-of-life improvements** - **Dynamic follow-ups default**: set a project-wide default for follow-up question depth from Question Guide settings. - **Improved publishing feedback**: publishing now updates the UI immediately without requiring a page reload. ## February 6, 2026 This week we shipped expanded language support across 42 languages, skip logic for screeners, and more accurate session timing estimates. **Expanded language support** Strella now supports 42 languages for conducting research. The full list includes: Belarusian, Bosnian, Bulgarian, Catalan, Croatian, Czech, Danish, Dutch, English, Estonian, Finnish, French, German, Greek, Hindi, Hungarian, Indonesian, Italian, Japanese, Kannada, Korean, Latvian, Lithuanian, Macedonian, Malay, Marathi, Norwegian, Polish, Portuguese, Portuguese (Brazil), Romanian, Russian, Serbian, Slovak, Slovenian, Spanish, Swedish, Tagalog, Tamil, Turkish, Ukrainian, and Vietnamese. **Skip logic for screeners** You can now set up skip logic conditions in your screeners. If a participant selects a specific answer, they can automatically jump to a different question, letting you create branching screener flows that adapt to each respondent. **More accurate session timing estimates** The estimated session duration shown to participants is now more accurate. Timing now accounts for screener length and, for studies with variants, only counts the questions a participant will actually see. ## January 30, 2026 This week we shipped an AI-guided project setup flow, a new Billing page with credits and transaction history, a new recruitment funnel view, and transcript search in the session modal. **AI Chat guided project setup** You can now create a research project through an AI guided chat flow. The chat will draft your question guide, screener and draft audiences so you can get from idea → launched project faster. **New Billing page (credits + transaction history)** Strella now has a dedicated Billing page where you can purchase credits and manage your organization’s billing in one place. From Billing, you can: - **Buy credits** up front and use them across projects (including participant recruitment fees and incentive payments) - **Review Transaction History** for every credit add/spend, with filters and CSV export for accounting You’ll also see your current credit balance in the top-right navigation across the Strella platform if your balance is above $0. **Recruitment funnel breakdown** The Recruit tab now includes a Recruitment Funnel view (Recruit → ⋮ menu → Recruitment funnel) to help you understand drop-off through the recruitment process (self-recruited or panel). It also includes pass-rate detail for each screener question so you can quickly spot where candidates are failing. **Transcript search in the session modal** The session modal transcript now supports searching with Ctrl+F / Cmd+F, matching native browser search behavior. Use Enter / Shift+Enter to jump between matches, and Escape to close search. ## January 23, 2026 This week we shipped video layout selection in session and insight modals, demographic-based segmentation in Analysis, clearer transcript media labels, and improvements to the participants table recruiting workflow. **Choose video layout in session and insight modals** Select how you want to view recordings in session and insight modals with layout options including split-screen, screen share, and user camera. This preference is saved per user. **Segmentation by demographic data** Analysis segmentation now supports demographic data and attribution (like UTM parameters), in addition to existing segmentation options of screener responses, survey questions, and quotas. **Media item labels in transcripts** Media items in the session modal transcript now display their labels instead of raw URLs, making them much easier to identify at a glance. **Improved participants table recruiting workflow** Participants are now sorted with uncompleted orders first, and completed order rows are visually grayed out so you can quickly spot what still needs action. We've also added a tooltip to "Invite Participant" to explain when you need to extend your audience to recruit more participants. ## January 16, 2026 This week we shipped a Chrome picture-in-picture window, faster quota fetching, improved sharing permission troubleshooting, and automatic cleanup of insignificant sessions in Results and Analysis tabs. **Chrome picture-in-picture window** Participants interviewing in a Chromium browser can now open a picture-in-picture window when they navigate to a different tab, letting them keep tasks and surveys in view while they work through them. **Faster quota fetching** Fetching quotas on the Quotas page is now significantly faster, avoiding the freezes some large projects were experiencing. **Improved sharing permission troubleshooting** Strella now detects when browser permission settings are blocking screen sharing, microphone, or camera access. If permissions are blocked, we’ll guide participants on how to unblock them so they can continue the interview. **Automatic deletion of insignificant participant sessions** Insignificant sessions are now automatically deleted and won’t count toward session limits or quotas. ## January 9, 2026 This week we shipped Quant Explorer, a new AI mode for charts and highlight reels, a redesigned transcript experience, session modal enhancements with auto-saved playback and URL timestamps, and granular participant info settings. **Quant Explorer mode** A new AI chat mode lets you explore your data quantitatively. Ask Quant Explorer to create charts or generate highlight reels. Results appear inline in the conversation. **Redesigned transcript experience** The transcript viewer has been rebuilt with a cleaner, faster interface: - **In-call surveys inline**: Survey responses now appear directly in the transcript flow - **Clickable timestamps**: Click any timestamp to jump to that moment in the video - **Faster scrolling**: Improved virtualization makes long transcripts scroll smoothly - **Consolidated speaker labels**: Consecutive utterances from the same speaker are visually grouped - **Updated selection toolbar**: Select text to copy, create highlights, or jump to that timestamp **Session modal enhancements** We've added several quality-of-life improvements to the session modal: - **Highlights**: "Notes" have been renamed to "highlights" across the platform - **Resume playback**: Video and audio playback position is automatically saved and restored when you return - **Click-to-jump**: Click any word in the transcript to jump directly to that moment - **Shareable timestamps**: Share session links with a timestamp parameter so they open at specific times. **Explorer timestamps** Explorer chat now links directly to specific moments in sessions using clickable timestamps instead of text quotes. **Granular participant info settings** Name and email collection are now separate settings. Previously you could only toggle both together. Now you can require just a name, just an email, or neither. Both are enabled by default. **Open sessions in new tabs** Ctrl/Cmd+click or middle-click on session cards to open them in a new browser tab. ## January 2, 2026 This week we shipped new Summary and Profile tabs in the session modal, streamlined AI moderator survey instructions, smarter screen share verification, and recording activity tracking for quality checks. **Session modal Summary and Profile tabs** The session modal sidebar now includes two new tabs as part of ongoing session modal upgrades. The Summary tab shows an AI-generated summary based on your Project Objectives, while the Profile tab displays participant profile information. **Streamlined survey instructions** For survey questions, the AI moderator now reads the full instructions only for the first question. Subsequent survey questions skip the repeated instructions, creating a smoother and less repetitive experience for participants. **Smarter screen share verification** Screen sharing is now more reliable during onboarding. Strella confirms that participants are sharing the correct screen by checking whether the Strella window is visible in their screenshare. **Recording activity quality checks** Quality check results now include a Recording Activity criterion that tracks camera usage throughout the session. This helps identify when participants disabled their camera and for how long, giving you more insight when reviewing session quality. ## December 26, 2025 This week we shipped email notifications for project events. **Email notifications for project events** Organization members can now receive email notifications triggered by key project moments. Configure this in Project Settings > Access & scheduling > Project Email Notifications. ## December 19, 2025 This week we shipped a redesigned video player with thumbnail previews, a mini player with control parity, and improved sharing. **New video player** We've rebuilt the video player to provide a better viewing experience. Hover over the timeline to see thumbnail previews, and pop out into a mini player that now has full control parity with the main viewer. Sharing videos got smarter too: choose the default playback speed, toggle subtitles, and set whether shared reels start at the current time or from the beginning. ## December 12, 2025 This week we shipped a redesigned settings panel, smarter participant approvals with AI-powered recommendations, session thumbnails with hover previews, and a refreshed AI Chat UI. **Redesigned settings panel** Settings that belong together are now grouped into clear sections: - **Project setup** (research plan, objectives, moderator language) - **Recruitment** (scheduling, share links, appearance) - **Interview** (setup and AI moderator options) - **Advanced** (AI models and experimental features). We've also added new options including screener control for share links, custom closed messages, and the ability for participants to pause screen sharing. **Smarter participant approvals** Strella now reviews each participant session and recommends whether to approve or reject the participant. Hover over any recommendation to see the reasoning, then approve or reject right from the session modal. Suggested actions are also visible in the participant table view. **Session thumbnails at a glance** Each participant session and analysis view now includes a thumbnail. Hover over it to see a gif of the entire session, giving you a quick visual summary of the participant experience. **Refreshed AI Chat interface** The AI Chat interface has been refreshed with a cleaner, more polished look. ## December 5, 2025 This week we shipped our mobile app globally across iOS and Android, added an AI copilot for question guide creation, and streamlined the session modal experience. **Run studies natively on mobile** The Strella mobile app is now available in 177 countries on iOS and Android. From exploratory research to mobile usability tests: run any type of research directly on participants' devices. **Create question guides with our AI Copilot** A new copilot in the question guide editor lets you build and iterate on your guide through conversation. Start from scratch or expand an existing guide by chatting with the AI directly. **Redesigned session modal** The session modal got a refresh: collapsible sidebar, click-to-rename sessions, next/previous navigation between participants, and a cleaner layout overall. **Faster participant acceptance** Accept or decline participants without leaving the session modal. Cycle through sessions in one flow instead of navigating back and forth. ## November 28, 2025 This week we shipped a centralized notes page in Results and question-based filtering in the Responses table. **All your notes in one place** The new Notes page in Results collects every note across your sessions, filterable by creator, sentiment, source, participant, and session tag. Qualitative analysis just got easier. **Filter responses by question** The Responses table now includes a Question column, so you can see which transcript segments tie back to your question guide. Filter to show exactly what you need across all sessions. ## November 21, 2025 This week we shipped dynamic follow-up visibility in the question guide tree, word-level transcript highlighting, bulk participant disqualification, and smoother transcript selection. **See dynamic follow-ups at a glance** The question guide side panel now shows which questions have dynamic follow-ups, including their level and the follow-up prompt. **Word-level transcript highlighting** Follow along with interview transcripts word by word. As audio plays, each word highlights in real-time so you know exactly where you are in the conversation. ##‍ Bulk disqualify participants Select multiple participants and disqualify them all at once instead of having to do it one by one. ##‍ Smoother transcript selection Selecting portions of the transcript is now faster and more fluid, making it easier to create your own interview notes. ## October 10, 2025 This week we shipped reusable audience templates for panel recruitment and full translation support across the platform. **Save and reuse audience settings as templates** You can now save your Recruitment Audience settings as templates and apply them when creating new audiences. Access saved templates from the Hire Participants flow or manage them from the Templates view. **Full translation support** Set a Researcher Language for your question guide and a Participant Language for how content is presented to participants. All study-related emails, screeners, and in-call screens are now fully translated, on both desktop and mobile. --- # Custom dictionary & pronunciation Source: https://app.strella.io/docs/product/configure/custom-dictionary The custom dictionary teaches the AI moderator how to say words it would otherwise mispronounce: product names, acronyms, people's names, industry jargon. It lives in **Settings → Custom dictionary**. ## Steps Click the gear icon, then **Custom dictionary**. For any word the AI mispronounces, add a row: - **Word**: the word as written, e.g. `Strella`. - **Pronunciation**: how the word sounds, spelled the way someone would read it aloud, e.g. `STREH-luh`. The moderator uses these when reading questions and reacting. Changes are held in your draft. Click **Save** to apply them. If you close Settings with unsaved edits, Strella asks whether to keep editing or discard the changes. Pronunciation changes apply to new interviews. ## Tips - **Spell it the way it sounds, not in phonetic alphabet.** Write `MAY-kuh`, not `/ˈmeɪ.kə/`. - **Test the pronunciation in [Preview](/build/preview).** A few minutes of preview tells you whether the entry actually worked. ## Related The rest of the moderator's voice settings. How the moderator reads your guide aloud. --- # How the AI moderator works Source: https://app.strella.io/docs/product/configure/how-the-ai-moderator-works The AI moderator is the voice that conducts each interview. It opens with a built-in greeting, walks through every item in your guide, asks dynamic follow-ups where you've enabled them, and closes the interview with a built-in goodbye, all over voice. ## The interview lifecycle When a participant joins, the moderator follows a fixed lifecycle: 1. **Built-in greeting.** The moderator opens with a fixed line: "Hi, I'm [moderator name], welcome to the interview." You can [rename the moderator](/configure/voice-language-behavior) from the default "Strella," but you can't edit or remove the greeting itself. If you want scripted framing in your own words, add a **Statement** as the first item in your guide — the moderator reads it right after greeting the participant. 2. **Walks through your guide top to bottom.** Each item plays out in its own way: - **Open questions and tasks**: asked conversationally; **Ask question verbatim** forces exact wording. - **Statements**: read aloud exactly as written, then move on. No response collected. - **Survey questions**: appear on the participant's screen; the interview advances only when the participant answers. The moderator can't answer for them or skip past a survey question on its own. - **Media blocks**: the media appears on screen, stays for the whole block, and the moderator works through any questions you attached. 3. **Asks follow-ups.** Each Ask-group item has a **Follow-up depth** setting (Off / Low / Medium / High ≈ 0 / 1 / 2 / 3 follow-ups). **Follow-up guidance** tells the moderator what to probe for ("dig into pricing objections"). The moderator composes follow-up wording live, so you can't script the exact text. 4. **Built-in closing.** After the last item, the moderator automatically thanks the participant and lets them know they can hang up. You can't customize or remove this. If you want a thank-you in your own words, add a **Statement** as the final item, and it plays just before the automatic goodbye. ## Returning participants If a participant leaves and comes back later, the moderator welcomes them back, the interview resumes where they left off, and any randomization or section-targeting assignments stay the same: they see the same variant and the same conditional sections they were on track to see. ## How participants interact - The moderator listens continuously while it's the participant's turn, with no push-to-talk by default, though you can enable it. - During tasks, the moderator stays out of the way while the participant works, then checks in. With **Adaptive pause detection** on, it waits in silence rather than filling gaps. - The moderator can't see what's on the participant's screen by default, so it asks them to confirm they can see media before moving on. **Agent vision** (beta) lets it periodically look at the participant's camera and shared screen while media is up. ## The two behavior modes | Mode | What it does | Use it when | | ------------ | ------------------------------------------------------------------------ | ---------------------------------------------------------------- | | **Adaptive** | Rephrases questions naturally, references prior answers, asks follow-ups | Most studies. Conversational. Default. | | **Strict** | Reads each question verbatim from the guide, minimal improvisation | Quant-leaning studies where exact wording matters across people. | ## Things to know - **The moderator doesn't pretend to be human.** If a participant asks, it confirms it's an AI. - **It uses the participant's first name where appropriate**, unless **Name anonymization** is on in [Recording modes](/configure/recording-modes). - **It handles multiple people on a call.** Interviews are usually one-on-one, but the moderator can detect and work with more than one voice in real time, which is useful for studies that need a parent and child, two teammates, etc. Not the standard use case, so test it before relying on it. - **It supports many languages.** Pick one for the whole study, several for participant choice, or all of them. See [Supported languages](/reference/languages). It can also understand individual words or proper nouns interjected in another language (e.g. brand names) even when conducting the rest of the interview in the study's main language. - **It translates your guide live.** Write your guide in any language; the moderator translates as it speaks based on the participant's chosen language. **Custom translation instructions** (in moderator settings) let you steer style, formality, and which terms to keep untranslated. - **No total length cap.** Interviews aren't cut off for running long; the duration the participant sees on the join screen is an estimate. The only time limit is the **Session timeout** (30 sec to 15 min) for inactivity. See [Recording modes](/configure/recording-modes). ## Related Where to change the moderator's settings. Honest limits, for setting expectations. Webcam, screen share, captions, and other call-level settings. Teach the moderator how to say jargon and proper nouns. --- # What the AI moderator can't do Source: https://app.strella.io/docs/product/configure/moderator-limits The AI moderator handles most interviews end-to-end. Here are the places it'll struggle, and what to do about them. ## When this matters If your study depends on something on this list, design around it before publishing, not after the first 10 interviews come back empty. ## Things it can't (or won't) do - **Customize the greeting or the goodbye.** Both are built in and can't be edited or removed. To add framing in your own words, put a **Statement** as the first item (it plays right after the greeting) or the last item (just before the goodbye). - **Skip questions on its own.** Even if a participant covers a later question in an earlier answer, the moderator will still ask the later one unless you've put it inside a conditional section. Use [conditional sections](/build/conditional-sections) to skip pre-empted questions. - **Branch the guide on what a participant _said_ in conversation.** Conditional sections work on **structured answers only** (screener answers, earlier survey answers, or quotas). For free speech, capture the driver as a survey question first, then gate later sections on that answer. Within a question, follow-ups _do_ react to what was said. That's what [Follow-up guidance](/build/interview-guide) is for. - **Auto-advance on a timer.** Items advance when the conversation completes or the participant acts. For activities with a natural finish line, use a goal-based Figma prototype; for pacing control, hand advancement to the participant via **Question flow → Participant-controlled**. - **Script the exact wording of follow-ups.** Follow-up wording is composed live. You control how many ([Follow-up depth](/build/interview-guide)) and what to probe ([Follow-up guidance](/build/interview-guide)). If a question must be asked in specific words, make it its own guide item with **Ask question verbatim**. - **Skip past a survey question or answer it for the participant.** Survey questions wait for the participant. If you want it skippable, make it an open question instead, or gate it with a [conditional section](/build/conditional-sections) so some participants never see it. - **Switch language mid-call.** The interview runs in the language the participant chose at the start. The moderator can recognize individual words in another language (brand names, proper nouns) but won't change the conversation language partway through. - **Hand-pick which participant gets which variant.** Variant assignment is automatic and random. If specific participants need specific content, use [conditional sections](/build/conditional-sections) keyed to screener answers or quotas instead of randomization. - **Interrupt mid-thought reliably.** It waits for natural pauses. Participants who talk in long uninterrupted paragraphs may finish a thought before the moderator can redirect. - **Make judgments your guide hasn't asked for.** It won't decide a participant is "lying" or "biased" and probe on its own. It follows your guide. - **See what's on screen** by default. **Agent vision** (beta) lets the moderator look at the camera and shared screen periodically. Without it, the moderator works from what the participant says, which is why it asks them to confirm they can see media before moving on. - **Read documents the participant uploads mid-call.** Ask them to share their screen with the document open instead. - **Run past your Session timeout for an unresponsive participant.** If nothing happens for the configured timeout (30 sec to 15 min), the call ends after a 30-second warning. Anything the moderator says or asks resets the clock. On a **Task**, where the moderator stays quiet while the participant explores on their own, the participant talking resets it too, so thinking out loud won't end the call. On other elements talking alone doesn't reset it, so an interview that has genuinely stalled still closes out. ## Related What it does well, for context on these limits. The mechanism for "show this only to participants who…" Design choices that work with the moderator's strengths. --- # Customize the participant experience Source: https://app.strella.io/docs/product/configure/participant-experience Participants see a series of pages: the share-link landing, the screener, the call, and a thank-you. Everything they read or click is configurable. ## Steps Click the gear icon in any study page. These two fields appear on the share-link landing page and registration page: - **External study title**: the participant-facing name of your study. It is the same field as **Project overview → Study title**, so edit it in either place. - **Description**: a short summary of what the study is about (max 300 characters). Helps participants decide whether to continue. Upload a PNG, JPG, or similar image (SVG isn't supported; max 4 MB). The logo appears on the landing page, screener, and call join screen. Strella resizes the image to 320 px on its longer side, which stays sharp at the size participants see it. An animated GIF keeps only its first frame. A transparent background works on every page. Opaque logos are fine if they match the page background. Two color pickers under **Button**: - **Button color**: background of every primary action button (Continue, Join, Submit). - **Text color**: the label color on those buttons. The preview button in the settings panel updates live as you change colors. Pick a combination with strong contrast: text on top of background needs to be readable. The interview-type and reward tags on the landing page take a pale tint of your button color. Strella darkens the tag text from the same color until it is readable on the tint, so a light button color gives darker tag text. Under **Font**, click the pencil and choose **Upload font**. Upload one font file: `.woff2`, `.woff`, `.ttf` or `.otf`, under 1 MB. A sample line shows the font. Participants see it on the share-link landing page, the registration page, the thank-you page and every page of the screener. The call and emails keep the Strella font. - **One file covers every weight.** Participants' browsers draw bold titles and labels from that one file, so upload your regular weight. - **Check your font license.** Upload only a font your license lets you use on a website. Strella doesn't check. - **A file the browser can't read** shows the default font instead, without an error. Open **Preview** to confirm the font loads. Click **Preview** at the top of the Appearance section. The landing page opens full-window with your unsaved title, description, logo, button colors and font. The preview always shows English text and all three registration fields (first name, last name, email), even when a share link turns one off. You can type in the form, but it doesn't submit. Click **Exit preview** to return to your settings. Under **Settings → Access & scheduling → Custom closed message**, write what participants see if they click your share link after the study has closed. Useful for "thanks, we're full: sign up for future studies at [link]" wrap-ups. To send participants to a specific URL after the interview, configure redirects per [share link](/recruit/share-links). Available destinations: - **After successful interview**: for qualified, completed participants. - **After screen-out**: for disqualified participants. - **After quota full**: for over-quota participants. - **After dropout**: for participants who quit mid-call. Each share link can have its own set. This helps when you have a "customer" link that sends people back to your dashboard and a "panel" link that sends people back to the panel. The next participant who opens your study sees your branding, and goes to the page you've set when they finish. ## Tips - **Test branding on the actual call page**, not just in the settings preview. Some color combinations look fine in one context and fail in another. - **The "Redirect URLs (Legacy)" settings tab is deprecated.