Use Studies to learn throughout an initiative

Use a study for questions like:

• Where would users click first?
• Which option do users prefer?
• What is confusing on this screen?
• Does this plan match what customers thought they were getting?
• Should this idea move forward, change shape, or stop?

Do not use a study for broad intake or ongoing voting. If customers need to submit or vote on ideas over time, use the Ideas Portal.

Choose AI-guided interview when natural follow-ups will produce better evidence than a fixed form. Zentrik turns the Study goal and Steps into a reviewable 2–5 minute interview plan with the intended topics, approximate duration, attached assets, and any missing inputs that block publishing.

The Steps view is the authored journey and plan. It is not the live interviewer. After you publish, the participant link opens the AI interviewer, which works through those topics one at a time, chooses natural wording, and asks a short follow-up when the answer needs more detail. The interviewer stays inside the objective and authored topics; it does not invent a new study.

The participant first sees the study disclosure, then chooses Use microphone or Type instead. In microphone mode the interviewer speaks and the participant can answer aloud; in typed mode the same conversation runs as text. Voice is optional, and the transcript—not the audio recording—is what the participant reviews and submits. During the interview, the participant can interrupt, pause both sides of the audio, mute their microphone, type a response, skip a topic, or finish early. When a topic refers to an attached screenshot or video, the interviewer can present that specific asset in the conversation before asking about it.

Finishing early shows how many topics remain and records those topics as skipped rather than pretending they were covered. Before submission, the participant reviews the transcript, can return to the interview, and can add a correction. If the study includes a closing rating, it appears as one final step and stays separate from the interview answers. Zentrik stores the transcript, topic coverage, and the assets that were shown. Generated findings cite the participant turns that support them or point to the completed structured activity that produced the evidence. The Study-level findings draft carries this bounded evidence forward so product teams can inspect the source instead of accepting an unsupported summary.

The authoring-to-learning path is: review the generated interview plan, publish the Study, share the participant link, inspect Responses and topic coverage, then generate or refresh Findings from the saved evidence. Findings are a reviewable draft; the product team still decides what to change.

Use a self-guided study when exact wording, fixed answer choices, or asynchronous completion matters more than conversational probing. Use an AI-guided interview for exploratory questions, workflows that benefit from examples and follow-ups, or visual material that is easier to discuss in context.

Choose Live interview when a product manager or researcher will lead the conversation. Zentrik keeps the work inside the normal Study lifecycle while giving the team a shared discussion guide, an interview inventory, and findings grounded in imported transcripts.

The short path

1. From Studies, open New → Live Interview. From an idea or initiative, open its Studies workspace and choose the same method.
2. Choose Draft with AI to prepare a guide from an intent, or choose Start blank when you want to shape the guide yourself. A blank draft is intentionally empty; it does not invent a research goal.
3. In a draft, use Overview → Add completed interview to attach a transcript that already exists in Signals. You can also open Guide first and complete the learning question, audience, opening, topics, and closing note.
4. Start the Study when the guide is ready. Then use Interviews → Add interview to either plan a future conversation or add one or several completed transcript Signals.

When the Study is active, use Add interview to either:

• plan an upcoming conversation with a suggested or searched account
• add one or several completed transcript Signals already in Discovery
• import a new transcript through any connected conversation source, then return to attach it

Draft studies can attach completed transcripts for context while the guide is still being shaped. Planning a future interview and copying its unique Pulse link become available after the Study starts.

Planning an interview does not create an empty Signal. It creates a place to manage the account, optional participant details, and a unique Pulse link that can be copied before or during the call. When the transcript arrives, attach it to the planned interview. Zentrik keeps the transcript on its canonical Signal and uses the Study relationship for guide coverage and findings.

Completed interviews can come from Fireflies, Granola, another supported conversation provider, or a manual transcript import. The source icon and label stay visible, and you can correct the participant or account before attaching a single interview. When adding several interviews at once, each keeps the accounts already associated with its Signal.

Use Interviews to see planned and completed conversations, copy Pulse links, resolve processing issues, and open the source transcript. Use Guide to refine the shared conversation structure. Guide revisions stay explicit so the team can see which version informed each interview. Use Findings to review themes, contradictions, gaps, and cited transcript turns before recording the Study's impact on the initiative.

Create a study from the New menu, or open an idea or initiative and use its Studies workspace. The draft separates the internal study goal from the participant-facing journey.

A useful draft usually has:

• a clear internal study name
• a study goal with participants, decision, and learning question
• optional welcome and thank-you pages with their own participant-facing copy
• one or more participant-facing steps
• an image, screenshot, or prototype frame when the question is visual
• an optional captioned YouTube, Vimeo, or Loom walkthrough for show-and-ask research
• a heatmap target area if you need to know whether users found the intended action
• a text, single-choice, or ranking question when you need a reason or preference
• an always-on or answer-specific follow-up when a structured answer needs explanation
• an optional end-of-study rating about the study, initiative, feature, or product, with wording tailored to the question

New rankings default to an unranked, randomized choice pool and require participants to confirm the order they built. This reduces anchoring without relying on drag and drop.

