Quick answer
A depth interview is a one-to-one, usually semi-structured qualitative conversation used to understand a participant's experiences, context, meanings and reasoning. Begin with a focused research question and purposive recruitment covering relevant variation. Obtain informed consent, use a flexible discussion guide, ask for the last real episode, reconstruct sequence and context, probe neutral follow-ups and distinguish observation from interpretation in notes. Analyze across interviews using traceable codes, cases, contradictions and negative evidence. Report patterns with sample boundaries and quotations that preserve context. Use interviews to generate and explain hypotheses, not to estimate prevalence or ask participants to forecast a market.
What a depth interview is
A depth interview is an individual qualitative conversation designed to explore a participant's experience, context and meaning in detail. It is usually semi-structured: a guide creates comparable coverage, while the interviewer follows relevant episodes and language rather than reading a fixed questionnaire.
The method is useful for sensitive topics, complex journeys, decisions involving several people and questions about why an observed pattern may occur. One-to-one privacy can make it easier to discuss uncertainty, failure, money or conflict than a group setting.
An interview produces situated accounts, not a transparent recording of the mind. People forget, rationalize and adapt stories to the conversation. Good practice grounds discussion in recent concrete events and combines accounts with behavior, artifacts or other methods.
Use interviews for the right question
Use depth interviews to discover language, motives, constraints, sequences, mental models and plausible mechanisms. They can explain what a funnel event means to customers or reveal why two people respond differently to the same proposition.
Do not use a small purposive sample to estimate how many customers hold a view, predict demand or select a winning concept by vote. Surveys, experiments and behavioral data answer prevalence and effect questions more directly after qualitative work defines what to measure.
Write the decision and knowledge gap. If the team already needs to see whether people can complete a prototype task, usability testing is the better primary method. If context and workaround matter, observation may be necessary alongside the conversation.
Build a depth-interview workflow
Frame a focused question, then create a sample matrix covering experiences likely to change the mechanism. Recruit people who recently lived the situation, not only available loyal customers. Decide stopping through evidence quality and case variation, not a ritual sample number.
Prepare consent, privacy handling, a topic guide and realistic session length. During the interview, establish context, reconstruct one or more episodes, probe decisions and close with reflection. Debrief immediately while situational detail remains fresh.
Analyze continuously enough to improve later interviews but avoid changing the question solely to confirm an emerging story. Keep raw evidence, notes, codes, interpretations and decisions connected so another researcher can inspect the path.
Frame
Define the decision, research question, participant experience and boundaries.
- What must the team learn?
- Why is an interview appropriate?
Recruit
Select information-rich participants across relevant experience and constraints.
- Who has lived the situation?
- Which variation could change it?
Explore
Build rapport, reconstruct real events and probe meaning without leading.
- What happened next?
- What made you say or do that?
Analyze
Compare cases systematically and preserve negative evidence and context.
- What pattern recurs?
- Which case challenges it?
Use
Translate findings into decisions and hypotheses while communicating limits.
- What should change or be tested?
- What cannot this sample establish?
Recruit for relevant variation
Use purposive sampling based on behavior, stage, outcome or constraint. A churn study needs people who left and those who stayed under comparable conditions. A new-category study may need noncustomers and rejecters, not just enthusiastic early adopters.
Screen with factual recent behavior before attitude. Confirm timing and role without revealing the desired story. Overly narrow screeners can manufacture a tidy audience; overly broad recruitment produces shallow interviews because participants lack the experience.
Plan inclusion for disability, language, digital access and assisted use where relevant. Offer accessible locations, formats, interpreters and fair incentives. Avoid recruiting through a gatekeeper who filters out critical voices.
- Decision and question written
- Method fit justified
- Sample matrix covers variation
- Recent experience screened
- Consent and privacy planned
- Guide uses neutral prompts
- Specific episodes reconstructed
- Artifacts handled safely
- Debrief completed
- Codes trace to evidence
- Negative cases examined
- Limits reported
Write a flexible discussion guide
Organize topics in a natural sequence: introduction and consent, participant context, recent episode, relevant trade-offs, reflection and close. Include research questions for the moderator but phrase participant prompts in ordinary language.
Begin broadly enough to avoid supplying an answer. Ask 'walk me through the last time' and follow with what happened, who was involved, what was available and what they did next. Ask for examples when participants offer general beliefs.
Pilot the guide and revise confusing, double-barreled or leading prompts. Do not ask whether people like the feature the team plans to build. Explore the underlying problem, current workaround and consequences before showing an idea.
Depth interview example
The renter study samples different outcomes and constraints, then reconstructs the latest real switching attempt. Bills, devices and sequence cues reduce abstraction. The interviewer learns what the participant did after abandonment, which can reveal offline resolution or competitor choice missing from analytics.