** Configure redirects per share link instead. The legacy fields still work for studies that already use them; in new studies, set redirects on share links. - **For studies you run regularly**, set branding once on a template study and duplicate it. Branding doesn't import via the [import flow](/build/import) because it's study-level, but duplicating a study preserves it. ## Related Title and description are also editable from the Project overview tab. Per-link redirect destinations. The closed-message setting is there. --- # Recording modes Source: https://app.strella.io/docs/product/configure/recording-modes Different studies want different things on the recording. Concept tests usually want webcam-only; usability tests want screen share; sensitive interviews might want neither. ## Steps Click the gear icon in the top-right, then **Interview setup**. **Allowed devices** controls which devices participants can join from: **Desktop and mobile**, **Desktop only**, or **Mobile only**. **Webcam recording**: - **Required to join**: the participant must turn on their camera. Use for emotional research, where facial reactions matter. - **Off**: the participant doesn't share video. Use for sensitive studies or when you only need voice. With webcam on, you can also turn on **Blur participant camera** to record the audio and a blurred video, for privacy without losing the conversation. Toggle **Screen recording** on if the study needs the participant to share their screen (almost always for usability work). Sub-options when screen share is on: - **Allow pausing screen share**: participants can hide sensitive screens mid-call. - **Default screen share option**: what the participant sees first in the screen-share picker (their **entire screen**, a window, or a tab). - **Screen share verification**: verify that participants are sharing the screen you expect, and catch anyone who shares the wrong thing. - **Blur participant screen share**: record a blurred version, for privacy. ![Settings → Interview setup → Recording: Camera recording and Screen recording toggles, with advanced screen-share options: default screen share option, screen share verification, and pausing.](/docs/configure/recording-modes.png) Three independent call controls: - **Default captions on**: show live captions during the call. Most studies leave this on. - **Session timeout**: end the call automatically when the interview goes idle past a threshold. Prevents the AI from waiting forever. On a **Task** the participant's own voice counts as activity, so thinking out loud while exploring a prototype won't cut the call short. - **Push to talk**: make the participant hold a button to speak (instead of voice-activity detection). Useful in noisy environments. - **Name anonymization**: replaces participant names with anonymous IDs throughout the study, and hides participant emails and social links everywhere: in Strella, in exports, in AI assistants connected to Strella, and in the developer API. This protection applies no matter what access a teammate or API key has been granted. - **Voice anonymization**: masks voice in recordings (where available). Useful for studies where the data may be shared more broadly. These settings apply to every new interview. Already-recorded interviews keep their original settings. ## Tips - **Test screen share with a real screen** in [Preview](/build/preview). Some operating systems behave differently, so test before publishing. - **Mobile screen share is supported** through Strella's iOS and Android apps. Participants on mobile install the app to share their device screen. - **Captions are useful even when you don't need them.** Participants with hearing differences benefit, and reviewing the live transcript helps with note-taking. - **Check the defaults before relying on them.** Strella ships with screen recording on, but several toggles on this tab are study-level: open the settings on a new study and confirm they match what you want. ## Related The full matrix of recording configurations. What participants need to join. Mental model for the interview itself. --- # Voice, language & behavior Source: https://app.strella.io/docs/product/configure/voice-language-behavior Open **Settings → AI moderator** to change anything on this page. Changes are held in your draft until you click **Save** at the bottom of Settings. If you close with unsaved edits, Strella asks whether to discard them. ## Steps Under **Moderator language**, choose one or more from the dropdown. Set **Default language** as the fallback when a session can't otherwise pick. - **One language**: every interview happens in that language. The voice picker filters to voices that speak it. - **Multiple languages**: the participant picks at the start of their call. - **All languages** (leave empty): same as picking every supported language. See [Supported languages](/reference/languages) for the full list. What you see depends on languages: - **One English-only study:** browse English voices by name and provider. - **One non-English study:** voices filter to ones that speak that language. - **Multiple languages or all-languages:** pick a gender (Feminine / Masculine / Neutral); Strella picks the right voice per language automatically. Click the play icon next to any voice to preview a sample. Set **Voice speed** with the slider. Participants can still adjust their own playback speed during the call. ![Settings → AI moderator → Voice: a voice sample with a Preview button, the Moderator voice dropdown, and a Voice speed control.](/docs/configure/voice-options.png) Use **Change display name** to override the default "Strella." The moderator introduces itself with that name, and the call UI refers to the moderator by it. **Question behavior**: - **Adaptive** (default): conversational; references prior answers; rephrases questions naturally. - **Strict**: reads each question verbatim; minimal improvisation. For one-off exact wording on a specific question, use the per-question **Ask question verbatim** toggle in your guide. It does the same thing for a single question. **Question flow**: - **Automatic** (default): the moderator decides when each question has been answered and moves on. - **Participant-controlled**: the moderator stays silent and waits for the participant to click next. Won't interrupt. Automatic is more conversational. Participant-controlled is useful for studies where you need long pauses (e.g. think-aloud usability work). **Reaction length** controls how _much_ the moderator says between answers: **Low** (brief acknowledgements), **Medium** (default, natural conversational warmth), or **High** (longer reflections). Setting it to **Low** is one way to make the moderator less effusive between answers; to change _which_ acknowledgements it uses, or stop it affirming answers altogether, use **Affirmation tone** (below). For Task steps, **Task completion handling** decides how the moderator knows the participant is done: - **Default**: the moderator judges based on what the participant says and confirms before moving on. - **End task button**: the participant clicks a button. - **Explicit confirmation**: the moderator waits for a verbal "done." A goal-based Figma task is the exception: reaching the goal screen ends the task automatically. **Adaptive pause detection** lets the moderator deliberately wait in silence, rather than filling every gap, when a participant is thinking or working through something. It resumes when the participant re-engages. Useful for tasks and think-aloud research. **Agent vision** lets the moderator periodically look at the participant's camera and shared screen while media is up, so it can react to what they're doing visually instead of just what they say. Blurred sources stay invisible to it. This is a beta feature. If you don't see the toggle, contact your account team to enable it on your organization. Three free-text fields shape the moderator's voice without touching the guide: - **Affirmation tone**: describe how the moderator acknowledges answers between questions ("neutral and brief, never enthusiastic"). Replaces the default acknowledgement style. - **Custom behavioral instructions**: guidance the moderator follows for the whole interview ("never suggest answers", "refer to the product as Acme One"). Won't override the built-in greeting or closing. - **Custom translation instructions**: steer how the moderator translates your guide ("use formal Dutch", "keep brand names untranslated"). All three are optional. Defaults work for most studies. The next interview will use these settings. Active interviews already in progress finish on their original settings. ## Tips - **You can't replace the greeting or the goodbye.** Both are built into the moderator. To add framing in your own words, put a **Statement** as the first or last item in your guide. - **One voice and language config per project**, not per question. If you need different audiences in different languages, allow multiple languages and let participants pick at the start of the call. - **Preview the voice before publishing.** Sample text won't catch jargon problems. Take a full preview interview to feel the actual pacing. - **Custom behavioral instructions can fight the moderator's defaults.** Keep them additive ("never suggest answers") rather than contradictory ("always tell them what others said"). ## Related The mental model behind these settings. The full voice list and which providers each comes from. Languages you can pick from. Help the moderator pronounce names and jargon correctly. --- # How AI helps you in Strella Source: https://app.strella.io/docs/product/get-started/ai-assistants Strella uses AI across the whole study lifecycle. Some are a chat you talk to; others run automatically in the background. Here's every major place AI helps, stage by stage. ## Setting up your study - **[AI-guided setup chat](/build/project-setup-chat)**: describe what you want to learn (or upload a brief) and it drafts a screener and a question guide. You keep editing it while the draft takes shape. - **[Build assistant](/build/build-assistant)**: the right-hand panel in Build. It reads your current draft and helps you edit it: rewrite a question, add a follow-up, suggest a section. ## Running the interview - **[AI moderator](/configure/how-the-ai-moderator-works)**: conducts every interview by voice. It greets the participant, walks your guide top to bottom, asks dynamic follow-ups, and closes out. It runs in many languages and translates your guide live, and (in beta) can glance at the participant's screen. - **Automatic transcription**: every recording is [transcribed](/analyze/review-interview) with speaker labels, synced to the video, and searchable. Regenerate it if speakers are mislabeled. ## Analyzing results - **[Per-interview analysis](/analyze/review-interview)**: each interview gets an AI-written summary, auto-generated notes, and quality-check badges that flag fraud and quality risks. - **[Themes & sentiment](/analyze/browse-analysis)**: across every completed interview, AI groups responses into themes by objective or by question, scores sentiment, codes open-ended answers into countable categories, and pulls representative quotes. - **[Results AI chat](/analyze/ai-chat)**: ask the analyst anything about this study's interviews. It cites its sources and can spin up a report or a highlight reel. See [AI chat best practices](/analyze/ai-chat-best-practices) for how to prompt it well. - **[Multi-project explorer](/analyze/ai-chat)**: the same chat, scoped across several studies in your organization, for questions that span more than one. ## Sharing findings - **[AI-generated reports](/share/reports)**: drafts a full writeup (executive summary, participant profiles, theme cards with quotes) and even a cover image. Edit any of it before you share. - **[Highlight reels](/share/highlight-reels)**: every theme has an auto-compiled reel of its source moments; the AI chat can build one on demand too. ## Bring your own AI - **[MCP](/mcp)**: connect an external AI assistant (Claude, ChatGPT, and others) to your Strella data so you can query your studies from the tools you already use. See [API, MCP & assistants](/account/api-mcp-assistants). ## At a glance | Stage | AI feature | What it does | | --------- | --------------------------------------- | ------------------------------------------------- | | Build | Setup chat, Build assistant | Draft and edit your screener + question guide | | Interview | AI moderator, transcription | Run each interview by voice, then transcribe it | | Analyze | Per-interview analysis, themes, AI chat | Summaries, quality checks, themes, sentiment, Q&A | | Share | Reports, highlight reels | Turn findings into a writeup or a reel | | Anywhere | MCP | Query your data from an external AI assistant | ## Things to know - **AI-generated insights cite their sources.** In the [analysis](/analyze/browse-analysis), the [Results AI chat](/analyze/ai-chat), and the multi-project explorer, click a quote to jump to that moment in the source interview. (The setup chat and build assistant don't link to interviews, because there aren't any yet.) - **Analysis only considers completed interviews.** The themes, the AI chat, and the explorer all ignore drafts and abandoned interviews. - **To change how the live moderator behaves, edit [Settings → AI moderator](/configure/voice-language-behavior).** The build assistant edits your guide; it doesn't run the interview. - **The moderator doesn't pretend to be human.** If a participant asks, it confirms it's an AI. ## Related The AI that conducts every interview. The in-Build AI for drafting and editing. Project chat + the multi-project explorer. AI themes by objective and by question. --- # Glossary Source: https://app.strella.io/docs/product/get-started/glossary A quick reference for the words Strella uses. Each term has a longer explanation linked. | Term | What it means | | ---------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Project** | One study. Has a screener, a question guide, a set of participants, and a results section. | | **Screener** | A short questionnaire that decides whether a participant qualifies for the interview. See [Build your screener](/build/screener). | | **Question guide** | The steps the AI moderator runs through with each participant: questions, tasks, statements, and media. See [Build your question guide](/build/interview-guide). | | **Step** | A single item in the screener or question guide. | | **Section** | A group of steps. A section can be conditional, so Strella shows it to only some participants, or randomized. | | **Task** | A step where the participant does something, such as visit a website, walk through a Figma prototype, or open a file. | | **Research objective** | A thing you want to learn. The AI organizes analysis around your objectives. See [Write research objectives](/build/research-objectives). | | **Interview** | One completed recorded session with a participant. | | **AI moderator** | The voice that conducts the interview live. See [How the AI moderator works](/configure/how-the-ai-moderator-works). | | **Participant** | Someone who takes the screener and (if they qualify) the interview. | | **Panel** | A third-party pool of recruitable participants. Strella integrates with Respondent and User Interviews. | | **Order** | A request you place with a panel for a batch of participants. | | **Share link** | A URL to invite your own participants. They click it, take the screener, and (if they qualify) are invited to take the call. | | **Quota** | A rule that caps how many participants of a given segment can complete the study. See [Set quotas](/recruit/quotas). | | **Funnel** | The stages a participant moves through, from clicking the link to finishing the interview. | | **Theme** | A pattern across multiple interviews that the AI analyst finds, such as "users hesitate at the pricing page." | | **Sentiment** | The AI's read of how positive or negative a participant was on a topic. Three bands: Positive, Neutral, Negative. | | **Highlight reel** | A shareable collection of highlights, from one or more interviews. See [Create & share a highlight reel](/share/highlight-reels). | | **Report** | A polished writeup of a study's findings, generated by AI and editable. See [Build & share a report](/share/reports). | | **Workspace** | A way to group studies inside an organization. See [Teammates & workspaces](/account/teammates-and-workspaces). | | **Organization** | Your company's top-level Strella account. Holds workspaces, studies, billing, teammates. | ## Related The lifecycle these terms fit into. Every kind of step you can add to a question guide. --- # Run your first study Source: https://app.strella.io/docs/product/get-started/quickstart This walks you from a blank study to one finished interview you can analyze. ## Before you start You'll need: - A Strella account - A research question or topic - 10 minutes to author the study, plus a few hours to a few days for participants to take the interview ## Steps From your **Projects** list, click **New project** and give it a name. ![Creating a study from the Projects list: clicking New project in the workspace.](/docs/get-started/create-new-project.gif) In the AI-guided setup, Strella opens a chat. Tell the chat what you want to learn, who the participants are, and what you want to study in the interview. It drafts a screener and question guide for you. Prefer to build it yourself? Click **Skip setup** in the chat to open Build with empty drafts and create the study by hand. The AI-guided setup drafts your research objectives: the 2 to 5 things you want to learn. The AI analyst uses these to organize themes later, so make them specific: "Understand which pricing tier B2B buyers consider first" beats "learn about pricing." Refine them anytime in **Settings → Research objectives**. In **Build**, the left panel shows your steps. Click **+ Add element** to add a screener question (above the divider) or an interview step (below it). For the screener, keep it short: 4 to 8 questions usually does it. For the interview, drop in open questions and tasks the AI will ask in order. Open **Settings → AI moderator** to customize Strella's voice. Defaults work for most studies. Click **Publish** in the top right of Build. Strella runs final checks and opens the study to participants. Publish before you recruit: the screener and guide need to be live before participants can take them. Go to **Recruit**. Hire from one of Strella's integrated panels, or bring your own participants with a share link. Once a participant finishes, **Results → Interviews** shows the recording. Open it, watch the call, read the transcript, and add notes. After a handful of interviews, **Results → Analysis** has themes grouped by objective and by question. Click any quote to jump to the moment in the call. You've shipped a real study. The next interviews come in without any work from you, and analysis updates as they complete. ## Tips - Use the **Preview** button in Build before publishing: you can take the interview yourself in 5 minutes and feel exactly what participants will feel. - Read your questions aloud before publishing. If one feels too long or too leading when you say it out loud, rewrite it, since reading them out is the fastest way to catch a question that won't land. ## Related The 3-panel layout that holds your screener, question guide, and the AI build assistant. What it can and can't do, and how it picks each follow-up question. --- # Welcome to Strella Source: https://app.strella.io/docs/product ## How a study works Write your research objectives. Author a screener and a question guide in one screen. An AI build assistant helps you iterate. Hire participants from a panel like Respondent or User Interviews, or bring your own with a custom share link. Set quotas to balance your sample. Publish. The AI moderator interviews each participant by voice, follows up, and guides them through any tasks or prototypes. Watch any interview, read the transcript, browse themes by objective or question, and chat with an AI analyst over the full set of responses. Generate a report. Build a highlight reel from the strongest moments. Share both as a public link. ## Where to start A 15-minute walkthrough end-to-end. Every place AI shows up in a study: setup, interviews, analysis, sharing. Screener, question guide, build assistant, all in one screen. The vocabulary you'll see across the app. ## For developers & AI assistants Strella has a public REST API and AI-readable documentation you can fetch directly, which is useful for building integrations or pointing an AI assistant at the docs: - [API reference](https://app.strella.io/docs/api), the interactive REST API reference with every endpoint, schema, and the API-key auth. - [llms-full.txt](https://app.strella.io/llms-full.txt): the complete product documentation as a single plain-text file, for ingestion by an AI assistant. - [llms.txt](https://app.strella.io/llms.txt): a short AI-readable index that links the full docs and the API contract. - [OpenAPI specification](https://app.strella.io/api/public/v1/openapi.json): the machine-readable API contract (OpenAPI 3.1). ## Need a hand Open the AI assistant in the corner of any docs page. It reads these pages and cites its sources. If you spot something out of date or missing, let us know. We use those gaps to prioritize what to write next. --- # Connect an AI client Source: https://app.strella.io/docs/product/mcp/connect Connecting takes about two to three minutes. Strella is listed in the Claude and ChatGPT connector directories, so for those two there is nothing to configure: open the listing, click connect, and sign in with your Strella account through a secure OAuth redirect. For every other client, add Strella as a custom connector using the same server URL: ``` https://app.strella.io/api/mcp ``` **Part of an organization?** Some organizations restrict which connectors people can add. If you can't find Strella in your connector list, or the connect button doesn't work, ask your workspace administrator to enable it. ## Claude (web, desktop & Cowork) Claude.ai, the Claude desktop app, and Claude in Cowork all share the same Connectors flow. Strella is a directory connector, so it's already in your list: there's nothing to add. Go to **Settings → Connectors** in Claude, or open Strella's listing directly at [claude.ai/directory/connectors/strella](https://claude.ai/directory/connectors/strella). Find **Strella** under **Not connected** and click **Connect**. ![The Claude Connectors list on the Not connected tab, showing Strella with a Community badge and a Connect button.](/help/claude-dir-step-1.png) Claude redirects you to sign in with your Strella account. Click **Allow** to complete setup. Strella is available in any new chat. Manage connector tools from the Connectors panel. After setup, try: `@Strella, start the connector tutorial for me` If your organization blocks directory connectors, add Strella by URL: 1. In **Settings → Connectors**, click the **+** button, then select **Add custom connector**. 2. Name it **Strella** and set the URL to `https://app.strella.io/api/mcp`. Leave advanced settings as default. 3. Find **Strella** under **Not connected**, click it, click **Connect**, and sign in. Authenticate with OAuth 2.0. Strella does not support OIDC, so leave your tool's OIDC toggle off. Connectors are available on all Claude plans. Update your app to the latest version. If you're part of an organization and don't see this option, ask your Claude workspace administrator to enable it. Disconnect the connector and add it again. If the problem continues, sign out of Claude and sign back in. You can also contact Strella support. Mention Strella in your prompt: "Using Strella, show me my projects." Claude may not pick the connector on its own the first time. ## ChatGPT Requires ChatGPT Pro, Plus, Business, or Education. Strella is listed in ChatGPT's plugin directory, so you create no app and paste no URL. Go to [Strella's plugin page](https://chatgpt.com/plugins/plugin_asdk_app_69bd87a9820081919a9bdbbac64c05ba?q=strella). You can also reach it from **Settings → Plugins → Browse plugins** and searching for Strella. ![Strella's plugin page in ChatGPT, showing the "AI native customer research" tagline, example prompts, and a Try in chat button.](/help/chatgpt-dir-step-1.png) Add Strella, then click **Try in chat**. ChatGPT redirects you to sign in with your Strella account. Click **Allow** to complete setup. Manage or disconnect Strella from **Settings → Plugins** at any time. After setup, try: `Show me my research projects in Strella` If you need to add Strella by URL rather than from the directory: 1. Go to **Settings → Plugins → Developer mode**, then click **Create app**. 2. Name it **Strella**, set the MCP Server URL to `https://app.strella.io/api/mcp`, and leave Authentication set to **OAuth**. Paste the URL before opening Advanced settings so the fields autofill. 