--- title: "Qualitative interview: how to run one, with examples" description: "What a qualitative interview is, the three types, how to run one step by step, 20 questions by stage, and an annotated excerpt with every follow-up explained." url: "https://raconte.ai/en/blog/qualitative-interview" --- # How to conduct a qualitative interview, step by step What a qualitative interview is, the three types, how to run one step by step, 20 questions by stage, and an annotated excerpt with every follow-up explained. September 30, 2026 Godefroy Key takeaways * A qualitative interview is a conversation where the person decides how to answer and the interviewer follows up on what they said. Without the follow-ups, you are reading out a questionnaire. * Ask about the last time the person actually did what you are studying: they then recount the events in the order they happened, instead of giving an opinion about their habits. * The person you interview should do most of the talking, around 80% of it. Each time you talk, you use up time they could have spent telling their story. * Silence is a question in its own right. Wait three seconds after an answer stops and people often add the part they had not planned to say. * Write a short summary within the hour after each interview. After three interviews in a row, you start mixing up who said what. A qualitative interview is a one-to-one conversation in which you ask open questions, let the person answer in their own words, and follow up on what they say, to understand how they experienced something and why they acted as they did. It usually lasts 45 to 60 minutes, runs from a short guide rather than a script, and produces stories you read and compare rather than numbers you count. Most of the skill is in the follow-ups, which is why two people with the same guide come back with very different material. The questions you prepare beforehand go into an [interview guide](/en/blog/interview-guide). This page covers the conversation itself: the three types of interview, the steps from recruitment to write-up, twenty questions sorted by the moment you ask them, and a short annotated excerpt that shows each follow-up at work. ## What a qualitative interview is A qualitative interview is an interview in which the person answering chooses the form of the answer. When you ask “walk me through the last time you did X”, the answer can take two sentences or ten minutes, go back in time, name people and tools you had never heard of. The interviewer then picks up on what was said and asks for more. A spoken questionnaire has neither the open answer nor the follow-up: its answers fit into boxes decided in advance. Steinar Kvale, a Danish psychologist at Aarhus University, and his co-author Svend Brinkmann, a psychologist at Aalborg University, describe two ways of thinking about it in _InterViews_, the standard textbook on the method. In the first, the interviewer is a miner digging for facts that already sit in the person’s head. In the second, the interviewer is a traveller who walks alongside the person and comes back with a story built during the conversation. Most product and UX researchers take the traveller’s view: what people tell you depends on how you listen and what you ask next. A qualitative interview records what people say about what they did. To see what they actually did, on screen or over three weeks, you need a usability test or a diary study. The interview is one of several [qualitative research methods](/en/blog/qualitative-research-methods), each suited to a different question. ## Structured, semi-structured, unstructured: the three types Interviews come in three types, set by how much you fix before the conversation starts. Structured Semi-structured Unstructured What you prepare Every question, in order Themes, one opening question each, planned follow-ups One opening prompt What you do during it Read the questions as written Follow the person’s story, check each theme gets covered Follow wherever the person goes Follow-ups Rare, often forbidden Constant, the core of the method Constant, with no set destination Typical length 15 to 30 minutes 45 to 60 minutes 60 to 90 minutes When to use it Many people, answers you want to compare line by line Understanding why people do what they do A subject you know almost nothing about Nearly all product research, UX research and customer research uses semi-structured interviews. The structured type is close to a survey read aloud, while the unstructured one suits the first weeks of exploring a field whose vocabulary you have not learned yet. You will also meet other names: in-depth interview, narrative interview, ethnographic interview. They describe a posture or a setting rather than a fourth type, since almost all of them run semi-structured. A focus group is a separate method, with several people at once and a group dynamic that changes what each person says. ## How to conduct a qualitative interview, step by step 1. **Write the question the study must answer, in one sentence.** “Why do customers stop using the export after a month” is a study question. “Learn about our users” is not one yet. 2. **Recruit people who lived the situation recently.** Someone who cancelled last week still remembers the sequence of events. A year later, all that is left is a verdict. Plan six to twelve people per segment. 3. **Prepare a semi-structured guide.** Three to five themes, one opening question each, two planned follow-ups under each. Copy [a full interview guide example](/en/blog/interview-guide#an-interview-guide-template-with-a-full-example) and adapt it. 4. **Open by framing the conversation.