Analysis compares cases rather than collecting memorable quotes. When two renters face the same document request but only one leaves, differences in authority, understanding or alternative routes become testable mechanisms rather than a generic friction claim.
A hypothetical energy-switching service sees renters abandon after starting a comparison, but analytics cannot explain whether price, trust, eligibility or documents create the break.
Interview recent renter completers, abandoners and people who chose another route, varying landlord arrangement, language, bill access and prior switching experience.
Ask participants to walk through the last attempt using their real devices and documents where safe, from trigger and search through the exact point of stopping.
Explore what they expected, which information they trusted, what they did next and how the landlord or bill holder shaped action, without pitching the service.
Build a case-by-stage matrix, attach source notes and contrast people who faced the same document request but made different choices.
Turn the strongest mechanisms into prototype and policy changes, then usability-test the flow and quantify prevalence with behavioral or survey data.
Interview accounts are reconstructed memories. Real artifacts and timeline prompts improve detail but do not turn recall into a complete behavioral record.
Moderate neutrally and listen deeply
Create psychological safety without signaling approval for a preferred answer. Use short neutral probes: tell me more, what happened next, what made that difficult, and how did you decide. Silence often produces detail that a rapid next question would suppress.
Distinguish what the participant did, thought, felt and now interprets. Clarify contradictions respectfully rather than smoothing them away. A mismatch between stated values and action can reveal constraints, not dishonesty.
Avoid teaching, pitching or defending during research. If a participant asks for help, follow the approved safety or support protocol and note how intervention affects the session. In sensitive work, participant wellbeing outranks completing the guide.
Analyze across cases with an evidence trail
Start with session debriefs and factual summaries. Code segments against the research question while allowing new concepts to emerge. Compare by case and stage, not just by theme, so the sequence and combination of conditions remain visible.
Use matrices to contrast outcomes, constraints and participant types. Search for deviant cases that challenge the provisional explanation. A theme repeated by five similar recruits is not automatically broader than one severe barrier found in an excluded path.
Separate evidence, interpretation and implication in the report. Quotes should illustrate a documented pattern with sufficient context, not substitute for analysis. Protect identity by removing unnecessary detail and controlling recordings and transcripts.
Turn findings into testable decisions
A finding states who experienced what, in which context, with supporting evidence and consequence. Link it to a decision and an owner. Prioritize by customer harm, strategic relevance, recurrence in the sample and corroborating evidence, not quote drama.
Translate mechanisms into hypotheses: changing an explanation should improve comprehension, or removing a document dependency should reduce assisted contact. Choose the next method based on the remaining uncertainty, such as usability, survey or experiment.
Report sample, recruitment, dates, mode, exclusions, guide, analysis approach and limits. Do not use percentages from a small purposive sample or claim saturation as proof that no other experience exists.
Limitations and common mistakes
Interview data is shaped by recall, social desirability, interviewer identity, wording and the research setting. Participants can explain an experience compellingly without identifying its true causal mechanism, and future-intent statements often fail to predict action.
Common mistakes include recruiting only customers, asking general hypotheticals, leading toward a roadmap idea, taking notes without source links, counting mentions as prevalence, ignoring negative cases and presenting personas from a handful of conversations.
Depth comes from specificity and analysis, not session length alone. Use interviews to understand worlds and generate mechanisms, then combine them with observation, behavioral evidence and quantitative validation where the decision requires it.
Ask about the last real time, reconstruct the sequence and follow the evidence. Specific memory is more useful than a polished answer about what people usually do.
Frequently asked questions
What is a depth interview?
It is a one-to-one, usually semi-structured qualitative conversation used to explore a participant's experience, context, meaning and reasoning in detail.
How many depth interviews are needed?
There is no universal number. Recruit across relevant variation and continue until the research question is sufficiently explained, important cases are covered and additional sessions add little decision value.
What questions work best in a depth interview?
Open, neutral prompts about recent concrete episodes work well, followed by short probes about sequence, context, evidence, trade-offs and what happened next.
Can interview findings be reported as percentages?
Not usually. Purposive qualitative samples are designed for depth and variation, not population estimates. Quantify prevalence with an appropriate survey or behavioral dataset.
What is the difference between an interview and usability test?
An interview primarily explores reported experience and meaning. A usability test observes participants attempting defined tasks with a product or prototype.
Sources and further reading
- GOV.UK Service Manual: Using In-Depth Interviews ↗Official practical guidance on planning, guiding, conducting and closing individual interviews
- Nielsen Norman Group: User Interviews ↗Publisher guidance on when interviews fit, question design and common moderation errors
- SAGE: Interviewing as Qualitative Research ↗Publisher source for in-depth interviewing, participant experience and interpretive rigor
- GOV.UK Service Manual: Taking Notes and Recording Research ↗Official guidance on observation notes, recordings, consent and evidence handling