3. Open **Advanced settings**, scroll to **OpenID support**, and uncheck **OIDC enabled**. A manually added app needs this off for Strella's sign-in to work. 4. Check the **"I understand and want to continue"** box, click **Create**, then sign in. Plugins require ChatGPT Pro, Plus, Business, or Education. Free plans do not support plugins. Update ChatGPT to the latest version. Disconnect Strella from Settings → Plugins and add it again from the plugin listing. If you added Strella manually, uncheck OIDC enabled. Start a new chat and mention Strella: "Using Strella, show me my projects." You can also click Refresh in Settings → Plugins → Strella to update the tool list. ## Figma Make Requires a Figma Dev or Full seat. In **Figma Make**, click the **+** button (Add context), then select **Connectors → Manage**. If you don't see the Connectors option, you may need a Figma Dev or Full seat. Alternatively, ask an organization admin to enable this connector for your team. Go to **Created by you** and click **+ Create**. - Name: **Strella** - Tagline and Description are optional - MCP server URL: `https://app.strella.io/api/mcp` - Skip Advanced settings Click **Next** after the first page, then **Create** on the second. Find **Strella** in the connector list and click **Connect**. Figma redirects you to sign in with your Strella account. Click **Allow** to complete setup. Manage connector settings here, or start a Figma Make chat with **@Strella**. After setup, try: `@Strella show me my research projects` Connectors require a Figma Dev or Full seat. Check your seat type in Figma settings. Refresh the page. If Strella still doesn't appear, go back to Connectors → Manage and check that the connector exists. ## Claude Code Run the following command in your terminal to add Strella as a connector: ```bash claude mcp add --transport http --scope user strella https://app.strella.io/api/mcp ``` Then run `claude /mcp` to authenticate with your Strella account. After setup, try: `Using Strella, show me my projects` Install Claude Code first. Visit claude.ai/download for installation instructions. ## Cursor Open (or create) the file `~/.cursor/mcp.json` and add the Strella configuration: ```json { "mcpServers": { "strella": { "url": "https://app.strella.io/api/mcp" } } } ``` Fully restart Cursor. A reload is not enough. Cursor then prompts you to sign in with your Strella account. After setup, try: `Using Strella, list my research projects` Save the mcp.json file and fully restart Cursor. A reload is not enough. Check Cursor's MCP settings for the Strella server. ## Other clients Strella works with any AI tool that supports connectors (MCP servers). Add a custom connector with the server URL `https://app.strella.io/api/mcp`, using OAuth 2.0 (no OIDC). Follow the official setup docs for your tool: **Productivity, design & prototyping** - [Microsoft Copilot Studio](https://learn.microsoft.com/en-us/microsoft-copilot-studio/mcp-add-existing-server-to-agent) - [Notion (Custom Agents)](https://www.notion.com/help/mcp-connections-for-custom-agents) - [Lovable](https://docs.lovable.dev/integrations/mcp-servers) - [Bolt.new](https://support.bolt.new/building/using-bolt/connect-mcp) - [v0](https://v0.app/docs/MCP) **Code editors & dev tools** - [Cursor](https://docs.cursor.com/context/model-context-protocol) - [Windsurf](https://docs.windsurf.com/windsurf/cascade/mcp) - [VS Code + GitHub Copilot](https://code.visualstudio.com/docs/copilot/customization/mcp-servers) - [Claude Code](https://docs.anthropic.com/en/docs/claude-code/mcp) - [Replit](https://docs.replit.com/replitai/mcp/overview) ## Authentication The Strella MCP server uses OAuth 2.0 via Clerk. When you connect, you're redirected to sign in with your Strella account, and the connection is scoped to your user permissions: you can only access organizations and projects you already have access to in Strella. - OAuth 2.0 with PKCE, compliant with the OAuth 2.1 draft and the MCP specification - Automatic token refresh - Scoped to your Strella role, so Viewers get read-only access Your Strella password is never shared with the AI provider: the sign-in uses a secure redirect, so only Strella ever sees your credentials. ## Technical details - **Transport**: Streamable HTTP (JSON-RPC 2.0) - **Auth**: OAuth 2.0, compliant with the OAuth 2.1 draft and MCP specification - **Rate limit**: the service is rate limited to prevent abuse. Under normal use you won't hit it; if you do, your AI tool waits and retries automatically. ## Want to learn more about MCP? The Model Context Protocol is an open standard for connecting AI assistants to external tools and data sources. Read the [MCP documentation](https://modelcontextprotocol.io/docs/getting-started/intro) to learn more. ## Related Every tool the connector exposes, with parameters and behavior. Refreshing tools, missing actions, and connection issues. --- # FAQ & troubleshooting Source: https://app.strella.io/docs/product/mcp/faq ## Getting started Strella works with Claude, ChatGPT, Figma Make, Claude Code, Cursor, and any other AI tool that supports connectors. No. Strella's connector runs in the cloud. You just connect it from your AI tool's settings, with no downloads, terminal commands, or config files (unless you're using a code editor like Cursor). About 2 to 3 minutes. You add the Strella connector in your AI tool, sign in with your Strella account, and you're done. The Strella connector is included with your Strella subscription at no additional cost. You do need a compatible AI tool subscription (e.g., ChatGPT Plus) that supports connectors. ## Using the connector Yes. You can connect Strella to Claude, ChatGPT, and other tools simultaneously. Each connection is independent. The Claude mobile app supports connectors, so you can use Strella on your phone through Claude. Other AI tools may add mobile connector support over time. Check your app's settings. Try refreshing the connector in your AI tool's settings. If that doesn't work, disconnect and reconnect Strella. You won't lose any data: it just re-establishes the link. That's set in your AI tool, not in Strella. AI tools typically ask for your approval before running a tool that changes data, and most let you set each tool to always allow, ask every time, or never allow. For read-only access, set the tools that change data to never allow. See [Security & privacy](/docs/product/mcp/security#does-the-ai-ask-before-changing-my-data). You probably have the Viewer role in Strella. Viewers can browse and analyze research through the AI, but can't create or edit anything. Ask an admin in your organization to change your role. ## Security & privacy When you ask about your data, the AI fetches it from Strella and includes it in the conversation. The AI provider (Anthropic, OpenAI, etc.) processes that data within the conversation: nothing is bulk-uploaded or stored separately. Most providers offer a "don't train on my data" setting in their privacy options. No. Your conversations with the AI are private to you, just like any other conversation in Claude or ChatGPT. The AI can only access data you already have permission to see in Strella. The connector uses your existing account permissions: it doesn't grant any additional access. Go to your AI tool's connector settings and remove Strella. Once removed, the AI can no longer fetch data from your account. You can reconnect at a later time. That depends on your AI provider and plan. Most providers offer enterprise or team plans where conversation data is not used for training. Claude Team/Enterprise and ChatGPT Business/Enterprise do not use your data for training by default. Check your AI provider's data policy for your specific plan. When you connect Strella, you sign in with your existing Strella account. Your password is never shared with the AI provider: the sign-in uses a secure redirect so only Strella ever sees your credentials. For a fuller walkthrough of data access and privacy, see [Security & privacy](/docs/product/mcp/security). ## Technical MCP (Model Context Protocol) is an open standard that lets AI assistants connect to external services like Strella. It defines how your AI tool discovers what Strella can do and how it requests data or takes actions on your behalf. Think of it as a shared language between your AI assistant and Strella. Without it, the AI would have no way to access your research data. MCP is supported by Claude, ChatGPT, Cursor, Figma Make, and a growing number of other AI tools. OAuth 2.0 is the industry standard for secure third-party sign-in, the same technology behind "Sign in with Google" buttons you see across the web. When you connect Strella to your AI tool, OAuth lets you sign in to your Strella account through a secure redirect. Your password is never shared with the AI provider. Strella also uses PKCE (Proof Key for Code Exchange), an additional security layer that protects the sign-in flow from being intercepted, even on public networks. OIDC (OpenID Connect) is an identity layer built on top of OAuth 2.0. Some AI tools offer an OIDC option in their connector settings. Strella does not use OIDC. We use standard OAuth 2.0 with PKCE instead. If your AI tool has an OIDC toggle (like ChatGPT's "OIDC Enabled" checkbox), make sure it is turned off when connecting Strella. Tools are individual actions the AI can perform through Strella, such as listing your projects, analyzing transcripts, or creating a new study. Your AI assistant already has built-in tools for things like web search and image generation; Strella's connector adds research-specific tools on top. You never need to call tools by name. Just describe what you want in plain language and the AI picks the right tool automatically. Yes. To protect service quality, the Strella connector limits how many requests your AI tool can make per minute. Normal use stays under these limits. If you reach one, the AI waits and retries automatically. The limit does not affect your access to the Strella web app. Yes. TLS (HTTPS) encrypts all communication between your AI tool and Strella. Nothing travels in plain text. ## Troubleshooting Your AI tool saves the list of Strella's tools when it first connects, and keeps it until you reconnect or start a new chat. The tools you already have keep working after a Strella update. To see new ones, use the steps below. Your AI tool remembers what Strella can do when it first connects, and keeps that list until you reconnect or start a new chat. In Claude.ai, click the … menu on the Strella connector and choose "Refresh tools list". In ChatGPT, open Settings → Plugins → Strella and click Refresh. In Claude Code, type `/mcp` and reconnect Strella. As a fallback, disconnect and reconnect, or start a new chat. AI tools cache connector definitions when they first connect, and a Strella update can leave that cache out of date. Try these steps in order: 1. **Refresh tools**: in Claude.ai, Connectors → Strella → … menu → "Refresh tools list"; in ChatGPT, Settings → Plugins → Strella → Refresh; in Claude Code, type `/mcp` and reconnect Strella. 2. **Reconnect**: if refresh isn't available or didn't recover the chat, disconnect and re-add the Strella connector. You may need to start a new chat afterward. 3. **Re-authenticate**: if errors persist, sign out of your AI tool and sign back in to refresh your session. Check that you're connected to the right organization. Ask "Which organization am I connected to?" and switch if needed. Also confirm you have at least one project with data in Strella. Check that you're using a supported AI tool plan (e.g., ChatGPT Pro/Plus, not Free). For authentication issues, contact us. The connector fetches fresh data every time you ask. If something looks outdated, ask again or refresh the connector in your AI tool's settings. ## Need help? If your AI assistant keeps hitting errors with Strella, or you have an idea for it, ask your assistant to send feedback to Strella. It writes it up, shows it to you, and sends it to our team once you agree. See [`strella_send_feedback`](/docs/product/mcp/tools#strella_send_feedback). Still stuck? Email us at [customersupport@strella.io](mailto:customersupport@strella.io). ## Related Per-client setup and configs. What data the AI can access and how sign-in works. --- # MCP server Source: https://app.strella.io/docs/product/mcp The Strella MCP server connects your research to the AI tools you already use: Claude, ChatGPT, Figma Make, Cursor, and any other client that supports the [Model Context Protocol](https://modelcontextprotocol.io). Once connected, your AI assistant can see your projects, interview transcripts, participant data, and session recordings. You ask questions in plain language, and it pulls the answers directly from Strella. Think of it like giving your AI assistant a window into your research. Instead of copying data between tabs or summarizing findings by hand, you can say "What did participants say about pricing?" and get a synthesized answer with quotes from your actual transcripts. It works both ways too: you can create projects, write interview questions, and set up screeners entirely through conversation. Your AI handles the structure; you focus on the research thinking. No installation required. The connector runs in the cloud: there's nothing to download or configure on your machine (unless you're using a code editor like Cursor). Connect it from your AI tool, sign in with your Strella account, and you're done in about three minutes. ## What you can do Everything below works through natural conversation. Start by exploring your data: nothing changes until you explicitly ask to create or edit something. **Explore and browse** - List your research projects with name, description, dates, and participant counts. - Look up participant demographics, interview summaries, and screener responses. - View interview scripts and questionnaire structure. - Watch session recordings and highlights in an embedded player. **Analyze and synthesize** - Ask research questions and get AI-powered summaries with quoted evidence. - Search across transcripts for specific topics, phrases, or patterns. - Compare what different participants said about the same topic. **Create and manage** - Create new research projects from a brief or description. - Add and edit interview questions, screeners, and project settings. - Draft recruitment orders against the Respondent and User Interviews panels. Creating and editing is available to everyone in your organization except Viewers, and your AI assistant typically asks for your approval before it changes anything. See [Security & privacy](/docs/product/mcp/security#who-can-make-changes-through-the-connector). ## How it works Under the hood, the connector gives your AI assistant a set of "tools", the same way it already has tools for searching the web or generating images. Strella's tools let it access, analyze, and update your research data. You never need to call tools by name; just describe what you want in plain language and the AI picks the right one. The server speaks the Model Context Protocol over Streamable HTTP, and authenticates with OAuth 2.0 through your Strella account. The connection is scoped to your existing permissions: the AI can only reach organizations and projects you can already access in Strella. ## Get connected Step-by-step setup for Claude, ChatGPT, Figma Make, Claude Code, and Cursor. Every tool the connector exposes, with parameters and behavior. Step-by-step workflows the AI loads for complex tasks like building guides. Copy-paste prompts for exploring, analyzing, and managing research. What data the AI can access, how sign-in works, and how to disconnect. Common questions about setup, privacy, and troubleshooting. ## Need help? Have questions or need a hand setting up the connector? Email us at [customersupport@strella.io](mailto:customersupport@strella.io). --- # Security & privacy Source: https://app.strella.io/docs/product/mcp/security The connector uses your existing Strella account and permissions. It doesn't grant the AI any access you don't already have, and Strella uploads nothing to the AI provider until you ask a question that needs it. ## What data can the AI access? The same projects, participants, and transcripts you can reach in Strella, with one limit on personal information. An admin can control how much participant personal information (names, emails, profile photos, demographics, recordings, and responses) the AI reads on your behalf, field by field. That control applies to AI assistants and the API, so the AI may read less participant detail than you do when you sign in to the dashboard directly. See [Participant data access](/account/participant-data-access). ## Is my research data stored by the AI provider? When you ask about your data, the AI fetches it from Strella and includes it in your conversation. The AI provider (Anthropic, OpenAI, etc.) processes it within that conversation, subject to their own data policies. Strella only sends data to the AI provider when you request it. Nothing is uploaded proactively or in bulk. Most AI providers also offer privacy modes or enterprise plans where conversation data is not used for model training. ## Who can see my conversations? Your conversations are private to you. Other people in your organization cannot see what you asked, what data was returned, or what insights the AI generated. This works the same way regular conversations do in Claude or ChatGPT. ## How does sign-in work? When you connect Strella, you sign in with your existing Strella account. This creates a secure link between your AI tool and Strella. Your Strella password is never shared with the AI provider: the sign-in uses a secure redirect (OAuth 2.0 with PKCE) so only Strella ever sees your credentials. ## Who can make changes through the connector? Everyone in your organization except Viewers. Viewers can browse, search, and analyze research through the AI, but Strella rejects any request from them that creates or edits something, such as a project, quota, share link, recruitment draft, or image upload. The connector follows your Strella role, so it never lets the AI do more than you could do yourself. An admin changes a teammate's role under **Settings → Members**. ## What can an image upload command do? When your AI uploads images into a question guide, Strella gives it single-use upload commands. Each command stores one image in your organization, and nothing else: it can't read your research or change a guide. A command stops working once it's used, or after 15 minutes. Until then it works like a password. Your AI shows you a command only if you ask for it, and your AI tool may also show it in the tool's results, so don't share that part of a conversation. ## Does the AI ask before changing my data? Your AI tool controls whether to ask, not Strella. Strella labels every tool as either read-only or one that changes data, and your AI tool uses that label to choose when to ask. AI tools typically ask for your approval before running a tool that changes data. Most also let you choose, tool by tool: - **Always allow**: the tool runs without asking. - **Ask every time**: you approve each use. - **Never allow**: the AI can't use the tool. For read-only access, set each tool that changes data to never allow. You'll find these options in your AI tool's settings for the Strella connector. ## How do I disconnect? Go to your AI tool's connector settings and remove Strella. Once removed, the AI can no longer fetch data from your Strella account. You can reconnect at a later time by adding Strella again and signing in. ## What about my organization's data policies? If your organization has specific data handling requirements, talk to your org admin before connecting. For questions about Strella's security practices, contact us at [customersupport@strella.io](mailto:customersupport@strella.io). ## Related More on privacy, encryption, and data training policies. How OAuth sign-in works for each client. --- # Tools & skills reference Source: https://app.strella.io/docs/product/mcp/tools The connector exposes **27 tools** across five categories. You never call tools by name: describe what you want in plain language and your AI assistant picks the right one. All tools authenticate via OAuth through your Strella account, scoped to the permissions you already have. Data tools support **field selection** to minimize token usage. In the tables below, defaults are marked, and fields marked _detail_ require an extra API call. The `id` field is always included. ## Tool categories at a glance | Category | Tools | | --- | --- | | Getting Started | `strella_get_guide` | | Research Data | `strella_search_projects`, `strella_project_detail`, `strella_list_participants`, `strella_participant_detail`, `strella_get_questionnaire`, `strella_list_workspaces`, `strella_list_recruitment_orders`, `strella_list_share_links`, `strella_list_custom_fields`, `strella_view_recording`, `strella_list_highlight_reels` | | Analysis | `strella_search_transcripts`, `strella_analyze_transcripts`, `strella_analyze_question_guide` | | Project Management | `strella_create_project`, `strella_edit_project`, `strella_prepare_image_uploads`, `strella_list_quotas`, `strella_edit_quotas`, `strella_edit_quota_groups`, `strella_write_draft_respondent_order`, `strella_write_draft_user_interviews_order`, `strella_search_recruitment_options`, `strella_edit_share_link`, `strella_edit_custom_field`, `strella_set_participant_custom_fields`, `strella_save_highlight_reel` | | Utilities | `strella_switch_organization`, `strella_skills`, `strella_search_help_docs`, `strella_send_feedback` | ## Getting Started ### `strella_get_guide` Initializes a Strella session. Returns a `context_id` plus a platform guide with step-by-step workflows. Every other tool needs the `context_id`. **Good to know** - **Called automatically**: this tool runs automatically when the Strella connector is first invoked in a conversation. You don't need to call it manually; your AI assistant handles it behind the scenes. - **Required first step**: returns a `context_id` that every other tool needs. Called once at the start of each conversation. - **Organization list**: shows which organizations you belong to. Most users have one; multi-org users pick which to work with. ## Research Data ### `strella_search_projects` Searches research projects by keyword, with field selection. Filters by workspace and date range. **Fields** (`id` is always included. **Default** fields come back unless you narrow them. _Detail_ fields require an extra call) | Field | Notes | | --- | --- | | `project_name` | default | | `participants_count` | default | | `description` | default | | `participant_facing_name` | | | `slug` | | | `workspace_id` | | | `created_at` | | | `updated_at` | | | `unqualified_count` | | | `qualified_count` | | | `invited_count` | | | `in_review_count` | | | `accepted_count` | | | `rejected_count` | | | `survey_responses_count` | | | `research_objectives` | detail | | `product_description` | detail | | `target_audience` | detail | | `participant_target_count` | detail | | `share_url` | detail | **Filters** | Filter | Description | | --- | --- | | `query` | Search keywords (fuzzy), natural language (deep), or a full Strella URL (url mode) | | `mode` | `fuzzy` (default), `deep` (semantic), `list` (browse all), `url` (exact URL lookup) | | `workspace_id` | Scope to a workspace | | `min_accepted_count` | Minimum completed participants | | `created_after` / `created_before` | Date range (ISO 8601) | ### `strella_project_detail` Gets a project summary. The tool always returns the base project settings. Use the `include` parameter to add data groups. **Optional includes** | Group | Description | | --- | --- | | `questionnaire` | Interview script the AI moderator follows | | `screener` | Qualifying quiz before the interview | | `interview_samples` | Up to 10 sampled interview summaries | | `project_context` | Secure context field used by the AI moderator | ### `strella_list_participants` Lists a project's participants, with field selection. Filters by status, name, and age range. **Fields** (`id` always included) | Field | Notes | | --- | --- | | `name` | default | | `status` | default | | `email` | | | `total_duration_seconds` | | | `interview_count` | | | `survey_response_count` | | | `updated_at` | | | `job_title` | detail | | `gender` | detail | | `age` | detail | | `level_of_education` | detail | | `race_ethnicity` | detail | | `household_income_in_usd` | detail | | `seniority` | detail | | `location` | detail | | `company_size` | detail | | `screener` | detail | **Filters** | Filter | Description | | --- | --- | | `status` | Comma-separated lifecycle states (default: `accepted,in_review`) | | `query` | Participant name (name mode) or Strella URL (url mode) | | `mode` | `list` (default), `name` (search by name), `url` (lookup by Strella URL) | | `age_min` / `age_max` | Age range (inclusive) | ### `strella_participant_detail` Gets detailed data for a single participant. Defaults to `demographics` + `interview_summaries`. **Field groups** | Group | Description | | --- | --- | | `demographics` | Name, email, status, all profile fields, and the participant's custom field values (default) | | `interview_summaries` | AI summaries, duration, session overview (default) | | `screener` | Screener quiz responses | | `sessions` | Session details with survey responses | | `task_results` | Per-Task Pass/Fail verdict for each task the participant performed, with the on-screen reason behind each verdict | ### `strella_get_questionnaire` Gets the questionnaire (interview script) for a project: the tree of questions, tasks, and sections the AI moderator follows. **Parameters** | Parameter | Description | | --- | --- | | `project_id` | The project whose interview script you want to view | **Good to know** - **Full vs. summary**: returns the complete, untruncated interview script. For a quick summary, use `strella_project_detail` with the `questionnaire` include instead. - **Question numbers**: each question is numbered in the order it appears in the script (`Q1`, `Q2`, and `T` for tasks; sections aren't numbered). Transcript search uses the same numbers when it labels lines. ### `strella_list_workspaces` Lists all workspaces in the active organization. **Good to know** - **No pagination**: returns all workspaces in a single call. - **Use case**: workspaces group projects by team or department, useful for scoping project searches. ### `strella_list_recruitment_orders` Lists recruitment orders for a project across both vendor panels. **Parameters** | Parameter | Description | | --- | --- | | `project_id` | The project to list recruitment orders for | | `panel` | Filter by panel: `respondent` or `user_interviews` | | `status` | Filter by order status (e.g., `pending`, `submitted`) | | `detail` | Set to `true` to include full targeting criteria | **Good to know** - **Progress tracking**: each order shows counts for qualified, invited, completed, and pending participants. ### `strella_list_share_links` Lists a project's share links with their resolved settings, participant-facing URL, and participant count. **Parameters** | Parameter | Description | | --- | --- | | `project_id` | The project whose share links to list | | `cursor` | Page cursor. Pass the previous page's `next_cursor` when `has_more` is true | | `limit` | Share links per page (up to 25) | **Good to know** - **Resolved settings**: each link shows its effective settings: a per-link override, falling back to the project value wherever the link inherits. - **What's included**: status, participant-facing URL, participant limit, screener requirement, auto-invite, collected fields, redirect URLs, and participant count. - **Paginated**: up to 25 links per page. Page with `cursor` when `has_more` is true. ### `strella_list_custom_fields` Lists a project's custom fields: the columns on its participants table, including Tags and AI-coded categories, in column order. **Parameters** | Parameter | Description | | --- | --- | | `project_id` | The project whose custom fields to list | **Good to know** - **What's included**: each field's name, type (`text`, `single_select` or `multi_select`), its options, and whether it is the participant identifier that matches a returning participant to their row. - **Values and labels**: each option has a label, which the table shows, and a value, which participants hold. The assistant writes values, not labels. - **Participant values**: a participant's values come with `strella_participant_detail` when the `demographics` group is requested. ### `strella_view_recording` Plays interview recordings in an interactive video player. Returns a ready-to-share `session_share_link`. **Parameters** | Parameter | Description | | --- | --- | | `participant_id` | The participant whose interview recording to play | | `start_time` / `end_time` | Optional. Play a specific clip instead of the full interview | **Good to know** - **Full interview**: pass one participant with no timing to watch all sessions with view switching (split-screen, webcam, screen-share). - **Clips**: pass `start_time` and `end_time` to play a specific segment. Timestamps are continuous across the full interview. ### `strella_list_highlight_reels` Lists a project's Saved Reels with clip counts, sentiment, and a link to open each one. Optionally embeds each reel's clips for content scanning. **Parameters** | Parameter | Description | | --- | --- | | `project_id` | The project whose Saved Reels to list | | `include_clips` | Set to `true` to embed each reel's clips (captions, participants, timestamps) so you can scan contents without opening each reel | | `sort` | Sort order. Newest first by default | | `cursor` / `limit` | Pagination (up to 25 reels per page) | **Good to know** - **Newest first**: reels are listed newest first by default; change the order with `sort`. - **Scan without opening**: `include_clips` embeds each reel's clips (never the quotes) so an assistant can find "which reel covers onboarding" in a single call. - **Deep link**: each reel includes a link to open it in Strella, where you can play, edit, share, and download it. - **Which reels are already shared**: each reel also carries its public share link when it has one, so your assistant can tell shared reels from unshared ones at a glance. Listing never creates a link; ask for one when saving. ## Analysis ### `strella_search_transcripts` Searches interview transcripts using a regular-expression pattern (JavaScript RegExp syntax), or finds the exact start and end of quotes you paste. **Parameters** | Parameter | Description | | --- | --- | | `pattern` | Search pattern (supports regex). Use `*` to match everything. Leave out when sending `quotes` | | `quotes` | Instead of `pattern`: up to 40 quotes, each as its words (and who said it, or the line it was said in, when known). Returns every place each quote was said, with its own start and end time | | `project_id` | Limit search to one project (recommended: faster and more focused) | | `participant_id` | Limit search to one participant's interviews | | `speaker_role` | `moderator` or `participant`. Leave blank to search both | | `date_from` / `date_to` | Only include sessions within this date range | | `context_lines` | How many surrounding lines to include around each match (0 to 10, default 0) | | `count_only` | On by default. Returns counts and a token estimate only, no text. Turn off to get actual matches | | `include_question` | Labels each line with the guide question it was said under, and whether it was the main question or a follow-up | | `item_id` | Search only what was said under one guide question. Matching everything (`.