** Say who you are, what you are trying to understand, that there is no right answer, and ask for consent to record. 5. **Start with an open question that asks for one specific, recent episode.** “Walk me through the last time” gets you more than “how do you usually”. 6. **Follow up on what the person said, in their words.** Repeat the word they used, ask what happened next, let a silence run. 7. **Keep your own hypotheses for the last ten minutes.** Test them once the person has told their story without your help. 8. **Close with an open question, then write a summary within the hour.** “What should I have asked you that I did not?” regularly opens a new subject. Kvale and Brinkmann spread the same work over seven stages, from choosing the theme to writing the report. Interviewing is only the third of the seven. A team that improvises the first two, the study question and the design, usually finds out while reading the transcripts that the interviews answered a different question. Three habits make most of the difference during the conversation. Ask one question at a time. “How was the setup, and did your team adopt the tool?” gets an answer to the second half. Steve Portigal, a research consultant and the author of _Interviewing Users_, keeps coming back to it: a stacked question makes the person choose which half to answer, usually the easier one. Let the person talk. Rob Fitzpatrick, the author of _The Mom Test_, puts the target at 80% of the speaking time for the person being interviewed. If you catch yourself explaining your product or your theory, the interview has turned into a pitch. Use their words. If they say “the report thing”, your next question uses “the report thing”, even if the feature has a proper name. When you correct their vocabulary, they hear that there is a right answer. ![The eight steps of a qualitative interview, from the study question to the summary written after the call](/.netlify/images?url=_astro%2Finterview-steps-en.BDfNpfSs.jpg&w=1376&h=768&dpl=6ad1cc428d394a5c89e0155b) ## 20 qualitative interview questions, by stage The questions follow the order of a conversation. Swap the bracketed parts for your subject. Kvale and Brinkmann name nine kinds of interview question, including introducing, follow-up and interpreting questions, and count silence as one of the nine. The opening, story, follow-up and closing questions cover most of those kinds. ### Opening: context and warm-up 1. Tell me about your role and what a normal week looks like. 2. How long have you been dealing with \[the subject\]? 3. How did you handle \[the task\] before you had \[the tool or process\]? 4. Who else is involved when \[the task\] comes up? 5. What made you agree to talk to me today? ### The story: one real episode 6. Walk me through the last time you \[did the thing you are studying\]. 7. What happened right before that? 8. Where were you, and what were you using? 9. What did you do next? 10. At which moment did it get difficult? 11. How did it end? ### Follow-ups: getting past the first answer 12. “\[Their last word\]?” (said as a question, then wait) 13. Can you give me an example? 14. What do you mean by \[their word\]? 15. What went through your mind at that moment? 16. You said \[their phrase\]. Tell me more about that. ### Closing: what you did not think of 17. If you could change one thing tomorrow, what would it be? 18. What have you tried that did not work? 19. Who else should I talk to about this? 20. What should I have asked you that I did not? The follow-up questions do most of the work. Ask the opening and story questions in the order of your guide, but use a follow-up whenever an answer stays vague. You will find other wordings in our bank of [45 open-ended questions for qualitative research](/en/blog/open-ended-questions-qualitative-research), sorted by research goal (discovery, onboarding, churn, pricing), along with the four phrasings that turn an open question into a closed one. ![The four stages of questions in a qualitative interview, with one example question each](/.netlify/images?url=_astro%2Fquestion-stages-en.BIUp9hOK.jpg&w=1376&h=768&dpl=6ad1cc428d394a5c89e0155b) ## A short annotated interview excerpt We made up this excerpt and its participant for a fictional study on why people cancel a meal-kit subscription. The excerpt shows how four follow-ups, one of them a silence, turn a vague first answer into a finding. > **Interviewer:** Walk me through the last week you had a box delivered. > > **Participant:** It was fine, really. The recipes are good. I just didn’t use it enough. “Fine” and “didn’t use it enough” are a verdict. If the interviewer asked why at this point, the answer would be a tidy reason, probably the price. The interviewer asks for the sequence instead. > **Interviewer:** When did the box arrive that week? > > **Participant:** Tuesday. I cooked one of the meals that night. Wednesday I had a late meeting so we ordered in. Thursday I don’t remember. Two of the kits ended up in the bin on Sunday. The answer now has days, an event and a fact the verdict hid: food thrown away. The interviewer repeats the word that stands out and says nothing else. > **Interviewer:** In the bin? > > **Participant:** Yeah. That’s the part I hated, paying for food and throwing it out. Every Sunday I felt bad about it. > > **Interviewer:** _(waits three seconds)_ > > **Participant:** And my partner cooks at the weekend but hates following recipes. So in practice it was me, three evenings a week, and I’m