Publishing Your Study

Publishing is the moment your study goes from a design document to a live research instrument. When you hit publish, Koji takes your research brief and creates an AI interviewer that's ready to have conversations with your participants. It also generates a unique interview link you can share immediately.

What Happens When You Publish

Publishing triggers several things behind the scenes:

1. Your AI Interviewer Is Created

Koji takes everything in your research brief — the problem context, methodology, target participant description, interview questions, probes, guardrails, and any context documents you uploaded — and configures an AI interviewer specifically for your study. This interviewer knows your research goals, follows your chosen methodology, uses the questions you approved, and adapts its probing based on participant responses.

The interviewer is unique to your study. It's not a generic chatbot — it's a purpose-built research conversationalist designed to execute your study plan.

2. Voice Agents Are Created (If Enabled)

If your study has voice interviews enabled, Koji automatically creates dedicated voice agents for each language your study supports. These voice agents handle the conversational flow, structured question capture, and real-time probing — all through natural speech. Voice agent creation happens automatically during publishing; no additional configuration is needed.

3. Question IDs Are Assigned

Each structured question in your study receives a stable identifier at publish time (if one hasn't already been assigned). These IDs flow through from the brief to the interview to the report, ensuring that your aggregate data — charts, distributions, averages — accurately maps responses across all interviews, even if question text is lightly edited later.

4. Your Interview Link Is Generated

You receive a unique URL that you can share with participants. When someone opens this link, they're taken directly into the interview experience. No accounts, no downloads, no friction — just a conversation.

You can share this link via email, embed it in a recruitment message, post it in a community, or distribute it however you normally reach participants. For more on distribution strategies, see Sharing Your Interview Link.

5. Your Study Becomes Active

Once published, your study moves to an active state. Interviews can begin happening at any time — whenever a participant clicks the link. You'll be able to monitor incoming responses from your dashboard.

Requirements for Publishing

Koji checks that your research brief has the essential components before allowing you to publish. Your brief needs:

• A problem statement — the core question or goal your study addresses
• Target participant information — at minimum, either the required experience or behavior of interest must be defined so the AI knows who it's talking to
• Key questions — at least a core set of questions for the AI interviewer to follow

Note that a methodology is helpful but not strictly required for publishing. If you haven't selected one, the AI interviewer will use sensible defaults based on your questions and problem statement.

If any required fields are missing or incomplete, Koji will prompt you to fill them in before publishing. This safeguard ensures every live study has enough structure to produce meaningful conversations. For more on how this validation works, see How the Quality Gate Works.

Pre-Publish Checklist

Before you hit publish, take a few minutes to review your study end-to-end. Once interviews start, you'll want to be confident that everything is right. Here's what to check:

Research Brief Review

• Problem statement — Does it clearly and accurately describe what you want to learn?
• Target participant — Is the description specific enough that you could use it as a recruitment screener? Does it focus on behavior and experience rather than demographics?
• Interview questions — Read every question out loud. Do they flow naturally? Are they open-ended and non-leading?
• Structured questions — Are scale ranges appropriate? Do choice options cover the likely answers? Are ranking items clear? See the structured questions guide for configuration tips.
• Probing configuration — Do follow-up questions dig into the right areas? Is anchor probing enabled on key scale questions?
• Question count — Is the total number of questions realistic for your target interview length?
• Language — Would your target participants understand and feel comfortable with the way questions are worded?

Context and Settings

• Context documents — Have you uploaded any relevant background materials? Are they current and accurate?
• Interview length — Is the target duration appropriate for the depth of questions you're asking?
• Interview mode — Is the mode (structured, exploratory, or hybrid) appropriate for your research goals?

Interview Experience

• Interview mode — Are you using voice, text, or both? Make sure the experience matches your participants' expectations. Learn more about voice interviews and text interviews.
• Branding — Have you customized the look and feel of the interview page to match your brand?

Logistics

• Recruitment plan — Do you know how you'll share the interview link? Do you have your participant list ready?
• Timeline — When do you need results? Allow time for participants to complete interviews at their own pace.
• Team awareness — Does your team know the study is going live? Will they be monitoring incoming results?

How to Publish

Once you're satisfied with your review:

1. Open your study in the design workspace
2. Review the research brief one final time in the artifact panel
3. Click the publish button
4. Koji validates your brief, assigns question IDs, creates the AI interviewer (and voice agents if enabled)
5. Your interview link is generated and ready to share

The process takes just a moment. Once published, you'll see your interview link prominently displayed so you can start sharing it right away.

After Publishing

Once your study is live, here's what to expect:

Monitoring Responses

As participants complete interviews, their responses appear in your study dashboard. You can review individual conversations, see how many interviews have been completed, and start identifying patterns. Structured question responses are automatically aggregated — you'll see charts and distributions building up as more participants respond.

Making Changes After Publishing

If you discover an issue with your study after publishing, you have options:

• Minor adjustments — You can update certain settings without unpublishing
• Significant changes — For substantial modifications to the research brief, you'll need to unpublish, make changes, and republish. Be aware that any interviews conducted before the change used the original brief.

Important: If you've already collected some interviews, think carefully before changing the brief. Mixing data from different versions of the interview can complicate your analysis. In many cases, it's better to finish the current round, analyze what you have, and create a new study for the revised approach.

Pausing and Resuming

If you need to temporarily stop accepting new interviews — perhaps to review early results before continuing recruitment — you can unpublish the study. The interview link will stop working, and existing data is preserved. When you're ready to continue, simply republish.

Common Scenarios

"I published but want to change one question"

If no interviews have been completed yet, unpublish, make the change, and republish. The same interview link will work. If interviews are already in, consider whether the change is significant enough to warrant starting fresh.

"My interview link isn't working"

Make sure your study is in a published (active) state. If you recently made changes and republished, the link should still be the same. If you're sharing the link and participants are having trouble, check that you're copying the full URL.

"I want to run the same study with a different participant group"

The best approach is to create a new study. You can use your existing study as a starting point and adjust the target participant description and any participant-specific questions. This keeps your data clean and your analysis straightforward.

"I'm not sure if my study is ready"

If you're unsure, go through the pre-publish checklist above. You can also ask the AI Consultant to review your brief one more time — say something like "Can you review the brief and flag anything I should fix before publishing?" It's always better to spend an extra ten minutes reviewing than to discover an issue after interviews are underway.

Next Steps

Once your study is published, it's time to get participants. Check out these guides for what comes next:

• Sharing Your Interview Link — strategies for distributing your link and recruiting participants
• Customizing Branding — making the interview experience feel like your brand
• Voice Interview Experience — what participants experience in voice mode
• Text Interview Experience — what participants experience in text mode
• How the Quality Gate Works — understanding the publish validation process

Further reading on the blog

• How to Run AI-Powered Customer Interviews at Scale — Learn how to conduct effective customer interviews at scale using AI. This comprehensive guide covers everything from planning and question
• Why AI Interviewers Are the Future of Customer Research — AI interviewers are transforming how product teams conduct customer research, enabling conversations at scale without sacrificing depth or q

<!-- further-reading:blog -->