*`) returns every answer to it | | `cursor` | Continues a one-question search on the next page of participants | **How it works** 1. **Filter**: narrows scope using your project, participant, date, and speaker filters at the database level, which is fast and efficient. 2. **Match**: applies your search pattern against the filtered transcript lines. 3. **Return**: sends back match counts, a per-participant breakdown, and a token estimate. If `count_only` is off, includes the matching text too. **Good to know** - **Preview first**: `count_only` is on by default. Check how many matches exist and how many tokens they'd use before fetching the full text. - **Wildcard (`*`)**: a bare `*` matches everything. Use it with a `participant_id` to retrieve someone's full transcript. - **Match limit**: results are capped at 100 lines. If results are truncated, narrow your filters or use a more specific pattern. - **Timestamps**: each match shows continuous timestamps (HH:MM:SS.s, rounded outward to a tenth of a second) across the participant's full interview; they don't restart between sessions. - **Exact quotes**: a matching line is a whole answer, which can run for minutes. To clip just a quote, your assistant sends `quotes` and gets back each quote's own start and end in seconds, ready for a reel. It matches past filler sounds such as "um" and capitalization or punctuation differences. A quote that occurs more than once, or that no line says, comes back marked as such, with what to try next, rather than as a guessed time. Recordings without word-level timing return the whole answer's time range, and your assistant tells you that clip plays the full answer. - **Token estimate**: the response includes an `estimated_tokens` field so you can gauge how much context window the results would use. - **Question labels**: with `include_question`, each line starts with a label such as `Q3 main` (the main question Q3 and its answer) or `Q3 f1` (the first follow-up to Q3 and its answer), plus a list of the questions those labels stand for. The numbers match the ones `strella_get_questionnaire` shows. Lines before the first question read `intro`, after the last `outro`, and lines from human-moderated or uploaded interviews `no-data`. - **Every answer to one question**: pass a question's `item_id` with the pattern `.*` to read what everyone said to it. Answers come a page of whole participants at a time, and each page says how many participants are left. ### `strella_analyze_transcripts` Analyzes the transcripts of a single project on the server, in parallel. Returns per-participant summaries and timestamped quote references for a query. **Parameters** | Parameter | Description | | --- | --- | | `project_id` | The project to analyze (required) | | `query` | Your research question, for example "What frustrations did participants mention about onboarding?" | | `participant_ids` | Optional. Analyze specific participants. If omitted, analyzes everyone in the project (up to 100) | | `question_ids` | Optional. Force specific survey question responses to be included. Overrides auto-detection | | `screener_question_ids` | Optional. Force specific screener question responses to be included. Overrides auto-detection | **How it works** 1. **Scope participants**: identifies which participants to analyze: all of them by default (up to 100), or the specific ones you listed. 2. **Detect structured data**: reads the project's questionnaire and screener, then uses AI to find which questions relate to your query (e.g., "What is their budget?" auto-detects budget-related screener questions). 3. **Analyze each transcript**: reads every participant's full transcript in parallel and produces a summary, relevant quotes, and an applicability judgment for each one. 4. **Attach structured responses**: for each participant, attaches their answers to the detected (or requested) screener/survey questions alongside the transcript analysis. **Good to know** - **Reads every transcript**: the tool reads the complete transcript for each participant, not just snippets. This is what powers the per-participant summaries and quotes. - **Auto-detects structured data**: based on your query, the tool automatically finds relevant screener and survey questions and includes those answers in the results. No configuration needed. - **Applicability filtering**: participants whose interview wasn't relevant to your query are marked as not applicable and excluded from results to save space. The count is still reported. - **Rate limited**: this tool is rate-limited per organization. Each response shows how many calls remain and when the limit resets. - **Single call**: pass all participants in one call; the server handles parallelism internally. Do not split into multiple calls. ### `strella_analyze_question_guide` Analyzes a project's question guide for objective coverage and quality issues. Returns prioritized suggestions and per-objective coverage. It leaves the guide unchanged. **Parameters** | Parameter | Description | | --- | --- | | `project_id` | The project whose question guide to analyze | **Good to know** - **Read-only**: nothing is written or committed. The guide is left exactly as it was; the suggestions are yours to apply or ignore. - **Draft first**: analyzes the unpublished draft guide when one exists, otherwise the published guide. - **What you get**: each research objective is rated for coverage (good, low, or none), with impact-scored suggestions for closing the gaps. Apply a fix with `strella_edit_project`. ## Project Management Tools here that create or change data are available to everyone in your organization except Viewers. See [Security & privacy](/docs/product/mcp/security#who-can-make-changes-through-the-connector). Several also need a skill loaded first: see [Skills](#skills). ### `strella_create_project` Creates a new research project (starts live). Returns a `project_id` for further edits. **Parameters** | Parameter | Description | | --- | --- | | `project_name` | Internal project name | | `participant_facing_name` | Optional. The name participants see. Defaults to the project name | | `description` | Short description participants see on the screener page | | `workspace_id` | Optional. Assigns the project to a workspace | | `moderation_mode` | Optional. `ai` (the default) or `human`, for a study where you run each interview yourself | | `skill_id` | The ID returned when the `question_guide_builder` skill loads | **Good to know** - **Requires skill**: load the `question_guide_builder` skill first; it provides step-by-step project creation guidance. - **Returns `project_id`**: use the returned ID to add questions, screener, and settings via `strella_edit_project`. - **Human-moderated projects:** the assistant writes the screener only, since a human-moderated study has no question guide. You then open the project in Strella, set your availability on the **Scheduling** tab, and publish from **Build**. The assistant can't do those two steps. The moderation mode is set at creation and can't be changed later. See [Schedule moderated interviews](/docs/product/recruit/schedule-moderated-interviews). ### `strella_edit_project` Edits an existing project (settings, question guide, and/or screener) in a single call. **Parameters** | Parameter | Description | | --- | --- | | `project_id` | The project to edit | | `base_hash` | Version check for questionnaire and screener sections. Rejects the edit when the published question guide or screener changed after your AI last read it | | `put` | Set to `true` on a section to replace it entirely instead of merging changes | | `draft_strategy` | What happens when the question guide has changes in the question guide builder that you haven't published. `block`, the default, rejects the edit and saves nothing. `erase` applies the edit and deletes those changes permanently, with no backup | **Good to know** - **Requires skill**: load the `question_guide_builder` skill first. - **Partial updates**: send only the fields you want to change; omitted fields stay as they are. - **Batch changes**: settings, questionnaire, and screener can all be updated in a single call. - **Version safety**: get `base_hash` from the `content_hash` returned by `strella_project_detail` or `strella_get_questionnaire`. Prevents conflicting edits. - **Unpublished guide changes**: a question guide edit saves nothing while the question guide builder holds changes you haven't published. Strella tells your AI to ask you whether to publish or discard them in Strella first, or to erase them. - **Changes the check misses**: the check doesn't see unpublished changes that only touch text formatting, a task's screen-verification screenshots or a section's audience percentage, and the edit replaces them. Publish those changes before you edit the question guide with your AI. ### `strella_prepare_image_uploads` Gives your AI single-use commands that upload image files for your question guide's image blocks, one command per image. Each upload stores the image in your organization, and the AI puts it into its image block when it edits the guide. **Parameters** | Parameter | Description | | --- | --- | | `count` | How many images to upload, up to 10 per call | | `skill_id` | The ID returned when the `guide_image_uploads` skill loads | **Good to know** - **Requires skill:** load the `guide_image_uploads` skill first. - **Needs the files and a command line:** it works only when your AI has the images as files and can run commands, as a coding assistant such as Claude Code can. In a plain chat, the AI adds the image block without its image and gives you a link that opens the guide builder on that block, where you upload the image. - **Same limits as the builder:** PNG, JPEG, GIF or WebP, up to 4 MB each. - **Single use:** each command uploads one image, then stops working. An unused command stops working after 15 minutes. Until then it works like a password, so your AI shows you a command only if you ask for it, for example to run the upload yourself. ### `strella_list_quotas` Lists participant quotas for a project. Returns all quotas in one call. **Parameters** | Parameter | Description | | --- | --- | | `project_id` | The project to list quotas for | **Good to know** - **What quotas do**: quotas set recruitment targets for specific segments, e.g., "interview 10 PMs and 10 engineers". - **Computed status**: each quota shows whether it's open or filled, based on completed interviews vs. limit. - **Quota groups**: can also return the project's quota groups, including empty ones. ### `strella_edit_quotas` Creates or updates a single participant quota. **Parameters** | Parameter | Description | | --- | --- | | `project_id` | The project this quota belongs to | | `quota_id` | Omit to create a new quota; provide it to update an existing one | | `limit` | Target number of participants for this segment | | `criteria` | Defines the segment using screener question answers (e.g., "role is PM") | **Good to know** - **Requires skill**: load the `quotas` skill first. - **Full replacement on update**: when updating, `criteria` is fully replaced. Fetch current criteria with `strella_list_quotas` first if you want to keep existing filters. - **Grouping**: pass `group_id` to put the quota in an existing quota group, or `new_group_name` to create a group for it. ### `strella_edit_quota_groups` Creates or renames a quota group. A quota group is one dimension of your sample plan, such as Gender or Income, and each participant fills one quota in every group they match. **Parameters** | Parameter | Description | | --- | --- | | `project_id` | The project the group belongs to | | `skill_id` | Proof-of-reading token from the `quotas` skill | | `group_id` | Omit to create a new group; provide it to rename an existing one | | `name` | Group name, unique within the project | | `quotas` | Create only: up to 20 quotas to create inside the new group | **Good to know** - **Requires skill**: load the `quotas` skill first. - **One call per dimension**: a split like "25 PMs and 25 engineers" is one call with two quotas. - **No overlap**: quotas in the same group can't share a participant, so the tool rejects criteria that overlap another quota in the group. - **Partial success**: if one of the new quotas fails, the tool keeps the group and the quotas it created before the failure, and the response names the one that failed. - **Deleting a group**: not available through the connector. Delete groups in Strella. ### `strella_write_draft_respondent_order` Creates or updates a draft recruitment order on the **Respondent.io** panel. **Parameters** | Parameter | Description | | --- | --- | | `project_id` | The project to recruit for | | `order_id` | Omit to create a new draft; provide it to update an existing one | | `name` | Public study title shown to potential participants | **Good to know** - **Requires skill**: load the panel-specific skill first; it provides live targeting options (countries, industries, pricing). - **Draft only**: creates a draft. The researcher reviews and finalizes in Strella. Payment and publication happen in the app, not via the tool. - **Returns `submit_url`**: direct the researcher to this link to review and submit the order. ### `strella_write_draft_user_interviews_order` Creates or updates a draft recruitment order on the **User Interviews** panel. **Parameters** | Parameter | Description | | --- | --- | | `project_id` | The project to recruit for | | `order_id` | Omit to create a new draft; provide it to update an existing one | | `name` | Public study title shown to potential participants | **Good to know** - **Requires skill**: load the panel-specific skill first; it provides live targeting options (countries, industries, pricing). - **Draft only**: creates a draft. The researcher reviews and finalizes in Strella. Payment and publication happen in the app, not via the tool. - **Returns `submit_url`**: direct the researcher to this link to review and submit the order. ### `strella_search_recruitment_options` Finds the vendor code for a targeting value you name. Use it for values too large to include in the write-draft skill prompt. **Parameters** | Parameter | Description | | --- | --- | | `panel` | `respondent` or `user_interviews` | | `slug` | The targeting category to search (e.g., skills, occupations, locations, job_titles) | | `query` | What you're looking for (e.g., "React", "London") | **Good to know** - **When to use**: only needed for targeting values too large to inline in the skill prompt (skills, occupations, locations, job titles). Other targeting values are available directly in the skill. ### `strella_edit_share_link` Creates or updates a single share link. A link inherits every project setting until you override it: status, participant limit, screener requirement, auto-invite, collected fields, or redirect URLs. **Parameters** | Parameter | Description | | --- | --- | | `project_id` | The project the share link belongs to | | `skill_id` | Proof-of-reading token from the `share_links` skill | | `share_link_id` | Omit to create a new link; provide it to update an existing one | | `name` | Link label (1 to 80 characters), unique within the project. Required when creating | | `url_slug` | Custom slug the participant URL derives from. Omit for a random token | | `status` | `active`, `disabled`, or `complete` (update only) | | `participant_limit_enabled` / `participant_limit` | Whether the link caps participants, and the cap | | `require_screener`, `auto_invite`, `collect_email`, `collect_names`, `send_reminder_emails` | Per-link settings | | `call_finished_redirect_url`, `screenout_redirect_url`, `quota_full_redirect_url`, `call_dropout_redirect_url` | Where to send participants at each exit point | **Good to know** - **Requires skill**: load the `share_links` skill first; it returns the `skill_id` this tool needs. - **Inherit or override**: every per-link setting defaults to the project value. Pass a value to override it, `null` to revert to the project value, or omit it to leave it unchanged. - **Changing the slug breaks old links**: changing `url_slug` regenerates the participant token, so every share URL you've already distributed for that link stops resolving. The new token takes a fresh random suffix, so re-sending the old slug does not restore it. ### `strella_edit_custom_field` Creates or updates a single custom field on the participants table. **Parameters** | Parameter | Description | | --- | --- | | `project_id` | The project the field belongs to | | `skill_id` | Proof-of-reading token from the `custom_fields` skill | | `field_id` | Omit to create a new field; provide it to update an existing one | | `name` | Column name, unique within the project. Required when creating | | `type` | `text`, `single_select` or `multi_select`. Required when creating | | `options` | Select options in display order, each a label with an optional value | | `is_match_key` | Makes a `text` field the participant identifier | **Good to know** - **Requires skill**: load the `custom_fields` skill first; it returns the `skill_id` this tool needs. - **Options replace the whole list**: send every option the field should keep. To rename an option without losing the participants who hold it, send its existing value with the new label. - **Changing the type keeps values as they are**: participants' stored values are not converted to the new type. - **No delete**: deleting a field removes every participant's value in it, so delete fields from the participants table in Strella. ### `strella_set_participant_custom_fields` Sets the same custom field values, such as tags, on up to 25 participants in one project. **Parameters** | Parameter | Description | | --- | --- | | `project_id` | The project the participants belong to | | `skill_id` | Proof-of-reading token from the `custom_fields` skill | | `participant_ids` | The participants to update, 1 to 25 | | `custom_fields` | Each field's id and the value it should hold: text, one option value, a list of option values, or `null` to clear it | **Good to know** - **Requires skill**: load the `custom_fields` skill first. - **Lists replace the whole value**: a `multi_select` value replaces the participant's current list, so the assistant reads their tags first to add one. - **Only custom fields change**: a participant's status, name and quotas stay as they are, and fields left out keep their values. - **Partial results**: the result lists which participants were updated, which were refused (not found, or holding an identifier value another participant already has), and which were not attempted when an error stopped the run. ### `strella_save_highlight_reel` Saves interview clips into a project as an editable Saved Reel: the same reels you can play, edit, share, and download in Strella. Use it to keep a set of strong moments your AI assistant found while reviewing interviews. **Parameters** | Parameter | Description | | --- | --- | | `project_id` | The project the reel is saved into. Clips can come from participants in any of your organization's projects | | `title` | Reel title shown on the Saved Reels page. Used exactly as written | | `description` | Optional summary shown in the reel's Overview | | `clips` | The clips in playback order (1 to 25). Each clip is a participant plus a start and end time, with a short caption and a sentiment. The reel keeps the order exactly | | `create_share_link` | Set to `true` to get a public share link back with the saved reel. Send it on its own with `reel_id` to get the link for a reel that is already saved | **Good to know** - **Requires skill**: load the `highlight_reels` skill first; it returns the `skill_id` this tool needs. - **Clips can span projects**: the reel lives in one project, but its clips may come from any project in your organization. When clips span several projects, your assistant asks which project to save the reel into. - **All-or-nothing**: if any clip is invalid (unknown participant, unplayable recording, out-of-range time), the whole save is rejected with a per-clip reason and nothing is saved. - **Continuous timestamps**: clip times are continuous seconds across a participant's full interview, the same frame `strella_search_transcripts` and `strella_view_recording` use. - **Returns a link**: the response includes an in-app link to the saved reel, where you can play, edit, share, and download it. See [Create & share a highlight reel](/docs/product/share/highlight-reels). - **One public share link per reel**: ask for it with `create_share_link` and the response also carries `share_url`, a link anyone can play without a Strella account. A reel keeps that one link: every later save through your assistant refreshes it to the reel's current clips, so the link you already sent never plays an old cut. To get links for several saved reels at once, your assistant lists them and asks for each reel's link in turn, with no video playing in the chat. ## Utilities ### `strella_switch_organization` Switches the active organization for all subsequent tool calls. Requires confirmation on each use. **Parameters** | Parameter | Description | | --- | --- | | `organization_slug` | The organization slug to switch to (slugs are listed in the guide response) | **Good to know** - **When to use**: only needed for users who belong to multiple organizations. Most users have one and never need this. ### `strella_skills` Loads detailed step-by-step instructions for a specific workflow, or lists all available skills. **Parameters** | Parameter | Description | | --- | --- | | `skill_name` | The skill to load (e.g., `question_guide_builder`, `quotas`). Omit to list all available skills | **Good to know** - **Auto-loaded for writes**: write skills (`question_guide_builder`, `quotas`, recruitment) are loaded automatically when your AI assistant detects a matching intent. The skill returns a `skill_id` required by the corresponding write tool. - **Detailed guidance**: skills provide step-by-step instructions, live config data, and validation rules for complex workflows. - **See Skills below**: each skill is documented in the [Skills](#skills) section with its description, auto-loading behavior, and scope. ### `strella_search_help_docs` Searches the Strella help docs (these product guides) with a regular-expression pattern, or fetches one full docs page by slug. Your AI assistant uses it to answer "how does X work in Strella" questions from the real documentation instead of improvising. **Parameters** | Parameter | Description | | --- | --- | | `pattern` | Search pattern (supports regex). Provide `pattern` or `slug` | | `slug` | Exact docs page slug (e.g., `recruit/quotas`) or docs URL. Returns that page in full | | `section` | Limit the search to one docs section (e.g., `recruit`, `analyze`) | | `context_lines` | How many surrounding lines to include around each match (0 to 10, default 2) | **How it works** 1. **Search**: greps every help-doc page for your pattern and returns matching lines with context, grouped with a per-page match count sorted by relevance. 2. **Follow up**: when excerpts aren't enough, the assistant fetches the best page in full via `slug`. 