never home three evenings a week. When the interviewer repeated “in the bin”, the participant said what bothered them most: throwing away food they had paid for, more than the price itself. After three seconds of silence, the participant added a constraint the interviewer had never asked about: the box needs someone at home to cook from a recipe three evenings a week, which nobody in this household does. The last follow-up asks what replaced the product, because a workaround shows the actual need. > **Interviewer:** What did you do the week after you cancelled? > > **Participant:** Back to the supermarket. I only buy for two dinners now. A cancellation survey would have logged “didn’t use it enough”, perhaps “too expensive”. The interview gives a product team something to act on: a two-dinner box, or a way to skip a meal before it ships. ## Qualitative vs quantitative interview A quantitative interview is a structured interview whose answers are coded as numbers: the same closed questions, in the same order, read the same way to every person, often over the phone. Census interviews and telephone surveys work this way. You get a count of how many people hold a view. The reasons behind it come out in a qualitative interview, through open questions and follow-ups. Qualitative interview Quantitative interview Questions Open, adapted to each person Closed, identical for everyone Follow-ups Constant None, to keep answers comparable Sample 6 to 12 people per segment Hundreds, sized for statistical precision Analysis Reading, coding, building themes Counting, cross-tabulating What you learn Why, and in which words How many, and how often The two work best in sequence. When you [combine qualitative and quantitative research](/en/blog/qualitative-vs-quantitative-research), run the qualitative interviews first, so the answer options of your later survey come from the words people actually used rather than from your guesses. ## From transcript to findings The analysis starts before you open the transcripts. Portigal recommends writing a top-of-mind summary right after each session, before the next interview blurs your memory. Note down the three things that surprised you, the quotes you want to find again and the questions you want to add. Then work from a [verbatim transcript](/en/blog/interview-transcription) with the speakers separated. A summary of an interview is already an interpretation, so codes placed on it describe your reading rather than the participant’s words. The usual path for analysing qualitative interview data looks like this: ```mermaid graph TD A["Recording"] --> B["Verbatim transcript, speakers separated"] B --> C["Summary written after each interview"] B --> D["Codes on short extracts"] D --> E["Codes grouped into themes"] C --> E E --> F["Themes checked against the transcripts"] F --> G["Findings, each backed by quotes"] ``` The step teams skip is going back to the transcripts. A theme you cannot back with three or four extracts from different people is an impression. In a separate article, we take one real extract through the six phases of [thematic analysis](/en/blog/thematic-analysis), the most common method for coding interview transcripts, from the first codes to a theme you can defend. You can code about ten transcripts by hand, in a spreadsheet. Past that, use [software built for coding qualitative data](/en/blog/top-qualitative-data-analysis-software), which keeps the link from each theme to the quotes behind it. ## Running qualitative interviews at scale without a moderator Most teams know how to run a good qualitative interview. What stops them is the cost of each one: a slot to book in two calendars, an interviewer who is free, an hour of conversation and another hour of notes. That cost repeats for every person, which is why a study planned for twelve interviews often stops at six. An asynchronous interview removes the booking. The person gets a link, opens it whenever it suits them, and answers out loud. We built [Raconte](/en) for that: you write the brief once, then an AI runs the voice conversation with each person. It asks one question at a time, follows up whenever an answer stays vague, and skips a question the person already answered in passing. Each interview comes back with the transcript, the audio, a summary and the sentiment of every participant message, plus reports aggregated across everyone who answered the same brief. People also say more when they speak. In a study of just over a thousand smartphone survey respondents, the survey methodologist Jan Karem Höhne and his co-authors [measured spoken answers](https://journals.sagepub.com/doi/full/10.1177/08944393231160961) averaging around 55 words, against around 21 for the same open questions typed. The format has limits that matter for a qualitative study. * An AI only hears the voice. An interviewer in the room also sees the face, the hesitation before a click or the sticky note on the monitor, and can ask about them. * An AI’s follow-ups are coarser than an experienced researcher’s. The AI follows up on vague answers and asks for episodes, but an experienced researcher also picks up a change in tone, a contradiction twenty minutes apart, or a subject the person is avoiding. * For an interview about a conflict, a bereavement or a negotiation, bring in a human interviewer, in the room or on a call. In practice, run the first five or six interviews yourself to learn the vocabulary and sharpen the guide, then turn it into a brief and send it to forty or a hundred people asynchronously, more than most teams