Best practices to use AI Moderated Interviewer

Conducting sessions with AI Moderated Interviewer is a fast, effective way to collect rich qualitative insights. However, the depth and quality of insights depend on how you set up the interview. In this article, you’ll find best practices to help you run effective interviews with AI Moderated Interviewer.

To begin, choose a clear, descriptive title or topic for your interview so the objective is immediately obvious. This helps participants anticipate the questions they’ll be asked and be better prepared while responding. The AI may also use this information to guide the conversation as it adapts to participant answers.

Avoid using internal company nomenclature or project names as the interview topic. It may be confusing to the participants and the AI Moderator.

If you have any disclaimers or terms and conditions that participants must acknowledge before starting the interview, add them to the Compliance and Legal section under Advanced settings. Participants won’t be able to begin the interview until they’ve provided consent by clicking “I understand.”

We don’t recommend collecting legal consent through conditional questions in the discussion guide unless you specifically need an audio or video recording from participants. If a participant declines consent mid-interview, you may end up with incomplete sessions.

This can be especially difficult to manage when you need to collect a specific number of completed interviews. You can save time by screening these participants before they begin the interview.

If there is any context you can provide to the AI that will help it conduct the interview, add it under Add research goals instead of the discussion guide.

Remember to mention the objectives you want to achieve through these interviews. For example,

Remember to write down points you would want to tell a peer who is helping you conduct the interview.

These research goals will not be visible to participants. The AI Moderator will use them to guide and adapt to the conversations.

You can choose how participants record their responses during an interview. You can either record the entire conversation (Real-time conversation), or let participants use push-to-record (Turn-based responses) to answer only when they’re ready.

Turn-based responses give participants more time to think before answering. If you want to observe a participant’s thought process as they work through each question, use the Real-time conversation option.

If participants are unlikely to have access to a quiet place while completing the interview, consider using the turn-based format.

A well-structured discussion guide helps the AI Moderated Interviewer lead natural, focused conversations. You can collect insights that map directly to your research goals. You’ll need to keep the guide ready before you can set up the interview.

In case you don't want to start from scratch, pick a Marvin template from the Discussion guide section.

You can also use Marvin’s AI as a research partner to help you prepare your guide.

Or consider uploading a guide that your team is already using. You can then edit it to match the AI moderated interview.

Mention a suggested interview duration. This helps the AI pace the conversation and ensures participants don’t feel rushed or fatigued. Make sure the AI moderator will be able to complete all the questions in this recommended duration.

Shorter interviews work well for screeners or validation studies, while longer sessions are better suited for exploratory or in-depth research.

Include brief explanatory text at the start of the guide to help the AI introduce the session. For example, you may want the AI moderator to read out a short overview you wrote. Or you could ask it to explain the objective of the interview in simple, participant-friendly language.

If you need to add instructions throughout the interview, include them in the relevant sections of the discussion guide rather than grouping them all at the beginning.

You don’t need to add guidance on how to gently bring participants back on topic. The AI moderator handles this automatically. The AI uses your instructions along with your research goals to adapt the conversation in real time.

Conditional questions allow you to branch the conversation based on participant responses. For example:

Make sure every question in your discussion guide is marked with a /. The AI uses it to distinguish between questions and additional instructions or context. Any unmarked text is treated as guidance for the AI rather than a question for the participant.

You can set one of three probing levels for each question:

Identify the few questions that directly tie back to your core research objectives. For these questions, consider using Probe more to signal to the AI that it needs to explore deeper.

Avoid using Probe more for questions that are complex but not critical, as this can lengthen the interview without adding meaningful insight.

Avoid overly long or complex questions. Participants are more likely to give incomplete answers or misunderstand what’s being asked. Instead, break complex topics into shorter, simpler questions and use probing to go deeper when needed.

Finally, make sure questions flow naturally from one topic to the next. Abrupt topic changes make it harder for the AI to adapt and guide the conversation smoothly, and can feel disjointed for participants.

Before sharing your interview with participants, run a test interview directly from the Settings page. You can check the overall flow and ensure you’ve covered everything in your discussion guide. We strongly recommend doing this as a final step before launch.

You can experience firsthand how the AI moderator introduces the topic, asks follow-up questions, and transitions between sections. This helps you spot unclear questions and missing context, and ensures the participant experience matches your expectations.

Marvin’s AI Moderated Interviewer does not automatically stop collecting interviews after a fixed number of responses. However, you can easily monitor incoming interviews and switch off the study at any time.

If you need to make changes mid-study, you can revise your discussion guide and continue collecting responses with the updated version. To limit the total number of interviews, we recommend sharing the interview link with a smaller, targeted group of participants rather than distributing it broadly.

You can manage AI Moderated Interviews from the Files tab of the corresponding project.

Once you’ve completed the interviews, use Marvin’s helpful in-built workflows to review and analyze your data efficiently.

Add file tags to organize interviews and make them easy to reference later.

Read AI-generated summaries to quickly spot key themes.

Create auto notes to capture important insights in each interviews.

Check project updates to quickly catch up on new interviews.

Together, these tools help you move from raw interviews to actionable insights faster, all within the same workflow.