If you want to understand why customers behave the way they do, not just that they do it, in-depth interviews are your most reliable tool. No survey can replicate the moment a respondent pauses, sighs, and says “it’s hard to explain” before revealing exactly what you needed to know.

This guide covers everything you need to run IDIs well: what they are, when to use them over other methods, how to write a discussion guide, how many interviews you actually need, and what to do with the data afterward. There is also a ready-to-use discussion guide template you can adapt for your next study.

What is an In-Depth Interview (IDI)?

An in-depth interview is a one-on-one qualitative research conversation between a trained moderator and a single respondent. Unlike a structured survey, IDIs are semi-structured: the moderator follows a guide but is free to probe, follow up and go wherever the conversation is most revealing.

The goal is depth, not breadth. Where a survey tells you what percentage of your customers feel frustrated an IDI tells you the specific moment the frustration started, what they tried before giving up and what a better experience would have felt like to them.

IDIs are used across market research, UX, product development, brand strategy and customer experience. They work equally well for consumer and B2B research and they scale better than focus groups for sensitive topics where social dynamics might suppress honest answers.

IDIs vs Focus Groups: How to Choose

Both methods are qualitative, but they serve different purposes. Focus groups generate social data. The dynamic between participants can surface tensions, trigger associations, and reveal how people talk about a topic with peers. They work well for concept testing, creative development, and gauging how ideas land in a group conversation.

IDIs generate individual data. You get each respondent’s unfiltered view without any group influence. Choose IDIs when:

•   The topic is sensitive: financial behaviour, health decisions, personal frustrations
•   Respondents are hard to recruit together: senior executives, niche professionals
•   You need to trace a full customer journey across multiple touchpoints
•   You suspect group dynamics will push people toward socially acceptable answers

For a fuller comparison of all three primary qualitative methods, read Focus Groups vs IDIs vs AI Interviews: When to Use Each.

How to Conduct an In-Depth Interview: Step by Step

1. Define your research objectives

Start with the business question, not the topic. “We want to understand the onboarding experience” is a topic. “We need to know why 40% of users drop off before completing setup” is a research objective. The more specific you are here, the tighter your discussion guide will be and the more useful your findings.

2. Recruit the right respondents

Your sample should reflect the people whose behaviour or attitudes you are trying to understand. For most consumer IDI studies, 8 to 12 respondents per segment is enough to reach thematic saturation. For more on this, see Sample Size for Qualitative Research: How Many is Enough?. Screen carefully. A respondent who does not meet your criteria can waste an entire interview slot.

3. Write your discussion guide

A discussion guide is not a questionnaire. It is a structured set of topics and probes that keeps the conversation on track while leaving room for the moderator to follow interesting threads. For a detailed walkthrough, including how the requirements shift when using AI moderation, see How to Write a Discussion Guide for AI Interviews.

4. Conduct the interview

Keep it conversational. The best moderators listen far more than they talk. Ask one question at a time. Let silence sit. When a respondent gives a short answer, probe with “tell me more about that” or “what do you mean by that” rather than moving on immediately. The most valuable moments in an IDI usually come after the first answer.

Aim for 45 to 60 minutes per session. Shorter interviews rarely reach the depth that justifies the method. Longer sessions lead to fatigue on both sides.

5. Record and transcribe

With consent, record every session. Relying on notes alone introduces recall bias and misses tone, hesitation, and emphasis, all of which carry meaning in qualitative research. Transcripts are essential for rigorous analysis.

IDI Discussion Guide Template

Here is a reusable framework you can adapt across projects. Sections are topic-based, not question-based, which keeps the conversation natural while ensuring coverage.

Opening (5 minutes)

Introduce yourself and the purpose of the session without revealing specific objectives. Confirm recording consent. Remind the respondent that there are no right or wrong answers. Start with easy warm-up questions: “Tell me about your role” or “Walk me through your typical week.”

Topic 1: Context and background (10 minutes)

•   Walk me through how you first encountered or started using [product/service/category]
•   What was going on in your life or work at that time?
•   What were you trying to solve? Probe: what had you tried before?

Topic 2: Core experience (20 minutes)

•   Tell me about a specific time when you [key behaviour you are studying]
•   Walk me through it step by step
•   What were you thinking at that moment? How did that make you feel? What did you do next?
•   What worked well? What felt frustrating or unclear?

Topic 3: Needs and expectations (10 minutes)

•   What would a better experience have looked like?
•   If you could change one thing, what would it be?
•   How does this compare to similar experiences you have had elsewhere?

Closing (5 minutes)

•   Is there anything we have not covered that you think is important?
•   Final open question: “If you had to describe this experience to a friend in one sentence, what would you say?”

How to Analyse IDI Data

Analysis is where most IDI projects lose quality. For a full breakdown of the analysis workflow, especially for AI-moderated interviews, see How to Analyse Qualitative Data from AI Interviews. The core principles are:

Start with transcripts, not memory. Read each one without trying to draw conclusions. Note themes, contradictions, and striking phrases as you go. Your memory of a conversation is already an edited version of it.

Use thematic analysis. Group observations into themes, then test each theme against all your data. A theme that appears in three interviews and contradicts two others is more honest than one memorable quote that felt representative.

Avoid the trap of counting. “Six out of twelve respondents mentioned X” sounds rigorous but is not. IDIs are not designed for frequency analysis. The question is not how often a theme appeared but what it means and why it matters for the business question you started with.

AI-Moderated IDIs: A More Scalable Approach

Traditional IDIs require a skilled moderator, which limits how many you can run affordably and how quickly you can field a study. AI-moderated interviews are changing both constraints. Platforms like Merren use Maya AI to conduct fully automated one-on-one interviews over WhatsApp. Maya follows a structured discussion guide and probes dynamically based on what each respondent says, much like a human moderator would.

Studies that would take three weeks with traditional fieldwork can go from brief to analysis in three days, with transcripts and thematic summaries generated automatically. This is not a replacement for human moderation in every context. But for scalable consumer insight, concept development, or early-stage discovery research, AI-moderated IDIs close the gap between qualitative depth and quantitative speed.

Common Mistakes in IDI Research

Leading questions are the most common problem. “Did you find the onboarding confusing?” presumes confusion. “How would you describe the onboarding experience?” does not. Review every question in your guide for any assumption embedded in the phrasing.

Over-recruiting the wrong profile wastes budget and produces noise. Eight highly relevant respondents will tell you more than fifteen loosely matched ones.

Skipping the pilot is a shortcut that costs you later. Run at least one pilot interview before full fieldwork to pressure-test your guide and timing.

Delivering findings without a clear “so what” frustrates stakeholders. For guidance on structuring your debrief, read How to Present Research Findings to Stakeholders. 

Related Reading

•   What is an AI Moderator? How It Works vs. Human Moderators
•   How to Write a Discussion Guide for AI Interviews
•   Sample Size for Qualitative Research: How Many is Enough?
•   Focus Groups vs IDIs vs AI Interviews: When to Use Each
•   How to Analyse Qualitative Data from AI Interviews
•   How to Present Research Findings to Stakeholders