Keep the study narrow. A 90-second answer is often better than a survey that tries to settle every product debate at once.

Use this path when someone needs to try a clickable flow before answering.

The short path

1. On the linked idea or initiative, choose Prototype.
2. Open the ready revision you want to test. If there is not one, build it in Zentrik or attach an existing prototype link.
3. Review the preview. Choose Test with people for a self-guided Study, or open the adjacent menu and choose AI-guided interview for conversational follow-ups.
4. Keep the goal and Steps focused on one task or decision. Preview, publish, and copy the participant link.

The Study is pinned to that exact revision. A newer prototype version will not change it. To compare revisions, start another Study from the newer version.

Participants first open the prototype in a new tab, complete the task, and return to the Study. They can reopen it while answering.

What Zentrik records

Zentrik keeps the tested prototype and revision with the Study. It also keeps submitted self-guided answers or the reviewed AI interview transcript, completion state, and participant or account context when the participant is identified.

Zentrik does not currently record every click or action inside the prototype. To learn where people click, add a Heatmap step with a representative screenshot. To understand paths, errors, or recovery, ask one direct follow-up question.

Use Linked work inside the study to attach the ideas, initiatives, opportunities, and product features that should inform it.

Zentrik can propose related work from the study goal, but every proposal stays in a review queue until a product manager attaches or dismisses it. Dismissed links are remembered so they are not immediately proposed again. You can also search the workspace and attach work manually.

A study always keeps at least one linked work item. Adding or removing another link does not move the study, duplicate responses, or rewrite its findings.

Preview the study before you publish it. Check the exact screen, question text, answer options, target areas, and step order.

Draft studies can be edited. Published study content is intentionally locked so responses stay comparable. If the prompt changes materially, clone the study or create a new one instead of editing live data.

When participant access is configured for your workspace, publish the study and copy the share link. Send it to the users or accounts whose judgment matters for the decision.

Studies can use the same customer identity model as the portal when a workspace is configured that way, but the feature is independent: linked work does not make the study part of a portal backlog.

Start with a small segment. A few relevant users can be more useful than a large generic audience when the question is tied to a specific workflow or product area.

Participants can answer open-text questions by typing or by selecting Speak inside the answer field. Voice is optional: the question and submission flow work the same way for participants who prefer to type.

The participant controls when recording starts and stops. While recording, the control shows elapsed time and clear actions to discard or finish. Zentrik then places the transcript in the normal answer field so the participant can review, edit, or replace it before submitting the study. In results, Transcribed from voice identifies how the answer began without changing how researchers read or use it.

Zentrik does not retain the audio recording after it is sent for transcription. The submitted response contains the reviewed text, not an audio attachment. Recordings stop automatically after two minutes to keep answers focused and uploads bounded.

If microphone access is denied, recording is unavailable, or transcription fails, the participant can retry or type the answer without losing the rest of the study. Voice answers currently apply to open-text questions; choice and ranking questions keep their existing interaction.

Results have three useful levels:

• Summary shows response count, clicks, target-area performance, and answer distributions.
• Responses preserves participant, account, answer, and notes context when available.
• Conclusions turns the evidence into a reviewable learning report.

For heatmap steps, look at where users clicked and whether they found the intended area. For question steps, read the answers before reducing the result to a count. The useful output is not just a chart; it is a better decision about the idea.

On an idea’s Studies tab, Learning summary brings the related Studies together. Use Review evidence to inspect the findings and source Studies behind it. Still uncertain names one material gap without presenting it as a fact. When New evidence appears, refresh the summary deliberately; the last confirmed summary stays visible until the refresh succeeds.

If the evidence supports another focused question, Plan next study prepares one proposal for review. Create draft creates an unpublished Study and opens it for editing. It never publishes or invites participants automatically.

Study evidence cannot prove that the team changed direction. On an initiative, use Record impact to state what the product team actually decided:

• confirmed the current direction
• changed scope or direction
• raised a risk
• decided that no change was needed
• needs a follow-up study before deciding

The initiative learning summary uses these human records for product impact. If none exists, Zentrik says No direction change has been recorded rather than inferring a pivot from responses or generated conclusions.

After responses arrive, generate or write conclusions on the study, then record the product decision on the initiative when one has been made.

Treat generated conclusions as a draft for the product team to review. Conclusions should summarize the evidence and its limits; the separate product-impact record states what changed.

The next step should be explicit. Common outcomes are:

• refine the idea
• link the study to an opportunity
• create or update an initiative
• move the decision into a planning review
• close the idea if the evidence argues against it

The goal is to keep customer intent attached while the work moves from learning into shaping, planning, and delivery.