have the time to interview one by one. ## Where to start Write your study question in one sentence, recruit six people who lived the situation in the last few weeks, and prepare three themes with one opening question each. Run the first interview, write your summary within the hour, and cut the questions nobody could answer. After six interviews, you usually know the words people use, so the guide stops changing. If scheduling and note-taking hold the study back, try Raconte: the first 60 interview minutes each month are free, no card required. After that, you [pay per minute of interview](/en/pricing). ## Frequently asked questions ### What is a qualitative interview? A qualitative interview is a one-to-one conversation in which the interviewer asks open questions, lets the person answer in their own words and follows up on what they say. Its goal is to understand how someone experienced a situation and why they acted as they did. It produces stories to read and compare rather than numbers to count. ### How do you conduct a good qualitative interview? Prepare a semi-structured guide with three to five themes. Open the interview by explaining what you are trying to understand and asking for consent to record. Ask for one recent, specific episode rather than general habits. Follow up in the person’s own words, let silences run, and keep your hypotheses for the last ten minutes. Close by asking what you should have asked, then write a summary within the hour. ### What are some good interview questions for qualitative research? The most useful ones ask for a real episode: “walk me through the last time you did X”, “what happened right before”, “what did you do next”. Follow-ups such as “what do you mean by that” or repeating the person’s last word as a question do most of the work. Avoid questions that can be answered with yes or no, and hypothetical ones about what the person would buy or use. ### What is a quantitative interview? A quantitative interview is a structured interview in which every person answers the same closed questions, in the same order, and the answers are coded as numbers. Telephone surveys and census interviews work this way. You get a count of how many people hold a view, while the reasons behind it come out in a qualitative interview. ### What is the 80/20 rule in interviewing? In research interviews, the 80/20 rule says the person being interviewed should do about 80% of the talking and the interviewer about 20%. Rob Fitzpatrick gives this split in _The Mom Test_. When you talk more than that, you are usually explaining your idea or suggesting answers, so the person starts agreeing with you. ### How long should a qualitative interview last? A semi-structured interview usually lasts 45 to 60 minutes. Less than 30 minutes leaves little room for a detailed story and its follow-ups. Beyond 90 minutes, the person tires and answers get shorter. An asynchronous voice interview is usually shorter: it covers a narrower subject and goes straight to the first question. ### How many qualitative interviews do you need? Plan six to twelve per homogeneous group. Greg Guest, an anthropologist at Family Health International, and his colleagues found nearly every theme within the first twelve interviews of their study, with the broad strokes visible after six, as long as everyone got the same guide. Very different profiles need their own six to twelve each. ### How do you analyze qualitative interview data? Work from verbatim transcripts with the speakers separated. Write a short summary after each interview, then code short extracts, group the codes into themes, and check every theme against the transcripts before you report it. Thematic analysis is the most common method for doing this. ### Can a qualitative interview be run without an interviewer? Partly. An AI can run a voice interview from a written brief, ask one question at a time and follow up on vague answers, which removes scheduling and lets you reach far more people. It hears the person without seeing them. Its follow-ups are also less subtle than an experienced researcher’s. Sensitive subjects still call for a human interviewer. Ready to launch your first interview? Describe your interview, share the link: an AI agent runs it by voice with each participant. You get back the transcript and analysis. [Try for free](/login) ## Continue reading [![Interview guide: the method and a full example](/.netlify/images?url=_astro%2Fthumbnail._6IslZAl.jpg&w=1376&h=768&dpl=6ad1cc428d394a5c89e0155b) July 29, 2026 ## Interview guide: the method and a full example The three interview families, a five-block structure, a complete guide you can copy, and the five question types that quietly ruin a study. ](/en/blog/interview-guide)[![45 open-ended questions for qualitative research](/.netlify/images?url=_astro%2Fthumbnail.x5EdDQ3f.jpg&w=1376&h=768&dpl=6ad1cc428d394a5c89e0155b) August 22, 2026 ## 45 open-ended questions for qualitative research A bank of 45 open-ended questions grouped by research goal, the reason each one is phrased the way it is, and the four ways a question closes back up. ](/en/blog/open-ended-questions-qualitative-research)[![Thematic analysis: the six phases, with a worked example](/.netlify/images?url=_astro%2Fthumbnail.DLYApfFv.jpg&w=1376&h=768&dpl=6ad1cc428d394a5c89e0155b) September 5, 2026 ## Thematic analysis: the six phases, with a worked example Braun and Clarke's six phases run on a real verbatim, the reflexive and codebook split, how many interviews you need, and what a language model misses. ](/en/blog/thematic-analysis)