3. **Cite**: every result carries a working link to the docs site, so answers cite their source. **Good to know** - **Docs, not interviews**: this searches the product documentation. For what participants said in interviews, the assistant uses `strella_search_transcripts` instead. - **Honest misses**: zero matches means the docs don't cover the topic; the assistant is told to say so rather than guess. - **Changelog on request**: release notes are excluded from search (they mention every feature name), but the full changelog is available by asking for it directly (`slug: changelog`). - **Match limit**: results are capped at 40 matching lines, at most 8 per page so one dense page can't crowd out the rest; the assistant narrows by section or pattern, or opens a page in full. ### `strella_send_feedback` Sends the Strella team feedback your AI assistant writes with you: a bug where a Strella tool failed, an idea for something Strella could do, or a note about something that was confusing. **How to start** - **From your AI tool's prompt menu**: AI tools that list connector prompts, such as Claude and Claude Code, show **Send feedback to Strella** (in Claude Code, as a slash command). You can add a sentence in your own words. - **By asking**: say "send this feedback to Strella" or "report this problem to Strella". This works in every AI tool, including ChatGPT. Your assistant loads the `feedback` skill first, which tells it what to write and to show you the draft. **Parameters** | Parameter | Description | | --- | --- | | `kind` | `bug`, `idea`, `confusing`, or `other` | | `goal` | What you were trying to do, in one or two sentences | | `details` | What happened and what would help (optional) | | `failed_calls` | For a bug: one line per Strella tool call that failed or confused the assistant, with what it sent and what came back (optional) | | `expected` | For a bug: what the assistant expected instead (optional) | | `suspected_cause` | For a bug: the assistant's best guess at the cause, and how sure it is (optional) | | `failed_tool_names` | Comma-separated names of the Strella tools involved (optional) | **Good to know** - **You approve it first**: your assistant shows you the full draft and sends it only after you agree. - **Private details are removed**: Strella replaces anything that looks like an email address, a phone number, or a credential such as an API key with a label before sending. Your assistant shows you the feedback exactly as sent. - **Participant details stay out**: the feedback refers to people and projects by id, and leaves out names and interview text. - **A reference to quote**: feedback usually returns a reference you can mention if you contact [customersupport@strella.io](mailto:customersupport@strella.io). - **Limit**: up to 10 pieces of feedback per person per hour. ## Skills Skills are detailed step-by-step guides that provide domain knowledge for specific workflows. Your AI loads a skill with the `strella_skills` tool, or omit the skill name to list every available skill. Many skills load automatically when the AI detects a matching intent. Two badges describe how a skill behaves: - **skill_id**: the skill returns a proof-of-reading token required by a write tool before it will run. - **dynamic**: the skill fetches live data from the API each time it loads (e.g., current pricing or targeting catalogs). ### `tutorial` Hands-on walkthrough you steer from a short menu: ask a research question across the interviews in a project, build a simple test project, or find a moment in a recording and save it as a highlight reel. Suggested for first-time users. _Automatically suggested to new users when the guide detects a first-time session._ ### `question_guide_builder`   `skill_id` Complete guide for creating and editing research projects, questionnaires, screeners, and moderator settings. Covers the full build and edit workflow including question types, follow-up depth, and screener qualification logic. _Automatically loaded when you ask to create or edit a project. The returned `skill_id` is required by `strella_create_project` and `strella_edit_project`._ ### `question_guide_exact_importer`   `skill_id` Import an existing research guide (Google Doc, Word, or pasted text) verbatim into Strella, with no rewriting or inferring. Use when you have an existing discussion guide to port 1:1. ### `question_guide_media_blocks` Media block examples and documentation: website, Figma, video, image, and embed blocks with tasks and questions. Load when building a guide that shows media to participants. ### `guide_image_uploads`   `skill_id` Upload image files into a question guide's image blocks. Covers when an upload applies, preparing the commands, uploading each image, writing every image into the guide in one edit, and what to do when an upload is refused. _Automatically loaded when your AI needs to upload an image. The returned `skill_id` is required by `strella_prepare_image_uploads`._ ### `question_guide_ab_testing` A/B testing setup with variant blocks and audience rules for segment-targeted sections. Load when the guide needs A/B comparisons or segment-specific questions. ### `plan_changes` Generate a reviewable markdown plan before creating or editing a project. Shows the proposed structure for approval before making any changes. ### `quotas`   `skill_id` Create and update participant quotas: segment targets defined by screener criteria with a fill limit and behavior. Covers DNF criteria format, full-replacement update semantics, and screener-coupling requirements. _Automatically loaded when you ask to create or edit quotas. The returned `skill_id` is required by `strella_edit_quotas`._ ### `share_links`   `skill_id` Create and update participant share links: per-link outreach URLs whose settings inherit from the project unless overridden. Covers create vs update, the override rule (a value overrides, `null` reverts to the project value, omit leaves unchanged), and the `url_slug` change that breaks already-distributed links. _Automatically loaded when you ask to create or edit a share link. The returned `skill_id` is required by `strella_edit_share_link`._ ### `custom_fields`   `skill_id` Create and edit custom fields on the participants table, and set participants' values in them. Covers option values and labels, full-replacement options, type changes, and the participant identifier. _Automatically loaded when you ask to add or edit a column or tag participants. The returned `skill_id` is required by `strella_edit_custom_field` and `strella_set_participant_custom_fields`._ ### `highlight_reels`   `skill_id` Save a set of interview clips into a project as a Saved Reel. Covers how clips are addressed (participant plus continuous timestamps), building the reel in playback order, the all-or-nothing validation that rejects a save if any clip is invalid, and how to get a reel's public share link. _Automatically loaded when you ask to save a highlight reel. The returned `skill_id` is required by `strella_save_highlight_reel`._ ### `research_explorer` Step-by-step workflow for exploring and analyzing research data. Provides structured guidance for navigating projects, participants, and transcripts. ### `capabilities` Complete reference of all tools and what they can do: read, analyze, write, and watch. Load when you want to see everything available. ### `whats_new` The connector's latest additions, newest first, with a prompt to try for each item. The list matches the [What's new](/mcp/whats-new) page, so your AI can answer "what's new?" or suggest something you haven't tried yet. ### `feedback`   `skill_id` Write feedback for the Strella team: a bug, an idea, or something confusing. Covers when to offer it, what each kind needs, keeping it to this conversation and free of participant details, and showing you the full draft before anything is sent. _Automatically loaded when you ask to send feedback. The returned `skill_id` is required by `strella_send_feedback`._ ### `write_draft_respondent_order`   `skill_id`   `dynamic` Draft a Respondent.io recruitment order. Fetches live panel config (countries, industries, demographics, pricing) each time it loads so targeting options are always current. _Automatically loaded when you ask to set up recruitment via Respondent.io. The returned `skill_id` is required by the write tool._ ### `write_draft_user_interviews_order`   `skill_id`   `dynamic` Draft a User Interviews recruitment order. Fetches live characteristic catalog (slugs, operators, pricing) each time it loads so targeting options are always current. _Automatically loaded when you ask to set up recruitment via User Interviews. The returned `skill_id` is required by the write tool._ ## Related Setup for Claude, ChatGPT, Figma Make, and more. See these tools in action with copy-paste prompts. --- # Use cases & example prompts Source: https://app.strella.io/docs/product/mcp/use-cases Everything here works through natural conversation. Copy a prompt, paste it into your AI assistant, and adjust the project or participant names to match your own data. The prompts are grouped by what they do: - **Explore**: browse your data safely. Nothing changes. - **Analyze**: get AI-powered insights from your interviews. Still read-only. - **Create**: creates or edits projects. Updates your Strella data, so it isn't available to Viewers. ## Explore | Prompt | What it does | | --- | --- | | Show me all my research projects | Lists every project in your organization with status, dates, and participant counts. | | Show me the details of the "User Onboarding" project | Returns project settings, research objectives, interview count, and status breakdown. | | List all workspaces in my organization | Shows how your projects are grouped by team, department, or client. | | Which of my projects have the most completed interviews? | Compares projects by completion rate to show your most active studies. | | Show me the questionnaire for the "Mobile App UX" project | Returns the full interview script: questions, follow-ups, and sections. | | Show me the full transcript for participant Jane Doe | Retrieves the complete interview transcript for a specific participant. | | Play the session recording for participant #3 in Project X | Opens an embedded video player with the interview recording. | ## Analyze | Prompt | What it does | | --- | --- | | What did participants say about the onboarding experience? | Analyzes transcripts for a project and returns themes, summaries, and quoted evidence. | | Summarize the key findings from Project X | Generates a high-level summary across all completed interviews in a project. | | What pain points did users mention most frequently? | Cross-references participant responses to identify recurring themes and frustrations. | | Search transcripts for mentions of "pricing" in Project Y | Finds specific keywords or patterns across all interview transcripts. | ## Create | Prompt | What it does | | --- | --- | | Create a new project called "Q1 Customer Feedback" | Sets up a new study that you can then configure with questions and a screener. | | Add three interview questions about user satisfaction to Project X | Updates the interview questionnaire with new questions. | | Update the research objectives for Project Y | Edits project settings to reflect new or refined research goals. | Create prompts work for everyone in your organization except Viewers. Your AI tool typically asks for your approval before it changes anything, depending on its settings. See [Security & privacy](/docs/product/mcp/security#who-can-make-changes-through-the-connector). ## Worked examples A closer look at a few common requests and the steps the connector takes for each. ### Explore research findings > Show me all my live projects and summarize the key findings from the most recent one. **What happens** - Lists your projects filtered by status. - Retrieves the most recent project's details and sampled interview data. - Shows high-level findings and insights, at your request. ### Analyze transcripts across participants > Analyze the transcripts in my 'Mobile App Usability' project. What positive feedback did participants share about the mobile app? **What happens** - Fetches all transcript data from the specified project. - Uses AI to summarize what all participants said. - Returns themed patterns with supporting quotes. ### Review a participant's interview > Pull up the details for participant John Smith. I want to see his interview summary and screener responses. **What happens** - Lists participants and finds them by name (if you have personally identifiable information (PII) scope). - Retrieves their interview summaries and screener responses. - Provides a complete review of their research session. ### Watch a session recording > Show me the recording for participant Jane Doe's most recent session. **What happens** - Looks up the participant and their sessions. - Opens an embedded video player with the session recording. ### Set up a new research project > Create a new project called 'Checkout Flow Usability' in my default workspace. Add 5 interview questions exploring pain points in the purchase flow, a screener to filter for users who have shopped online in the last 30 days, and configure the AI moderator to probe for specific friction points. **What happens** - Creates a new project in your workspace. - Configures the question guide, screener, and AI moderator. - Asks you for any missing details and suggests next steps. - Leaves the project ready for recruitment and interviews. ## Related The tools behind each of these prompts. Get set up before trying these prompts. --- # What's new Source: https://app.strella.io/docs/product/mcp/whats-new ## October 27, 2026 **New** - Strella connector can blur a question's recording, read a question word for word, or wait for the participant before moving on. - Try: "Read the pricing question exactly as written." - Strella connector can set a Figma task's type, start screen and goal screen. - Try: "Make the checkout task end on the confirmation screen." - Strella connector can mark a choice exclusive, like "None of the above". - Try using: `edit_project` - Strella connector can set a share link to ask for first names only. - Try: "Only ask for first names on my newsletter link." ## October 22, 2026 **New** - Strella connector shows every project, share link, guide and screener setting. - Try: "Check the skip logic on my screener." - Strella connector can open an embed or Figma prototype at a new page when a question starts. - Try using: `edit_project` **Fixes** - `edit_project` no longer says nothing changed after a skip logic edit. ## October 15, 2026 **New** - Added `prepare_image_uploads`, so an assistant with file access, like Claude Code or Cursor, can upload images into a guide's image blocks. - Try: "Put these three screenshots into the image blocks of my onboarding project." - Strella connector shows who booked an interview on a human-moderated project, when it starts, and the link to join it. - Try: "Who is booked for an upcoming interview on this project?" - Strella connector can turn on "Block interaction until answered" for an embed or Figma prototype in a guide. - Try using: `edit_project` **Fixes** - `list_participants` no longer drops recording and summary fields when you repeat the same request. ## October 13, 2026 **New** - Strella connector can read and write matrix questions in a screener, including skip logic based on a matrix answer. - Try using: `edit_project` - Strella connector can add an image block to a guide before its image exists, so you can upload the image in the guide builder later. - Try using: `edit_project` ## October 6, 2026 **New** - Strella connector can tell you which guide question a transcript line belongs to, and whether it was the main question or a follow-up. - Try: "What came up in the follow-ups to the pricing question?" - Strella connector can read every answer to one guide question across a whole project. - Try: "What did everyone say when asked about onboarding?" ## September 15, 2026 **New** - Strella connector can create a public share link when saving a highlight reel. - Try: "Create a shareable reel of the best moments from this interview." - Strella connector can add a text block to a question guide: written copy participants read and react to. - Try using: `edit_project` ## September 8, 2026 **Fixes** - Strella recordings no longer freeze between clips of a reel. ## September 1, 2026 **New** - Strella connector can group quotas so a participant only fills one quota per group. - Try using: `edit_quota_groups`, `edit_quotas`, `list_quotas` **Fixes** - `list_participants` no longer drops `updated_at` when a caller asks for it. ## August 20, 2026 **New** - Strella connector can create and read NPS questions. - Try using: `edit_project` ## August 13, 2026 **New** - Strella connector can turn the screenshot-deterrent watermark on or off for a project's media. - Try using: `edit_project` **Fixes** - `edit_project` no longer wipes a media block's URL when saving a guide. ## August 11, 2026 **New** - Added `search_help_docs`, so a connected assistant can look up Strella help articles directly. - Try: "How do I use a report in Strella?" ## August 7, 2026 **New** - Strella connector can turn greeting participants by name on or off. - Try using: `edit_project` - Strella connector can create and read matrix questions. - Try using: `edit_project` **Fixes** - `edit_project` no longer rejects a guide write because of a rating-scale question. ## August 4, 2026 **New** - Strella connector can pull clips from several projects into one reel. - Try using: `save_highlight_reel` ## July 30, 2026 **New** - Added `analyze_question_guide` (replacing `analyze_guide`), which reviews a question guide against a project's research objectives. - Try: "Does my guide cover all of my research objectives?" ## July 28, 2026 **New** - Strella connector can list, view, save, and edit highlight reels. - Try using: `list_highlight_reels`, `save_highlight_reel`, `view_recording` ## July 21, 2026 **New** - Strella connector can list, create, and update participant share links. - Try using: `edit_share_link`, `list_share_links` - Strella connector can turn auto-invite on for a recruitment order. - Try using: `write_draft_respondent_order`, `write_draft_user_interviews_order` ## July 15, 2026 **New** - Strella connector can create and read card sort questions, including randomized card order. - Try using: `edit_project` ## July 9, 2026 **New** - Strella connector can create and read file-upload questions. - Try using: `edit_project` ## June 25, 2026 **New** - Strella connector can read whether a participant passed or failed each task, and filter by it. - Try using: `participant_detail` ## June 23, 2026 **New** - Strella connector separates general project context from moderator-only context more clearly. ## June 12, 2026 **New** - Strella connector can hand out a shareable link to a participant's interview recording. - Try using: `participant_detail`, `view_recording` ## June 9, 2026 **Fixes** - `edit_project` no longer saves a moderator voice speed outside the allowed range. ## June 2, 2026 **New** - Strella connector can read and set device and interview settings: push to talk, camera and screen blur, screen-share verification, captions, session timeout, and participant-name anonymization. - Try using: `create_project`, `edit_project` ## May 20, 2026 **New** - Strella connector can create and read rating scale questions (Likert, faces, stars), and cap how many options a multiple-choice question accepts. - Try using: `edit_project` ## May 12, 2026 **New** - Added the `question_guide_exact_importer` skill, so an assistant can port an existing guide from a Google Doc or Word file into Strella if format matches one-one with Strella primitives. - Try: "Import this discussion guide exactly as designed." - Strella connector can gate a guide section to an audience defined by screener answers, including slider ranges. - Try using: `edit_project` - Strella connector can write screener skip logic and limit how many variants each participant sees. - Try using: `edit_project` - Strella connector can filter projects, participants, and recruitment orders by when they last changed. - Try using: `search_projects`, `list_participants`, `list_recruitment_orders` **Fixes** - `edit_project` no longer wipes a screener's existing skip logic when saving. ## April 28, 2026 **New** - Strella connector can list, create and update draft recruitment orders. **Fixes** - `edit_project` section conditions with slider ranges, negations, or combined joins are now handled correctly. ## April 23, 2026 **New** - Strella connector can list, create and update participant quotas. - Try using: `list_quotas`, `edit_quotas` - Strella connector can set the language participants start an interview in. ## Initial releases (February 25 to April 14, 2026) - Strella connector shipped with base set of tools & capabilities --- # Control access & scheduling Source: https://app.strella.io/docs/product/recruit/access-and-scheduling The **Access & scheduling** settings tab controls who can join your study and when it closes. Most teams set these once early in a study and forget about them. ![Settings → Access & scheduling: toggles to close the project, schedule a close date, set a participant limit, add a custom closed message, and auto-disqualify by quota.](/docs/recruit/access-scheduling.png) Auto-invite, the screener requirement, and email/name collection are set per share link, not study-wide. See [Manage share links](/recruit/share-links). For panel recruitment, auto-invite is set on each panel order. ## Steps Click the gear icon in any study page. Three independent toggles end recruitment automatically: - **Close project to new participants**: flip this immediately to block all new arrivals. The study stays open for existing in-flight participants. - **Schedule a close date**: pick a date and time; the study closes automatically at that point. - **Set a participant limit**: the study closes automatically once N participants have completed. Use any combination. First condition that's met closes the study. Two toggles work alongside your [quotas](/recruit/quotas): - **Auto-disqualify when quota filled**: when a quota fills, any further participants matching it are auto-disqualified instead of sitting in Qualified waiting for review. - **Auto-disqualify when no quota matches**: participants whose answers don't fit any active quota are auto-disqualified. (No effect when you have no quotas.) Use these to keep the Participants tab clean when you have tight segment caps. When the study is closed (manually or via schedule/limit), participants who click your share link see a default "study closed" page. Turn on **Custom closed message** to write your own, usually a thank-you plus a link to your next study, your site, or a wait-list signup. Whatever combination you've configured runs in the background. You can re-open the study anytime by flipping **Close project** back off, provided your scheduled close date hasn't passed and you're under the participant limit. You can also close or reopen a study straight from its project-list row on the dashboard: click its status and pick **Close project** or **Reopen project**. Reopen there is disabled (with the reason) when a scheduled close date or participant limit is what's keeping the study closed. ## Tips - **A participant limit is safer than a panel order limit.** An order limit caps how many candidates the panel sources. A participant limit caps how many complete the interview, which is what most teams care about. When your study reaches its participant limit or closes, Strella pauses its panel orders, so the panel stops sending people. Raise the limit or reopen the study within 5 business days and the orders resume. Otherwise, Strella closes each order and [refunds](/recruit/hire-from-panel) the spots you didn't use. While the study is full or closed, you can't add participants to a panel order. - **Strella releases panel participants who arrive too late.** If someone from a panel order opens the study after it filled or closed, Strella releases them back to the panel. They move to the **Disqualified** tab with a note that the study was full or closed when they arrived. ## Related The quotas that drive the auto-disqualify rules. What happens when auto-invite is off. The closed-message branding lives here. --- # Approve, reject & invite Source: https://app.strella.io/docs/product/recruit/approve-reject-invite The **Participants** table in Recruit is where you triage incoming candidates. Each row is one person. Tabs along the top group them by status: **Qualified**, **Invited**, **Completed**, **Disqualified**. ## Steps The default view shows everyone who's currently **Qualified**: they've passed the screener but haven't been invited yet. People keep applying while you have the page open. Select **Refresh**, at the right of the toolbar next to the **⋮** menu, to pull in everyone who has applied since you opened it and update the counts on each tab. Scroll right in the table to see each participant's screener answers in their own columns, or click a row to open the detail side panel (answers, demographics, and metadata). When available, the **AI rating** column shows an overall 1 to 5 star score. Hover the rating for the per-criterion breakdown: each criterion (**Research Fit**, **Consistency**, **No Gaming**, **Engagement**) is marked **Pass**, **Warning**, or **Fail** with a short reason. ![The Participants table with an AI Rating column, and the hover breakdown showing Research Fit, Consistency, No Gaming, and Engagement each marked Pass.](/docs/recruit/participant-rating.png) With a participant selected (in the row or the detail side panel): - **Invite**: sends them an email with a link to join the interview. They click it and start, with nothing to schedule. Someone who came in through your own share link or CSV needs an email address for this, so a participant without one has no **Invite** button, and a bulk invite skips them and tells you how many it skipped. - **Reject**: disqualifies them from the study. Useful when screener logic let them through but they don't fit on closer review. For bulk actions, multi-select rows and use the bulk action bar at the top. Auto-invite is a setting on each audience, not a toggle on this table. Open **Recruit → Audiences**, then use a row's **⋯** menu: **Manage** for a panel order, or **Manage share link → Link settings** for a share link. Switch on **Auto-invite qualified participants** and Strella invites anyone the AI rates 4 to 5 stars automatically. The Qualified tab stays empty for that audience; those participants land directly in Invited. Use auto-invite when you trust the screener to filter cleanly and you don't need to review case-by-case. Tags are study-specific labels you can add to any participant (e.g. "key account", "follow up later"). Add them in the **Tags** column of the Participants table, or from the participant's **Profile** tab when you open their interview. Useful for filtering after the interview when you want to compare reactions across a custom segment. Invited participants get an email with a link to join the interview. They click it and start, with no time to book. ## Tips - **AI ratings are guidance, not policy.