In-depth interviews (IDIs): when one long conversation beats a survey

In-depth interviews fit when you need to know why people buy, switch or stop. How many to run, how to write the guide, and how to report what you found.

Instant Expert EditorialPublished Updated 6 min read

In-depth interviews are the right tool when you need to know why people buy, switch or stop, before you know the reasons well enough to count them in a survey. An in-depth interview (IDI) is a one-to-one research conversation, usually 30 minutes to two hours, about one person's experiences, decisions and reasons. Pathfinder International's guide calls the method intensive individual interviews with a small number of respondents (Pathfinder International).

Most IDIs are semi-structured: the interviewer works from a guide but can change the order, skip questions and follow anything unexpected. Nielsen Norman Group notes that the semistructured interviews UX researchers run are what market researchers call depth interviews (Nielsen Norman Group). Same conversation, different business cards.

Use IDIs when you need one person's reasons

Pathfinder says IDIs often put other data in context, such as figures showing that something changed, and help refine the questions for a later survey.

MethodChoose it whenWatch out for
In-depth interviewsYou need each person's reasons and the sequence of a decisionSmall samples; time to run and analyze
Focus groupsYou want reactions to a concept, name or message, and discussion between people helpsGroup influence; weak for detailed personal behavior
SurveyYou know the likely answers and need to count how common each isCan't explain answers or follow up

Pathfinder also prefers individual interviews to focus groups when people may not talk openly in a group, or when you need individual opinions. That covers much B2B research, where buyers may not want to discuss vendors, budgets or internal decisions in front of people from other companies. Focus group research covers when the group format is worth it.

A made-up payroll study

This example is hypothetical. A payroll software company sees that restaurant groups with 5 to 20 locations cancel at a higher rate than its other customers, usually in their first year. The data shows when they leave, but not why or what they switch to.

The team plans 45-minute video IDIs with three groups:

  • Finance managers at restaurant groups that cancelled in the past six months
  • Finance managers at similar groups that stayed
  • A few people who sell payroll software to restaurants and have seen switching decisions from the other side

1. Plan interviews per group, then stop when the stories repeat

There is no fixed number, but a 2022 systematic review of studies that tested when interviews stopped producing new findings put that point at 9 to 17 interviews, and noted that most of those studies had fairly similar participants and narrowly defined objectives (Hennink and Kaiser, Social Science & Medicine). Pathfinder's rule of thumb points the same way: you have done enough when you keep hearing the same stories, themes and issues.

Count per group, not per study: a handful of leavers and a handful of stayers, with room to add more if their reasons keep varying (how many customer interviews to plan).

2. Recruit the person who made the decision

In a restaurant group, the person who lived through the switch might be the controller, the finance director or an owner who still runs payroll; titles vary. Confirm early in each call what they personally did. Use research screener questions to check fit before booking; research participant recruitment covers channels and incentives.

3. Write a guide of no more than 15 open questions

GOV.UK's guidance on in-depth interviews suggests listing the topics you need to cover, ordering them so the conversation flows, writing a starter question and possible follow-ups for each, and testing the result on a colleague (GOV.UK). Pathfinder suggests no more than 15 main questions, open-ended rather than yes-or-no, with factual questions before opinion questions.

For the payroll study, a starter question might be: "Walk me through how you decided to change payroll providers. What happened first?" Follow-ups trace who was involved, what they compared and what decided it (how to write a discussion guide).

4. Run the session and mostly listen

GOV.UK's advice for the session:

  • Get informed consent and run through your introduction before you start.
  • Begin with a few easy questions, such as what their job involves.
  • Ask open, neutral questions, and follow up with simple prompts like "Can you tell me more about that?"
  • Focus on stories and real examples rather than how things should happen.
  • Stay quiet. The more you talk, the less they will.
  • Leave the guide when something new and relevant comes up.

Leave time at the end to clarify. GOV.UK notes interviews can run in person, by phone or by video. For recording, see how to ask for recording consent.

5. Report counts of people, not percentages

Pathfinder lists the method's main limitations: answers can be biased, interviews take time to run, transcribe and analyze, and results usually can't be generalized because samples are small and not random.

  • Summarize right after each interview, as Pathfinder recommends, and check facts you can verify independently. If a finance manager says their old provider charged per location, check that provider's pricing.
  • Compare across people. Look for reasons that recur among the leavers and are absent among the stayers. See affinity mapping and how to analyze customer interviews.
  • Count people. Pathfinder warns that numbers and percentages can give the impression that results apply to a whole population. "Four of the six leavers we interviewed switched after a payroll tax error" reports what you found. "67% of restaurant groups" claims more than six interviews can show.

Once interviews show the main reasons, a survey or product data can tell you how common each one is (qualitative market research, interviews versus surveys).

Your next step

Write down the decision your study informs, the two or three groups whose experience bears on it, and five starter questions about things they did. Book three interviews per group, then decide whether you need more.

Your own leavers and stayers are the easiest groups to reach, because you have their emails. The payroll salespeople are harder. On Instant Expert you can describe them, such as "sales reps who sell payroll software to restaurant groups," or paste the LinkedIn URLs or emails of people you already identified, and offer a 45-minute call to match the study design. You set what each person receives, our fee is added on top (a $100 offer costs you $125), and you're charged only when someone books. Each person has 7 days to pick a time and gets up to two reminders, so expect bookings to arrive over the first week rather than all at once. Start with finance professionals in restaurants and sales professionals in payroll software.

The call runs on a video link we provide, and our privacy policy says calls are recorded from when the requester joins, to settle attendance and disputes. That doesn't replace the consent step in section 4: ask for your own permission before you record or transcribe anything for the study.