** A low-rated participant might still be a great interview if the topic fits their experience. Hover the rating for the per-criterion breakdown before rejecting on the overall star count alone. - **The Completed tab is your bench** of finished interviews: click in to jump to the recording, transcript, and AI analysis. (Or use [Analyze → Review a single interview](/analyze/review-interview) directly.) ## Related See the table's tabs as conversion steps. Cap how many of each segment can pass. What to do once an interview finishes. Stop the payout and flag fraud after an interview. --- # Bring your own participants Source: https://app.strella.io/docs/product/recruit/bring-your-own If you already know who you want to talk to (customers, prospects, employees, partners), you don't need a panel. You bring them in via a share link or a CSV upload. ## Before you start You'll need: - A list of people to invite, or a place to share the link. - _Optionally_, a published screener: add one to qualify people before they interview; skip it for a known list (your own customers, say) you don't need to screen. ## Steps Click **Share** in the study header. Strella creates a share link participants can click. By default, the link does everything in one sitting (screener, qualification, and interview) with nothing for the participant to schedule. See [Manage share links & redirects](/recruit/share-links) for multiple links and per-link customization. Drop the link into an email, a calendar invite, a slack message, a "feedback" page on your site, wherever your audience is. In **Recruit → Participants**, open the three-dot menu and choose **Import CSV**, then select **Add new participants**. The other option, **Update existing participants**, only adds custom data to people already in your study. Include a name column and an email column. On the mapping step, confirm the email column maps to **Email Address**. Strella invites people by email, so you can't invite an imported person without one from the Participants table. Send that person the share link yourself, and [mark an ID column as the participant identifier](/recruit/url-parameters#match-the-same-person-across-visits) so their click joins the row you imported. Other columns become participant metadata that you can filter and tag on. In **Recruit → Audiences**, open the share link's row or choose **Manage share link** from its menu, then turn on **Auto-invite qualified participants** for that link. Strella can then invite qualified participants who arrive through that link automatically, within remaining audience capacity and quotas. This suits high-trust audiences, such as your own customers. Leave it off for unknown lists if you want to vet each candidate first. As people click your link (or you import them), they appear in **Participants**. The tabs along the top are **Qualified**, **Invited**, **Completed**, and **Disqualified**. ## Tips - **If you use a screener, it's the gatekeeper.** Even for a known list, the screener still runs: set quotas and qualification logic so an unknown contact in your CSV doesn't slip through. (No screener? Everyone who joins goes straight through.) - **For internal studies**, use [custom share links](/recruit/share-links) per audience (one for customers, one for employees, etc.) so you can tell who came from where. - **CSV imports don't send anything automatically.** The imported people wait in **Qualified** until you invite them there, which emails them, or until you send them the share link yourself. ## Related Multiple links per study, per-link customization. What to do once participants arrive. --- # Watch the funnel Source: https://app.strella.io/docs/product/recruit/funnel The funnel is one chart that tells you where participants drop. If you're getting clicks but few completed interviews, this is the first place to look. ![The recruitment funnel: stage-by-stage bars from All through Completed Session, with conversion percentages and counts, plus a breakdown table below.](/docs/recruit/funnel.png) ## Steps In **Recruit**, open the **⋮** menu beside the tabs and choose **Recruitment funnel**. It shows each stage with the count of participants who reached it. If you don't see the menu, the funnel isn't turned on for your organization yet. The full set of stages, in order: 1. **Clicked link**: everyone who opened your share link. 2. **Started screener**: answered at least one screener question. 3. **Completed screener**: answered every screener question. 4. **Qualified**: passed the screener and any active quotas. 5. **Registered**: provided their contact details (when you collect name/email). 6. **Invited**: received the invite to join the interview. 7. **Opened invite**: opened the invite link. 8. **Started interview**: joined the call. 9. **Completed interview**: finished the interview. Each stage shows its count plus conversion from the previous stage. The funnel skips stages that don't apply to your study configuration (see below). Use the date filter at the top to look at a specific window: last 7 days, last month, custom range. Expand a stage to see breakdowns by [share link](/recruit/share-links) or [quota](/recruit/quotas). ![Hovering a funnel stage to reveal its breakdown by recruitment source.](/docs/recruit/funnel-demo.gif) ## How the funnel varies by study config Not every study uses every stage: - **No screener**: the "Started screener" and "Completed screener" stages are skipped. Participants go directly from arrival to Qualified (if access is open) or Invited. - **Auto-invite on**: when [Auto-invite](/recruit/access-and-scheduling) is on, qualified participants are invited without manual approval; the time between Qualified and Invited collapses. - **No name/email collection**: share-link studies that don't collect a name or email skip the registration step entirely. Participants can go straight from qualifying to the call. The stage list above is the full default; your funnel may show fewer stages depending on which steps your study config skips. ## Tips - **Big drops at "Started screener"** usually mean the landing page or first question discourages people. Make question 1 the easiest one. - **Big drops between "Completed screener" and "Qualified"** usually mean your screener is too strict, or quotas are filling fast. Loosen one criterion at a time. - **Big drops between "Invited" and "Started interview"** usually mean the invite isn't landing: it went to spam, or there's no reminder nudging people back. See [Access & scheduling](/recruit/access-and-scheduling). ## Related The reference of every stage and what it means. The source-level breakdown is here. --- # Hire from a participant panel Source: https://app.strella.io/docs/product/recruit/hire-from-panel Use a panel when you don't have a list of your own participants. Strella places the order, the panel sources matching candidates, and Strella handles the participant payouts. ## Before you start You'll need: - A published screener: orders require one. An order uses the published version of your screener and guide, so publish any edits before you pay. - An idea of your audience (industry, role, demographics): you'll target on these. - An idea of how many interviews you want. ## Steps In **Recruit → Audiences**, click **Recruit participants** on the **Recruit from our panel** card. After you pick a panel, the flow has three steps: Audience → Details → Payment. Choose between **Respondent** and **User Interviews**. They have different strengths: | Panel | Strengths | | ------------------- | ---------------------------------------------------------------------------------------------------------- | | **Respondent** | Faster, lower cost, broader international reach, strong general-population coverage. | | **User Interviews** | Better B2B coverage; tighter quality controls reduce the rate of fraudulent participants. | Pick the one that fits your study. You can also place two orders if you want sample from both. Pick **B2B** or **B2C** as the market type and describe your audience. Each panel exposes its own set of filters (demographic, professional, geographic), and they're presented inline in the order flow. A live feasibility forecast updates as you tighten filters and tells you whether your target sample is realistic at your incentive. On Respondent, each filter has a limit on how many options it takes: | Filter | Market | Limit | | --------------- | ----------- | ----- | | Research topics | B2C | 2 | | Job titles | B2B | 10 | | Job functions | B2B | 3 | | Industries | B2B | 5 | | Locations | B2B and B2C | 25 | Respondent rejects an order over any of these limits, so the order won't move on to the next step until every filter is within its limit. Filters suggested from your audience description stay within the limits and use only options from Respondent's own lists. On Respondent, a filter either limits who can apply or only sets who Respondent invites first: - **Who can apply:** location, age, gender, education, ethnicity, job function and seniority, plus household income on B2B orders. A panelist who doesn't match can't apply. - **Who Respondent invites first:** research topics, industries and company size, plus household income on B2C orders. A panelist who doesn't match can still apply, so add a screener question for anything that must be true. - **Job titles** tell Respondent which job family to look in. Panelists in any family your titles point to can apply, even with a different title. - **Sample size** is how many interviews you want. - **Incentive amount** is what each participant gets paid. The forecast updates as you change it. - **Estimated interview length** sets the recommended incentive. Each panel sets its own minimum per participant: **$5 on Respondent**, **$1 on User Interviews**. The incentive field names its minimum and won't take less, and the recommended amount never falls below it, so a very short interview still recommends at least the minimum. Respondent also caps the incentive at **$200** per participant, and the field won't take more. Title, description, screen-out copy. These show in the panel's invite to potential participants. Be honest about what the study involves, since over-promising costs you in screener drop-offs. On Respondent, the description is up to 300 characters. The final step shows your full order plus the total cost. Submit to send the order to the panel. If your screener or guide has edits you haven't published, a warning appears above the payment options. Click **Open screener** to publish them in a new tab, then come back to this tab to pay. The warning doesn't stop you from paying. The order appears in **Recruit → Audiences** as **Active** while the panel sources candidates. Candidates show up in your **Participants** tab as they pass the screener. Invite them from the **Qualified** tab to send them the interview. ## Tips - **Save an order as a template** to reuse its audience configuration on future studies, which is useful for ongoing research with a fixed audience. (In the order's action menu: **Save as template**.) - **Don't over-tighten the filters.** Each filter that decides who can apply narrows the pool. If the feasibility forecast turns red, loosen one before submitting. - **Some order details can be edited after launch** (e.g. raising the sample size). Open the order from **Recruit → Audiences** to see what's still mutable. Some fields lock once the panel has begun recruiting. - **You're refunded for spots you don't use.** If an order doesn't fully fill, closing it refunds the unused participants (the shortfall against your target × the cost per participant) back to your credits or original payment method. ## Related Required before placing an order. What to do when candidates start arriving. Cap segment counts inside an order. --- # How recruitment works Source: https://app.strella.io/docs/product/recruit/overview Recruit is the tab where you decide _who's_ in your study. Whether participants come from a paid panel or your own list, they all appear in the same place: the **Participants** table. ## The two paths ### Hire from a panel Strella integrates with two participant panels. You place an order through Strella, specify your audience, and the panel sends you qualified participants. Strella handles payouts. **Use when** you don't already have access to your target audience. ### Bring your own You generate a share link in Strella and send it out yourself to people you already know. They click in, take your screener, and (if they qualify) are invited to take the call. **Use when** you already have a list: customers, prospects, employees, a partner audience. ## How they connect Every participant, whether from a panel or BYO, goes through the same pipeline: 1. **Lands on the screener.** They answer your screener questions. 2. **Qualifies or doesn't.** Based on screener answers and quotas. 3. **Invited to the interview.** Invite them from the **Qualified** tab (or let [auto-invite](/recruit/access-and-scheduling) do it) and they get a link to join. 4. **Takes the interview.** They click the link and the AI moderator runs them through your guide. There's no slot to book. In a human moderated study they book a time from your availability instead; see [Schedule moderated interviews](/recruit/schedule-moderated-interviews). 5. **Shows up in Results.** Recording, transcript, themes. You'll see all participants, regardless of how they got here, in **Recruit → Participants**, with tabs for **Qualified**, **Invited**, **Completed**, and **Disqualified**. ![The Recruit → Participants table on the Completed tab: finished sessions with play buttons, and the Qualified / Invited / Completed / Disqualified tabs across the top.](/docs/recruit/participants-table.png) ## The Recruit sub-tabs | Tab | What it shows | | ---------------- | ------------------------------------------------------------------------------------------------------------- | | **Scheduling** | Human moderated studies only: the availability participants book from. | | **Audiences** | Your panel orders and share links: status, how many have completed, and drafts you haven't finished. Start new ones here. | | **Participants** | Every individual, across all panels and share links. Invite, disqualify, send reminders, pay. | | **Screener** | Only if you use the classic builder, and never on human moderated studies: the screener editor. In the new builder you edit the screener in Build. | | **Quotas** | Quotas you've defined for this study and how full each one is. | When it's turned on for your organization, the **⋮** menu beside the tabs opens the **Recruitment funnel**: stage-by-stage conversion from clicking the link, to completing the screener, to completing the interview. ## Things to know - **You can mix both paths.** A study can pull half its sample from a panel and the other half from your own list. - **Screener and quotas apply to both.** They're study-wide, not per-source. - **Panel orders have lead time.** Hiring from a panel takes hours to days depending on demand. BYO is instant, as soon as you share the link. ## Related Place an order with Respondent or User Interviews. Share a link or import a list. Balance your sample. See where participants drop off. --- # Pay participants you recruited Source: https://app.strella.io/docs/product/recruit/pay-participants Participants you recruit yourself, with a share link or a CSV import, are paid from the **Pay Research Participants** window once their interview is done. You choose who to pay, set one payment method and one incentive per participant, and Strella applies them to everyone you selected. Panel participants from Respondent or User Interviews are paid differently: their incentive is prepaid with the recruitment order. See [Report a participant](/recruit/report-participant) for how their payout works. ## Pay a group of participants Tick the participants you want to pay, then click **Pay Participants** in the bar at the bottom of the table. The button appears when at least one ticked participant is one you recruited yourself and hasn't been paid yet. Everyone starts ticked. Untick anyone you don't want to pay right now, or use the checkbox in the header to tick or untick everyone at once. Pick a **Payment method** and type the **Incentive per participant**. Every ticked row shows the method and amount that participant will get, and the summary beside the fields shows how many are selected and the total. Click **Confirm**. It stays disabled until there is someone to pay and, for a payment through Strella, an incentive above $0. You can also pay one participant at a time from their interview under **Results**. The same window opens with just that participant. ## Payment methods Each method sets what Confirm does for the ticked participants. | Method | What Confirm does | | ------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------- | | Pay via Strella | Takes you to checkout to pay the total, then Strella sends each participant their incentive at their email address. | | Pay via Strella (Credits) | Pays the total from your organization's credit balance, after a second confirmation that shows the amount. | | No payment / Pay manually | Marks the participants as attended without any payment through Strella. Use it when you pay them yourself or the study offered no reward. | When you choose **Pay via Strella (Credits)**, a summary under the list compares the total with your available credits. If the total is more than your balance, the window shows **Insufficient credits** and Confirm is disabled until you add credits on the Billing page. ## Who gets paid - **Ticked participants** get the method and incentive you chose. - **Unticked participants**: Strella leaves them out entirely. It does not pay them or mark them as attended, and they stay unpaid in the table so you can pay them later. - **Participants with no verified email address** can't receive a payment through Strella, because Strella sends the incentive to their email. Whatever method you pick, Strella only marks a ticked participant without one as attended, and their row says **No verified email address**. - **Panel participants** in the same selection appear under **Hired Participants** as **Prepaid**. Confirming marks them as attended on their panel, which releases their payout on most orders. ## Tips - **Set the amount once.** You don't need to enter the incentive for each person: the one amount applies to everyone ticked. - **Pay in groups if amounts differ.** To pay some participants a different amount, untick them, confirm the first group, then open the window again for the rest. --- # Set quotas Source: https://app.strella.io/docs/product/recruit/quotas A quota is a rule that says "at most N participants who answered X to question Y." Once a quota fills, further matching participants are disqualified (or flagged for review, depending on your auto-disqualify settings). ## How quotas work A quota has three parts: - **Criteria**: which screener answers it tracks. Quotas reference structured screener questions: **Multiple choice** (single-select), **Multiple choice multi-select**, and **Slider** answers. Free-text answers aren't usable as quota criteria. - **Quota limit**: how many matching participants can complete. Once the count reaches the limit, the quota is full. - **Quota behavior**: what happens at the limit: - **Close when full** (default): new matches don't count toward the quota. - **Allow overflow**: let them through, but mark the quota over. Multi-criteria quotas join criteria with **AND** by default, but you can switch any criterion to **OR**. Use OR to match either of two answers (e.g. "5 to 10 employees" expressed as two range criteria joined with OR). ## Steps Click **+ Add quota** to open the quota editor. Use something descriptive, like "PMs at SMB" or "Mobile-first users." Names show up in the funnel breakdown later. Pick a screener question and the answers that match. For a "PMs at SMB" quota: - Question: _What's your role?_ → matches **Product manager** - AND Question: _How big is your company?_ → matches **under 200 employees** Each criterion picks one screener question. To combine multiple, add more criteria. They're joined with AND. Pick a **Quota limit** (e.g. **10**) and a **Quota behavior** (usually **Close when full**). The Quotas tab shows each quota's progress. Click in to see which participants count toward it. ![The Quotas tab: each quota (Drivers, Public Transit Riders) with its completed count and a fill-progress bar.](/docs/recruit/quotas.png) Strella checks quotas automatically as new participants pass the screener. You take no manual action. ## Configure what happens at the limit Two related toggles are in **Settings → Access & scheduling**: - **Auto-disqualify when quota filled**: when a quota fills, further matching participants are auto-disqualified instead of sitting in Qualified awaiting review. - **Auto-disqualify when no quota matches**: participants whose screener answers don't match any active quota are auto-disqualified. Useful when your quotas define the only segments you want. Both default to off. Turn them on if you want quotas to apply automatically. ## Editing quota-referenced screener questions Once a screener question is referenced by a quota, you can't restructure it (add/remove/rename options) without first removing the quota. The one thing you _can_ still change is the **qualify / disqualify** mark on each option, because those aren't part of what the quota tracks. If you need to edit the question, delete the quota first, edit, then recreate the quota. ## Tips - **Don't set quotas before you know your audience.** Run a few interviews first to see the distribution, then add quotas if needed. ## Related The screener questions quotas can reference. The auto-disqualify toggles that drive quota behavior. See quota fills broken out alongside conversion. --- # Report a participant Source: https://app.strella.io/docs/product/recruit/report-participant Panel participants (from Respondent or User Interviews) are paid automatically once their interview completes, so there's no payout button to click. If someone turns out to be fraudulent or low quality, you have a short window to **report** them, which stops the payout and flags them with the panel. ## How auto-payment works When a panel participant finishes their interview, Strella schedules their incentive payout automatically. Strella does not pay right away. It waits out a review window first, which gives you time to check the interview and stop payment if something is wrong. If you do nothing, Strella pays the participant at the end of the window. Reporting them before the window closes cancels that scheduled payout. (Participants you bring in yourself with a share link aren't paid through this flow.) ## The review window You have **3 business days** from when the interview completes to report a participant. The window counts business days, so **weekends don't count**: an interview that finishes on Friday can be reported through the following Wednesday. Once the window closes, Strella pays the participant and you can no longer report them. In the Participants table the **Report participant** action is disabled after payout, with the tooltip _"Participant has already been paid out"_. While the window is open, the tooltip shows the deadline date. ## Report a fraudulent participant Report a participant from the **Completed** tab of the Participants table, which lists finished interviews. Find the participant whose interview you want to review. Click the row to open the detail panel, or use the row's action menu. In the row's action menu (or the detail panel's menu), choose **Report participant**. The action is available only while the participant is inside the 3 business day window. Select a **Reason for reporting**. For a fraudulent participant, choose **Suspected fraud or misrepresentation**, the only reason that files a permanent report against their panel account. The other reasons (low-quality responses, didn't follow instructions, and so on) remove the participant from your study without a panel-level report. Add specifics in **Additional details**, tick the confirmation box, and click **Report & Delete Interview**. ## What happens when you report Reporting is final. When you submit a report: - **The interview recording is permanently deleted.** - **The participant isn't paid** for the interview, and Strella cancels the scheduled auto-payment. - **The recruitment slot is freed on the panel**, so it can be filled by someone else. - **Strella's support team investigates** the report. - For **Suspected fraud or misrepresentation**, a permanent strike is filed against the participant's panel account. ## Reject vs report - **Reject** happens before the interview, on the Qualified tab. It disqualifies a candidate who doesn't fit. No interview has happened and no payment is involved. See [Approve, reject & invite](/recruit/approve-reject-invite). - **Report** happens after the interview, on the Completed tab. Use it when someone has already completed an interview and you need to stop their payout and flag fraud. ## Tips - **Watch for the reminder email.** The business day before a participant's payout is due, Strella emails you so you can review and report anyone who needs it before end of business. - **Report as soon as you spot a problem.** The window is only 3 business days. Once it closes, the payout can't be reversed. - **Reserve "Suspected fraud or misrepresentation" for genuine fraud.** It files a permanent strike on the participant's panel account. For an interview that was simply weak, pick a quality reason instead. ## Related Triage candidates before the interview. Where panel payouts and incentives are set. --- # Schedule moderated interviews Source: https://app.strella.io/docs/product/recruit/schedule-moderated-interviews In a human moderated study you and your teammates run each interview yourselves, so participants book a time from your availability instead of joining whenever they like. Set who runs the study and when each of you is free on the **Scheduling** tab. The study can't go live until at least one time is open to book. ## Where Scheduling lives Create the project with **New project**, then choose **Human-moderated project** in the **Create a project** dialog. An AI assistant connected to Strella, or Strella's API, can also create a human moderated project and write its screener. Your availability and publishing still happen here, on **Scheduling** and **Build**. For a human moderated study, **Recruit** opens on **Scheduling**, and the tabs follow the order you set a study up in: **Scheduling**, **Audiences**, **Participants**, **Quotas**. ## Set up your schedule At the top of the panel on the right, **Researchers** lists everyone who runs this study's interviews. The list starts with you. Click **Add researcher** to add a teammate. Under **Booking mode**, pick **All together** or **Round robin**. See [Run interviews with teammates](#run-interviews-with-teammates). The rest of the panel holds the rules every bookable time follows: timezone, session length, notice and more. Each setting is described in the table below. On the calendar, drag down a day from the time you become free to the time you stop. The **Add availability** form opens with those times. Under **Researcher**, pick whose availability it is (you, to start). Click **Save**, and the block appears in green, labelled with that researcher's first name. A block participants can't book shows in yellow, marked **Unavailable** with the reason. For example, a block sooner than a 24-hour **Minimum notice** reads **Requires 24-hour notice**. Hover over the reason to see how to fix it. You can also click **Add availability**, or a day's header, and pick the researcher, the day and the **From** and **To** times yourself. Anyone on your team can add availability for any researcher on the study. Your first block also saves the settings on the right, so you never have to save them separately before adding availability. Drag a block to another time or day, or drag its top or bottom edge to change when it starts or ends. Strella saves the change when you let go and rounds dragged times to the nearest quarter hour. If a block overlaps another block of the same researcher, Strella joins the two into one block. A block keeps its researcher. To give the time to someone else, delete the block and add one for that researcher. Click a block to change its times in the form, or to delete it. On a phone, tap a day to add availability and tap a block to edit it. Go back to **Build** and click **Publish**. If no time is open to book yet, Strella shows **No times to book yet** with a **Go to Scheduling** button that takes you straight back here. ## Schedule settings Strella turns each availability block into bookable start times using these settings. Change any of them and click **Save schedule settings**. | Setting | What it controls | | ---------------------------------------- | -------------------------------------------------------------------------------------------------------------- | | **Timezone** | Your availability is in this timezone. Participants see times in their own. | | **Interview duration** | How long each interview is. | | **Allow participants to reschedule** | Participants can pick a new time themselves. When off, use **Cancel and re-invite**. | | **Minimum notice** | How far ahead participants must book. | | **Maximum interviews** | Most interviews per day. Leave it empty for no limit. | | **Buffer time** | Break after each interview. | | **Start time intervals** | How often interviews can start: every 15, 30 or 60 minutes. | | **Researchers** | Who runs the interviews. See [Run interviews with teammates](#run-interviews-with-teammates). | | **Booking mode** | **All together**: all researchers need to be available. **Round robin**: any researcher can be available. | ## Run interviews with teammates Add teammates from your organization under **Researchers**, then click **Save schedule settings**. Everyone's availability and busy times show on the calendar under their first name. - **To move a researcher to the top, click Make organizer on their row.** In **All together**, the first researcher is the **Organizer**. Invites come from the Organizer, and Strella asks the Organizer whether the participant showed up. - **All together** (the default): all researchers need to be available, and everyone joins. - **Round robin**: any researcher can be available, and one of them hosts. Strella spreads interviews across the team. Participants don't see which researcher hosts. - **Each researcher's own calendar counts.** Strella doesn't offer a time when that researcher is busy on their connected calendar or already in an interview on any study. - **Everyone in an interview gets their own invite**, with a link that joins as a researcher, not as a participant. Share that link, never the participant's. - **Removing a researcher** deletes their availability on this study, so Strella stops offering their time. The researcher stays in interviews already booked. - **Everyone joins the same call** and can speak, share their screen, and send chat messages. Any researcher can end the interview for everyone. - **A researcher can leave without ending it.** While another researcher is still in the call, ending the call offers **Leave call**, which takes you out and leaves the interview running. You can rejoin from the same page. - **Each researcher keeps their own name in the transcript.** Strella notes when each researcher was talking during the call and credits each interviewer voice to the researcher who was speaking. A voice it can't match with confidence shows as **Interviewer** instead, so a teammate's name never lands on someone else's questions. ## Connect your calendar Connect Google Calendar or Outlook once. Strella never offers participants a time when you're busy there, on any study where you're a researcher. 1. Go to **Settings › Integrations** and click **Connect** beside **Google Calendar** or **Outlook**. You can also click **Connect calendar** on your own row under **Researchers** on a study's Scheduling tab. 2. Sign in and allow access. You come back to the page you started from. Strella reads only when you're busy, never what an event is or who is in it. Your busy times show on each study's calendar over your availability, and participants can't book them. Events you've marked as free don't block a time. Teammates on the same study see your busy times, never what they are. If your calendar stops accepting the connection, your row and **Settings › Integrations** show **Reconnect**. Strella doesn't check your busy times until you reconnect. To stop using a calendar, click **Disconnect** in **Settings › Integrations** or on your row. **Disconnect** removes the calendar from every study where you're a researcher, not just this one. Without a connected calendar, leave busy times out of your availability, or a participant can book a slot you already have taken. With a work Microsoft account, your organization's IT admin may need to approve Strella before you can connect Outlook. ## How participants book A participant from a panel order or a share link takes your screener first. Once they qualify, they pick a time from your open slots, and a calendar invitation goes to their email address. Participants can join from 5 minutes before their booked time. If they open their link earlier, Strella shows their booking with a countdown, then opens the call for them when it reaches zero. You can join the call at any time. Participants who join through a share link always enter their name and email, because a booking needs somewhere to send the invitation. A time counts as open only when it falls inside an availability block, respects your minimum notice, buffer, and daily cap, and isn't already booked. An availability block that sits entirely inside your minimum notice leaves nothing to book. ## Manage booked interviews Booked interviews appear on the Scheduling calendar in red with the participant's name. Participants always pick their own interview time, so a booked interview can't be dragged. Click one to see who is in it and act on it: - **Researchers** lists everyone in the interview, the host marked. - **Join interview** opens the call in a new tab. - **Change host** makes another researcher the host. - In **Round robin** you can pick anyone on the study. The new host gets the calendar invite and the reminders, and Strella cancels the interview on the old host's calendar. Participants can then book the old host at that time again, so take it off their availability if they're not free. - In **All together** you pick someone already in the interview, and everyone stays in it. - Strella doesn't tell the participant, and won't pick someone who is in another interview or busy on their calendar at that time. - **Mark as attended** or **Mark as no-show** records attendance once the session has started. - **Cancel and re-invite** cancels the session and asks the participant to pick a new time from your availability. A participant from Respondent gets the invitation from Respondent. Anyone else gets an email with a link to book again. - **Cancel interview** cancels it, cancels the participant's invite, and frees the time for someone else. Their booking link no longer lets them book. To let them pick another time, use **Cancel and re-invite**. - **Remove participant** opens the [report flow](/recruit/report-participant), which removes them from the study. It shows once their booked interview time has passed. Before then, cancel the interview instead. Both cancel options ask you to confirm first, and say what the participant will get. ### Cancelling a Respondent interview Respondent's own rules set whether a participant recruited through Respondent is paid for a cancelled interview: - **4 or more hours before the start:** cancelling means they aren't paid. - **Less than 4 hours before the start, or after it:** cancelling still pays them in full, as if the interview had happened. If they didn't show up, use **Mark as no-show** instead, which isn't paid. - **Cancel and re-invite** isn't available less than 4 hours before the start, because Respondent doesn't allow it. - **A late cancel uses up that participant's place.** Because Respondent has paid them, Strella marks them as paid, they count toward your study's participant limit, and closing the order doesn't refund their place. They don't appear as an interview in Results. In **Recruit** › **Participants** they're listed under **All** as **Paid, cancelled late**. --- # Manage share links & redirects Source: https://app.strella.io/docs/product/recruit/share-links A share link is the URL participants click to enter your study. You can have several, which is useful for splitting your sample by source (customers vs prospects), or for sending different audiences to different post-interview destinations. Each link carries its own settings (whether to run the screener, collect an email or name, auto-invite, send reminders, and cap participants) and its own redirect destinations. A link inherits the study default for any setting you leave untouched, so you only change what's different for that audience. ## Steps Every link you've created is listed there, alongside any panel orders, each as its own row. Click **Create link** on the **Recruit with a link** card, then give it a name (for your internal reference; participants never see it) and, if you want a custom URL, a slug. Strella generates the rest of the link. Open the link row's **⋯** menu and choose **Manage share link**, then **Link settings**. Each setting uses the study's value until you change it here, and a change applies to this link only: - **Limit number of participants**: stop this link accepting new responses once it reaches its own limit. A link with its own limit closes itself when that limit is met; a link without one keeps taking responses until the study-wide participant limit closes the whole study. - **Require screener**: ask people who use this link to complete the screener first. - **Collect email addresses** and **Collect participant names**: ask arrivals for their email or first and last name. Without names, Strella numbers each participant (Participant #1, Participant #2) so you can tell them apart in results. Participants never see this number. - **First name only**: shown when name collection is on. Ask participants for just their first name: the registration form drops the last-name box, and a `last_name` sent through [URL parameters](/recruit/url-parameters) is ignored for this link. - **Auto-invite qualified participants**: invite qualifying participants automatically instead of waiting for your review. Works with or without email collection, since participants can be invited anonymously. - **Send reminder emails**: remind invited participants to complete the study. Requires **Collect email addresses**, since the reminder is emailed. To stop just this link without closing the whole study, use the status pill on the row itself (next to the **⋯** menu) and choose **Disable link**, rather than the Link settings panel. From the same **Manage share link** menu, choose **Redirects** to override the study's default redirects for this link. The redirect URLs you can customize per link: - **Call finished**: where qualified, completed participants go. - **Screenout**: where disqualified participants go. - **Quota full**: where over-quota participants go. - **Dropout**: where mid-call quitters go. Enter the full URL including `https://`, or leave a field blank to inherit the study default. Useful when a "customer" link sends people back to your dashboard, but a "prospect" link sends them to a sales page. Click the copy icon next to the link. Send it in email, embed it on your site, share it in chat, anywhere your audience is. Each share link tracks separately in the **Recruitment funnel** (the **⋮** menu beside the Recruit tabs), which is useful for measuring which sources convert best. ## Tips - **One link per source.** A link per acquisition channel makes funnel analysis much cleaner, because you can see at a glance which sources delivered. - **A link inherits the study default until you change it.** Leave a setting untouched and the link follows the study; change it and only that link is affected. - **Reminder emails need email collection.** Reminders are emailed, so turning off **Collect email addresses** turns them off. Auto-invite is unaffected: participants can be invited anonymously, so it works with email collection off. - **Tag arrivals with [URL parameters](/recruit/url-parameters).** Add `?key=value` to the end of any link to record where someone came from, pre-fill their details, set their language, or pass your own IDs straight through. - **Deleting a link doesn't delete participants** who came through it. The link just stops working for new arrivals. - **Closing the study stops every link.** Choosing **Disable link** on a single link's status pill stops only that link and leaves the rest of the study open. - **A closed or full study shows an error page right away.** Anyone who opens a link to a closed or full study, or a disabled link, sees the error page before any form appears. The page shows your custom closed message if you wrote one. - **Strella doesn't verify who a participant is.** A link works for anyone who has it, and an email or match key identifies an attempt rather than proving the person's identity. Someone who enters the email from an earlier attempt can continue that attempt, so share a link only with the audience you intend. ## Related The path that uses share links. Logo, colors, custom messages that share-link participants see. Track conversion per link. Tag participants and pass IDs through the link. --- # Pass data with URL parameters Source: https://app.strella.io/docs/product/recruit/url-parameters A share link can carry extra information about whoever clicks it. Add parameters to the end of the link's URL and Strella captures each one as a **custom field** on that participant. It is a tidy way to record where someone came from, who they are in your own systems, or which campaign sent them. Parameters connect Strella to your other tools. Tag a link by channel, pass a customer or panelist ID to the interview, and send those same values back out when the participant finishes. ## How it works Take any share link and add parameters to the end. Start with a `?`, write each one as `name=value`, and join additional ones with `&`: ``` https://app.strella.io/onboard/your-link-id?source=newsletter&campaign=spring_2026&external_id=12345 ``` Every parameter you add becomes a custom field on the participant who clicks it. Each one shows up automatically as its own column in **Recruit → Participants**, named after the parameter, with no setup required. The example above gives every participant from that link a **source** of `newsletter`, a **campaign** of `spring_2026`, and an **external_id** of `12345`. Strella captures the values the moment someone clicks, keeps them through the screener and registration, and saves them on the participant. You can **filter, sort, and tag** on them like any other column, and they appear on each participant's profile. Strella ignores case in parameter names, so `Source` and `source` fill the same column. Pick one spelling per field and use it every time. ## Reserved parameters A few names have a built-in meaning. Use them to set the participant's identity and language instead of creating a custom field: | Parameter | What it does | | -------------------------- | ----------------------------------------------------------------------------------------------- | | `first_name`, `last_name` | Set the participant's name and pre-fill (and lock) the name boxes on the registration form. | | `email` | Sets the participant's email and pre-fills (and locks) the email box on the registration form. | | `lang` | Sets the language for the screener and interview. Use a language code: `es`, `fr`, `de`, or `zh`. | A fully personalized link looks like this: ``` https://app.strella.io/onboard/your-link-id?first_name=Sam&last_name=Rivera&email=sam@acme.com&lang=es ``` Because these names are reserved, don't use them for your own tracking. A parameter called `email` always fills the email box. It does not create a separate custom field column. A few details on `lang`: - It only works for [supported languages](/reference/languages), and the screener and interview must have that language available. - If a study has language selection turned off, the link's `lang` is ignored and everyone gets the study's default language. - When `lang` is absent, participants get the study's default language (English if none is set). ## Match the same person across visits By default, every click creates a new participant. If you're passing an ID you already own (a CRM contact ID, a panelist ID), you can tell Strella to treat that field as the person's identity, so a repeat visit **updates the existing participant instead of creating a duplicate**. In **Recruit → Participants**, open **Manage columns** and add or edit a text custom field (for example, `external_id`). Turn on the **Use as participant identifier** toggle. From then on, Strella matches anyone who arrives through a link carrying that field to the existing participant with the same value, instead of adding a new row. This also stitches together people you've already imported. Upload a CSV with an `external_id` column, mark that field as the identifier, then send links carrying each person's `external_id`. Their click attaches to the row you imported instead of duplicating it. You can't turn on the identifier toggle while duplicate values already exist for that field. Resolve them first: select the duplicates in the table and use **Merge Duplicates** from the bulk menu, or delete the extras, then enable it. ## Send the data back out Strella can send the values you pass in to your other tools. - **Redirect URLs.** When a participant finishes, screens out, hits a full quota, or drops out, Strella appends their custom fields to the [redirect URL](/recruit/share-links) as parameters. That lets you hand an `external_id` (or any other field) back to your panel or downstream tool so both sides stay in sync. Language parameters are not forwarded. - **Their call link.** Each participant's unique interview link carries their custom fields too, so **Copy call link** gives you a URL with everything already filled in, ready to paste into an email or message. ## Limits & good practice - **One value per parameter.** Repeating the same name in a link (`?color=red&color=blue`) is ignored. Use a single value per field. - **200 characters max** per value. - **Keep it clean.** For security, Strella rejects values containing semicolons or script-like content, and the participant cannot enter. Use plain identifiers, names, and labels. - **Parameters are visible in the URL.** Don't pass anything secret or sensitive that you wouldn't want the participant (or their browser history) to see. - **Use one link per source.** Parameters tag *who* arrived. Separate [share links](/recruit/share-links) keep the [funnel](/recruit/funnel) counts separate for each channel. Use both together. ## Related Create links per source and set where participants land afterward. Share links and CSV imports feed the same Participants table. See how each source converts. What to do once tagged participants arrive. --- # Browser & device requirements Source: https://app.strella.io/docs/product/reference/browser-device-requirements ## Participants (taking an interview) | Requirement | Detail | | -------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Browser** | Chrome, Edge, Safari, or Firefox on a computer. On a phone, the phone's browser, unless the study needs Strella's mobile app (see **Device**). | | **Device** | Desktop or laptop is supported everywhere. A phone (iOS or Android) can join when **Allowed devices** in [Interview setup](/configure/recording-modes) includes mobile. An AI-moderated study that records the screen runs on Strella's mobile apps. Every other study, a human-moderated one included, runs in the phone's browser. | | **Internet** | A stable connection. The AI tolerates short hiccups but extended disconnection ends the call. | | **Microphone** | Required. Built-in laptop mics work; external mics are better. | | **Webcam** | Required only if **Webcam recording** is on for the study. | | **Screen share** | Required only if **Screen recording** is on. Mobile screen share requires Strella's iOS or Android app. A human-moderated interview never asks the participant to share their screen. | | **Figma prototype access** | The prototype must be **published** in Figma; participants don't need a Figma account to view it. | ## Researchers (authoring + reviewing) | Requirement | Detail | | --------------------- | --------------------------------------------------------------------------------------------------- | | **Browser** | Any modern desktop browser. | | **Device** | Desktop or laptop recommended. Tablet works for review; authoring on phone is possible but cramped. | | **Internet** | Stable broadband. Video playback at full quality needs reasonable bandwidth. | | **Figma integration** | Required to use Figma prototype steps. See [Connect Figma](/account/figma). | ## What happens when participants deny permissions Permission prompts are the most common reason a participant gets stuck before the interview starts. The call requires microphone access, plus camera access or screen share if your settings need them. For screen share, you control which surface participants are prompted to share first. **Settings → Interview setup → Default screen share option** offers **Window** or **Entire screen**. Choose Window when a task only needs one app, or Entire screen when participants move across apps. See [Recording settings](/reference/recording-modes). | Denied permission | What the participant sees | | -------------------------------------- | ----------------------------------------------------------------------------------------- | | **Microphone** | The call can't start. The participant must grant microphone access or leave the call. | | **Camera** (when webcam is required) | Same as the microphone: the participant can't proceed without granting it. | | **Screen share** (when required) | The participant is asked to retry. Repeated denial blocks the interview from starting. | | **Screen share dismissed** (no choice) | Most browsers treat dismissal the same as denial. The retry prompt explains how to allow screen share. | If participants drop out at this step, the most common fixes are: - **Browser-level permissions blocked sitewide**: they need to clear the site permissions for the share-link domain and reload. - **Working from a restricted device**: corporate laptops sometimes block screen share at the OS level; ask them to try a personal device. - **Mobile browser instead of the app**: on an AI-moderated study that allows mobile and records the screen, direct them to install the Strella iOS or Android app. A human-moderated interview runs in the phone's browser and needs no app. ## Related The toggles that set which permissions Strella asks for. What participants see before the permission prompts. --- # Supported languages Source: https://app.strella.io/docs/product/reference/languages Strella supports interviews in about 50 languages. Pick one for the whole study, several to let participants choose, or leave the list empty to allow all. ## Languages | Code | Language | | ------- | --------------------- | | `en` | English | | `ar` | Arabic | | `be` | Belarusian | | `bs` | Bosnian | | `bg` | Bulgarian | | `ca` | Catalan | | `zh` | Chinese (Simplified) | | `zh-TW` | Chinese (Traditional) | | `hr` | Croatian | | `cs` | Czech | | `da` | Danish | | `nl` | Dutch | | `et` | Estonian | | `fi` | Finnish | | `fr` | French | | `de` | German | | `el` | Greek | | `he` | Hebrew | | `hi` | Hindi | | `hu` | Hungarian | | `id` | Indonesian | | `it` | Italian | | `ja` | Japanese | | `kn` | Kannada | | `ko` | Korean | | `lv` | Latvian | | `lt` | Lithuanian | | `mk` | Macedonian | | `ms` | Malay | | `mr` | Marathi | | `no` | Norwegian | | `fa` | Persian | | `pl` | Polish | | `pt` | Portuguese | | `pt-BR` | Portuguese (Brazil) | | `ro` | Romanian | | `ru` | Russian | | `sr` | Serbian | | `sk` | Slovak | | `sl` | Slovenian | | `es` | Spanish | | `es-CO` | Spanish (Colombia) | | `sv` | Swedish | | `tl` | Tagalog | | `ta` | Tamil | | `th` | Thai | | `tr` | Turkish | | `uk` | Ukrainian | | `ur` | Urdu | | `vi` | Vietnamese | ## Things to know - **Voice availability varies by language.** When you select a single non-English language, the voice picker filters to voices that speak it. For multi-language studies, you pick a gender and Strella selects the right voice per language automatically. - **The participant interface also appears in the selected language.** The screener and call screens show in that language. - **Regional dialects use the base language.** Transcripts recognize regional speech such as Quebec French or Mexican Spanish when you pick French or Spanish. The table lists only Brazilian Portuguese and Colombian Spanish separately. Pick Spanish (Colombia) for a moderator with a Colombian voice and a participant interface written in Colombian Spanish. ## Related Configure the language for your study. Voices grouped by provider. --- # Recording settings Source: https://app.strella.io/docs/product/reference/recording-modes Recording behavior comes from independent toggles in **Settings → Interview setup**. Combine them to fit your study. | Setting | What it controls | | --------------------------------- | ------------------------------------------------------------------------------------------ | | **Allowed devices** | Desktop and mobile, desktop only, or mobile only. | | **Webcam recording** | Required to join / Off. | | **Blur participant camera** | Replaces the participant's video feed with a blurred version (audio unaffected). | | **Screen recording** | Enabled / Disabled. | | **Allow pausing screen share** | Participant can hide sensitive screens mid-call. | | **Default screen share option** | What the participant sees first in the screen-share picker (Window or Entire screen). | | **Screen share verification** | Require a QR-code check to make sure they're sharing the right screen. | | **Blur participant screen share** | Records a blurred version of the shared screen. | | **Default captions on** | Show live captions during the call. | | **Push to talk** | The Participant holds a button to speak, instead of Strella detecting when they start talking. | | **Session timeout** | End the call automatically after a period of participant inactivity. | | **Name anonymization** | Replaces participant names with anonymous IDs throughout the study. Hides emails and social links everywhere, including AI assistants and the public API. | | **Voice anonymization** | Mask voice in recordings (where supported). | ## Related How to set these toggles for a specific study. What participants need on their end. --- # Screener question types Source: https://app.strella.io/docs/product/reference/screener-question-types The screener supports five question types. Pick a row when you click **+ Add element** in the screener region of [Build](/build/build-screen). | Type | What it does | Usable as quota criteria? | | ------------------- | -------------------------------------------------------------------------------- | :-----------------------: | | **Multiple choice** | Single- or multi-select from a list. Mark each option to qualify or disqualify. | Yes | | **Short text** | One-line written answer (e.g. job title). | No | | **Long text** | A paragraph answer (e.g. "describe your role"). | No | | **Number slider** | A numeric value on a sliding scale. | Yes | | **Legal agreement** | An NDA or consent the participant must accept to continue. | No | ## How qualification works On a **single-select** multiple choice, mark each option **Qualify** or **Disqualify**: - **Qualify**: picking this answer lets the participant continue. - **Disqualify**: picking this answer screens the participant out. On a **multi-select** multiple choice, each option is **Must select**, **May select**, or **Disqualify**. Participants must tick every "Must select" option to qualify, "May select" options don't affect qualification, and ticking a "Disqualify" option screens them out. Disqualified participants still finish the rest of the screener before being sent to the goodbye page, so they can't tell which question screened them out. Use skip logic with destination **End of screener** to override this for a specific question. For other question types, qualification logic isn't part of the question itself. Use quotas if you need to filter on the answer. ## Related How to add and configure these questions. Use multiple-choice and slider answers to balance your sample. --- # Participant statuses & funnel stages Source: https://app.strella.io/docs/product/reference/statuses-and-funnel Strella has two related views of participant state: **statuses** (in the Participants table) and **funnel stages** (in the funnel chart). ## Participants table tabs | Tab | Meaning | | ----------------- | --------------------------------------------------------------------------- | | **Qualified** | Passed the screener and any quota checks. Hasn't been invited yet. | | **Invited** | Received the invite link to join. Hasn't completed the interview yet. | | **Completed** | Finished the interview. The recording is in Results. | | **Disqualified** | Failed the screener, was rejected manually, or got disqualified by a quota. | ## Funnel stages The funnel reads top to bottom; each stage is a subset of the one above. Some stages are skipped depending on study config. See [Watch the funnel](/recruit/funnel) for the variations. | Stage | Meaning | | ---------------------- | ---------------------------------------------------------------------------------- | | **Clicked link** | Everyone who opened your share link. | | **Started screener** | Answered at least one screener question. (Skipped if the study has no screener.) | | **Completed screener** | Answered every screener question. (Skipped if no screener.) | | **Qualified** | Passed the screener and any active quotas. | | **Registered** | Provided contact details. (Skipped if you don't collect name/email.) | | **Invited** | Received the invite to join the interview. | | **Opened invite** | Opened the invite link. | | **Started interview** | Joined the call. | | **Completed interview**| Finished the interview and produced a recording. | An interview counts as **Completed** once the participant reaches the final question. If they leave at that point without submitting an answer to the last question, the interview still shows as completed. Reaching the end is what marks it done, not answering every question. ## Related The funnel page itself. Where you act on participant statuses. --- # Step types Source: https://app.strella.io/docs/product/reference/step-types When you click **+ Add element** in [Build](/build/build-screen), the menu groups step types into five categories. Each row below corresponds to one entry in that menu. ## Ask Steps where the AI moderator talks with the participant. | Type | What it does | | ----------------- | --------------------------------------------------------------------------------- | | **Open question** | A free-form question the AI asks. Supports dynamic follow-ups and verbatim mode. | | **Task** | A task the participant performs while the AI watches. Pairs with media. | | **Statement** | The AI says something and expects no response. Useful for transitions and context. The AI speaks it aloud and never shows it on screen. To display words, use **Text** under Usability. | ## Survey Structured-response steps. The AI pauses, and the participant taps or drags their answer. | Type | What it does | | ------------------- | -------------------------------------------------------- | | **Multiple choice** | Single-select from a list of options. | | **Ranking** | Drag options into preferred order. | | **Open text** | Short typed answer. | | **Slider** | Number on a sliding scale (min, max, step configurable). | | **Rating scale** | Likert, face, or star scale (where enabled). | ## Structure Steps that affect the shape of the interview, not its content. | Type | What it does | | ----------------- | -------------------------------------------------------------------------------------------------------------------------------------------------- | | **Section** | A container for grouping steps. Can be conditional (Strella shows it only to some participants) based on screener answers, quotas, or in-call survey answers. | | **Randomization** | A split into branches. Each participant takes one branch at random. See [Randomize sections](/build/randomization). | ## Usability Steps that show the participant something. | Type | What it does | | ------------- | ----------------------------------------------------------------------------------------------------------------------------------------- | | **Prototype** | Embeds a Figma prototype. Supports free-explore or goal-based tasks and captures heatmaps. Requires the [Figma integration](/account/figma). | | **Website** | Opens a URL in a new tab. Pair with screen share to see what the participant does. | | **Image** | Shows an image inside the call window. | | **Video** | Plays a video inside the call window. | | **Embed** | Embeds an iframe inside the call window. For tools and demos that support embedding. | | **Text** | Shows written copy inside the call window for the participant to read. You type the words, so there is nothing to upload. | ## Test Interactive exercises the participant works through, then submits. A Test element can hold its own follow-up questions, which stay available while it's on screen. | Type | What it does | | ------------- | ---------------------------------------------------------------------------------------------------------------------------------- | | **Card sort** | Participants drag a set of cards into groups, revealing how they organize your ideas. Open, closed, or hybrid grouping. See [Add a card sort](/build/card-sort). | | **Tree test** | Participants click through an outline of your labels to find where a task would be done, revealing whether your navigation leads them to the right place. See [Add a tree test](/build/tree-test). | ## Related How to add and arrange these steps. The reference for the other half of Build. --- # Voice options Source: https://app.strella.io/docs/product/reference/voice-options The voice picker in **Settings → AI moderator** groups voices by provider. Use this reference to choose a provider for your needs. | Provider | Strengths | Speed range | | -------------- | ----------------------------------------------------- | ----------- | | **OpenAI** | Widest language support, consistent quality. | 0.5× to 2.0× | | **Inworld** | Expressive voices, limited language coverage. | 0.5× to 1.5× | | **ElevenLabs** | Premium quality, good multilingual support. | 0.7× to 1.2× | | **Cartesia** | Fast, natural-sounding voices, wide language support. | 0.5× to 2.0× | ## How the voice picker works The picker filters to voices appropriate for your study's language settings: - **One English study**: every English voice across all providers, browsable by name. - **One non-English study**: voices filter to ones that speak that language. - **Multiple languages or all-languages**: pick a gender (Feminine, Masculine, or Neutral); Strella selects the right voice per language at interview time. You can preview any voice with the play icon. Strella reads a sample line so you can hear the timbre and pacing before committing. ## Things to know - **Participants can adjust playback speed themselves** during the call, within the provider's range. Your setting is the default. - **Spanish (Colombia) has its own Inworld voices.** Pick `es-CO` and Inworld offers only Mateo (masculine) and Sofia (feminine). Studies in plain Spanish keep Inworld's other Spanish voices and don't get these two. OpenAI voices work for Spanish (Colombia) too, but Google, ElevenLabs and Cartesia voices don't. - **Inworld voices won't appear for languages they don't support.** The picker hides them; you don't have to figure out compatibility manually. - **Some voices are gated** behind feature access. If you don't see ElevenLabs voices, your organization may not have them enabled yet. Contact your account team. ## Related How to pick a voice for your study. The full language list. --- # Edit a study after launch Source: https://app.strella.io/docs/product/run/edit-after-launch You don't have to get everything right before publishing. Almost every field (screener, question guide, media, AI moderator settings, branding, quotas) is still editable once the study is live. ## When this matters Mid-study edits are useful when: - Your first interviews reveal a question that's confusing or leading. - You realize you forgot to ask about something important. - Recruitment is going slower than expected and you need to loosen the screener. ## How edits behave - **Edits only affect future participants.** Anyone who has already started or finished an interview keeps the version they took. Strella does not move in-flight participants to the new version. - **Version history keeps a record of your edits.** Build saves every published version and takes periodic backups as you work, so you can look back at an earlier guide, preview it, and restore it. See [Version history](/build/version-history). - **Analysis can mix versions without a warning.** If you remove a question after 10 interviews and run 10 more, the analysis shows the question with answers from only the first half. Use the participant filter in [Analysis](/analyze/browse-analysis) to filter by date if you need a clean cohort. ## Closing the study You can't unpublish a live study. To stop recruitment, **close** it instead: - **Manually, from Settings**: **Settings → Access & scheduling → Close project to new participants**. Existing in-flight participants keep going; new arrivals are blocked. - **Manually, from Share**: the **Share** dialog has the same **Close study to new participants** toggle, if that's where you already are. - **Manually, from the dashboard**: click the study's status on its project-list row and choose **Close project** (or **Reopen project**), without opening the study. - **Scheduled**: set a close date or a participant limit on the same Settings tab; the study closes automatically when the condition is met. Closing is reversible: flip the toggle back to reopen, provided your scheduled close date hasn't passed and you're under the participant limit. When a schedule or participant limit is what's holding a study closed, the dashboard's **Reopen project** option is disabled and names the reason, since reopening there only clears a manual close. ## Related What "live" means. Where to set Close project, scheduled close, and participant limit. Where the impact of mid-study edits shows up. See, preview, and restore earlier versions of your guide. --- # Publish your study Source: https://app.strella.io/docs/product/run/publish A study is a draft until you publish. Drafts are invisible to participants: your share link doesn't work, and panel orders can't be placed. Publishing makes it live. ## Before you start You'll need: - A question guide with at least one step. - The settings you care about saved (AI moderator, recording modes, branding). A screener is optional. Add one only if you want to qualify participants before the interview. ## Steps Build validates your screener and question guide in real time. Anything blocking (missing question text, a broken skip-logic destination, an empty required field) is highlighted in red where it appears. Errors must be fixed before you can preview or publish. Top-right of [Build](/build/build-screen). The publish dialog summarizes what's about to go live: screener, question guide, settings. Confirm to publish. ![Publishing a study: clicking Publish in Build and confirming in the publish dialog.](/docs/run/publish-study.gif) [Share your link](/recruit/bring-your-own) or [hire from a panel](/recruit/hire-from-panel). Both paths require a published study. The study status in the header changes to **Live**. Participants can join. ## Tips - **Always [preview](/build/preview) before publishing.** - **Publish during your working hours** if you're using a panel. Participants tend to arrive in the first hour or two after the order is submitted, so having you available to triage helps. - **Already-in-flight participants are unaffected by future edits.** They finish on the version they started. ## Related Do this before publishing. What's safe to change once you're live. --- # Create & share a highlight reel Source: https://app.strella.io/docs/product/share/highlight-reels ## How reels work Every theme in [Analysis](/analyze/browse-analysis) has an auto-generated reel sitting alongside it. Its clips are the highlights that make up the theme. You don't have to "create" a reel; it's already there. **Saving a reel** pins it to **Results → Saved Reels** so it lives in a stable place and stops auto-refreshing as new interviews come in, which is useful when you want to send a reel to a stakeholder and keep it fixed. **Sharing and saving are separate.** Send a reel from the AI chat the moment it's built, without saving it anywhere. Saving is what puts it on the Saved Reels page so you can come back to it. **A reel can pull clips from more than one project.** Strella files it in a single project, but its clips can come from interviews in other projects, most often when the AI chat or the Strella connector builds a reel across your studies. Each clip from another project plays in sequence with the rest and carries a chip naming its project. The reel's clip count and shared link include that clip like any other. ## Steps Three paths: - **From Analysis**: open a theme in [Analyze → by objective or by question](/analyze/browse-analysis) and play the reel attached to it. - **From the AI chat**: ask [Strella's AI chat](/analyze/ai-chat) to "build a highlight reel of the strongest moments about X." It creates a reel directly. - **From the Strella connector**: ask an AI assistant connected through the [Strella connector](/docs/product/mcp/tools#strella_save_highlight_reel) to save clips it found into a project. It lands as a Saved Reel you can play, edit, and share here. Click **Save** on a reel you want to keep. It appears in **Results → Saved Reels** and won't auto-refresh as you add more interviews. Skip this step if you only want to send the reel to someone. Reels are editable: easiest on the **Saved Reels** page, though you can also edit from the AI chat. You can: - **Trim a clip**: drag the handles on a clip to extend or shorten the quote, or reposition it. - **Reorder clips**: drag to change the sequence. - **Remove clips**: drop ones that aren't representative. - **Add clips**: pull in more highlights. ![Reordering clips in the reel editor: dragging clips into a new sequence.](/docs/share/reel-reorder.gif) ![Trimming a clip in the reel editor: dragging the handles to shorten the quote.](/docs/share/reel-trim.gif) Each clip is built from a **highlight**. To control whether the moderator's question plays before the participant's response, toggle that on the highlight, and the reel rebuilds from it. From the reel: - **Share**: generates a public link anyone can play without a Strella account. - **Download**: Strella builds the reel as an MP4 in the background, cut exactly where the player cuts it. A message shows its progress on every page. When the MP4 is ready, the message offers a **Download** button, and Strella emails you the link too. The MP4 carries the interview captions as subtitle tracks: English, plus the interview's own language when it was not held in English. The tracks are off until you pick one in your video player (QuickTime: View, then Subtitles; VLC: Subtitle, then Sub Track). **Download all reels** on a project's Saved Reels page gives every reel in the ZIP the same caption tracks. In the AI chat both sit in the **⋮** menu on the reel card, as **Share reel** and **Download reel**. Strella files a reel whose clips all come from one project there automatically. When its clips span several projects, Strella asks which one the reel belongs to, since that choice sets the branding on the shared page. Either way, Strella does not add the reel to that project's Saved Reels. A shared link is a snapshot of the clips as they were when you created it, so editing the reel afterwards doesn't change what your recipient sees. Share again to send the newer cut. A reel saved through your AI assistant (the Strella connector) works differently. It carries one public link you can ask for when saving. Every later save through the assistant refreshes that same link to the reel's current clips. Edits made here in Strella do not refresh it, but the next save through the assistant does. Teammates who open the shared link while signed in to Strella get a download button on the player itself, so they can pull the MP4 without opening the project. They just need to belong to the workspace the reel came from, and they don't have to be switched into it. Reels are full-quality video, easy to drop in a Slack thread or a presentation. The reel is ready to send. The public link stays live until the study is closed. ## Tips - **Keep reels under 3 minutes.** Stakeholders tune out fast. Pick the 4 to 6 strongest clips, not every clip you have. - **Mix sentiment.** A reel of only positive reactions reads as a sales pitch; mixing in one or two negative reactions makes it feel honest. - **Add a one-line context** above the reel link when you share it: "Here's what blockers people hit at sign-up" frames what to watch for. ## Related Where highlights, the building blocks of reels, start. The text counterpart. The faster path to a reel. --- # Build & share a report Source: https://app.strella.io/docs/product/share/reports A report is a generated writeup of a study's findings: an executive summary, a profile of who took part, and a section for each finding with its evidence, plus a cover. Evidence can be a quote, a chart, or a highlight reel. Share a report by link, or download it as a PDF or a Word file. ![A generated report in Results → Reports: the Document view with a title, participant and insight counts, Research Objectives, a numbered Key insights list, and a Participants section.](/docs/share/reports.png) ## Steps The list shows every report you've built for this study. Click **Create report** to start a new one. In a study with no reports yet, the button reads **Create with AI**. Either button opens the [AI chat](/analyze/ai-chat) in a new tab with "Generate a report" already in the message box. Send it as is, or say what the report should cover ("a report on pricing feedback"). The chat writes the report and opens it beside the conversation. Click **Add to project** to save it to Results → Reports. The report opens with its title and one paragraph naming what the study set out to learn. The sections follow in this order: - **Executive summary**: the main point of the study in one sentence, then 3 to 5 bullets, one for each key finding. - **Who we spoke to**: who took part, described in words. - **Main findings**: one section per finding. Each opens with the claim in bold, followed by its evidence: a quote, a chart, or a highlight reel. See [Evidence in a report](#evidence-in-a-report) below. - **Strategic recommendations**: a table of actions, highest priority first, each tied to the finding behind it and saying why it matters for your decision. Scroll sideways to see every column, and use the ⋯ menu above any table to copy or download it. - **Limitations** and **Conclusion**. Tell the chat how you want it and the report follows: very short, one page, bullets, your own sections, another language or a particular audience. You can't edit a report inline. To change the wording of a report you built in the AI chat, open the **⋯** (more) menu and click **Edit**. That takes you back to the chat in a new tab, with the report open beside it, so you can ask for the change there. If that chat no longer exists, Strella opens a new chat for this study instead. For hand-edits, download the report as a Word file and edit it there. Open the **⋯** (more) menu and click **Update cover**. The dialog has two tabs: - **Upload**: use your own image. - **Create with AI**: describe the image you want, and Strella generates it. From the report header: - **Share**: every report has one link. Under **Access**, choose who can open it: - **Strella only**: people in your organization, after they sign in. Anyone else sees a page-not-found. - **Anyone with the link**: no sign-in needed. The link isn't listed or searchable, but anyone it's forwarded to can open it. Changing this never changes the link, so a link you've already sent keeps working. Once a report has more than one version, **Shared version** picks what the link opens: **Latest** always shows the newest version, while picking a specific version keeps the link on it even after you refine the report again. New links start on **Latest**, and switching between **Strella only** and **Anyone with the link** never changes the version. If you don't want outside readers to see edits as you make them, pick a specific version before you share. - **Download**: in the **⋯** (more) menu, open **Download** and pick **PDF** or **Word (.docx)**. The report is ready to share. With **Anyone with the link**, stakeholders can view it in a browser without a Strella account. ## Evidence in a report Each finding carries the evidence behind it. Every kind of evidence links back to the interviews it came from. ### Quotes and citations - **Strella shows a sentence's sources in one chip.** The chip shows the first participant's name, plus **+N** for the other participants quoted for that sentence. Two quotes from one participant count once, so a chip can show **+1** and still hold three quotes. Click it to page through each quote in full. Click a quote to jump to that moment in the source interview. - **A block quote ends with the participant's name.** Click the name to open the quote, with a link to play that moment. - **Copy a quote with its name.** Click the copy icon on a block quote. Strella copies the quote and the participant's name as they read on the page, ready to paste into a deck. ### Charts A chart shows how participants split across the categories of an [AI coding](/analyze/code-responses), meaning the categories Strella groups an open question's answers into. - **The grey line under the title says what the chart counts.** For example, "AI coding · 12 participants coded". It counts only participants with an answer to that question, so it can be smaller than the number of people in the study. - **Counts stay current.** Strella recounts every time someone opens the report, so a chart grows as more interviews complete. Deleting the coding replaces the chart with a notice. - **The biggest category comes first.** A coding with two to four categories, where each participant has one answer, draws as a donut. Other codings draw as a list of bars. - **Long charts group their small categories.** A chart with eight or more categories shows the six biggest, then one **Other** bar for the rest, with a line underneath naming each one and its count. When a participant can give several answers, there is no Other bar, and the line lists each left-out category with its own count. - **Highlighted bars match the text.** When a finding talks about particular categories, those bars stay in color and the rest turn grey, and a line under the chart says so. In a list of bars, the highlighted ones also move to the top. - **Download a chart from the ⋯ menu in its header**, as an image with its caption or as a spreadsheet (.xlsx). The spreadsheet lists every category, including the ones grouped into Other. ### Highlight reels A report can embed highlight reels saved in the study. - **A reel plays inside the report**, in Strella and on the report's share link. - **A PDF or Word download can't play video.** There, a reel shows its title, its clip and participant counts, and a **Watch highlight reel** link. - **Sharing a report shares its reels.** When you set a report to **Anyone with the link**, or download it, Strella creates a public link for each reel it embeds. Anyone who has the report can watch them. ## Tips - **Use the AI chat for tightly scoped reports.** Asking "build a report about pricing feedback" makes a focused report instead of a sprawling one. - **Ask for a report on one group.** Ask the AI chat for "a report on just the renters" and it counts and quotes only the participants who gave that screener or survey answer. Update now keeps counting that group. This report has no charts or saved reels, because those cover the whole study. - **Get the executive summary right.** Stakeholders read the executive summary first, so check that it states your main finding. To reword it, ask the AI chat to revise it, or download the report as a Word file and edit it. - **Every save is a version.** Saving a refined report from the AI chat adds a version to the same report instead of a second tile, and so does **Update now**. Use the clock icon next to the **⋯** menu to view or download an earlier version. ## Related The other shareable artifact: video moments tied to a theme. The themes that reports are built from.