How to Conduct User Interviews: A Practical Guide

On this page

• What Is a User Interview and When Should You Use One?
• How to write a discussion guide that actually works
• Facilitation techniques that unlock honest answers
• Recruiting the right participants
• Common mistakes and how to avoid them
• What to do after the interviews: a brief synthesis primer
• Frequently asked questions

What Is a User Interview and When Should You Use One?

A user interview is a qualitative, conversational research method for exploring attitudes, motivations, and behaviours. You speak directly with people to understand why they think, feel, or act as they do — not to measure how many people share a preference or to observe whether a product works.

That distinction matters. User interviews differ from usability testing, which is observational: you watch someone attempt a task to assess whether a design functions. Interviews are conversation. They differ from surveys in the opposite direction: surveys offer scale and quantitative weight; interviews offer depth and nuance. You rarely replace one with the other.

User interviews are the right tool when you are in early discovery, mapping a user journey, or trying to understand the problem space before you build anything. They answer questions like: What does this person’s working day actually look like? What prompted them to seek a solution? What trade-offs are they managing?

They are the wrong tool when you need statistically representative data, need to measure task success rates, or need to validate that one design performs better than another. Reaching for interviews in those situations produces misleading results, not better insight.

Within a broader UX research methods toolkit, interviews sit at the exploratory, generative end. Use them to open questions, not to close them.

---

The most common reason user interviews produce thin insight is that they start without a clear research question. A business question — “should we add feature X?” — is not a research question. A research question is: “How do people currently manage this task, and where does effort accumulate?” One is a decision; the other is something you can actually go and learn.

Before you recruit a single participant, define your scope explicitly. Write down what you are studying and what you are not. Scope creep inside a discussion guide produces sessions that feel broad but reveal little.

On sample size: for qualitative research, 5–8 participants per distinct user segment is a practical rule of thumb. You are not aiming for statistical significance; you are aiming for thematic saturation — the point at which new sessions surface familiar themes rather than new ones. If you have two clearly different user segments, plan 5–8 per segment.

Keep the room lean. One facilitator and one note-taker is the minimum and often the ideal. A panel of observers — whether in the room or lurking on a video call — can create subtle pressure that makes participants perform rather than disclose.

Build time into your plan for synthesis between sessions. Rushing from one interview to the next without a debrief erodes the quality of your analysis. See our thinking on research cadence and planning for a fuller treatment.

---

How to write a discussion guide that actually works

A discussion guide is not a script. It is a structured scaffold that keeps your sessions comparable across participants without preventing the facilitator from following a genuinely interesting thread. Think of it as a map, not a set of train tracks.

A well-structured guide moves through five stages:

1. Warm-up — open, low-stakes questions about the participant’s role, context, and routine. The goal is rapport, not data. These questions settle participants into conversation and signal that you want them to talk freely.
2. Context-setting — orient the participant to the topic area without leading them towards your hypotheses.
3. Core topic exploration — the bulk of the session.
4. Task walkthrough — only if you are asking participants to interact with a prototype or artefact.
5. Wrap-up — a chance for participants to add anything they felt was not covered, and to close warmly.

For core questions, always favour behavioural prompts over attitudinal ones. “Tell me about the last time you had to do X” retrieves a real, specific episode. “Do you prefer X or Y?” invites opinion performance rather than honest reflection.

Build these five question types into your guide:

• Grand-tour questions — broad openers (“Walk me through a typical week of managing this”)
• Example-eliciting questions — “Can you give me a specific example of that?”
• Contrast questions — “How does that compare to how you did it before?”
• Devil’s advocate questions — “Some people find that approach time-consuming — does that match your experience?”
• Silent probes — planned pauses; write them into the guide as a reminder to wait

Timebox each section inside the guide document itself. A 60-minute interview needs a 45-minute guide. The gap is not wasted time — it is space for tangents, which is often where the most useful material lives.

---

Facilitation techniques that unlock honest answers

Active listening is a set of deliberate behaviours, not a personality trait. Mirror the participant’s last few words as a question to invite them to continue. Nod steadily regardless of whether the answer confirms or contradicts your hypothesis — confirmation bias leaks through body language and vocal tone. After a participant pauses, count silently to three before speaking. Most interviewers jump in too early, and the most revealing additions come in the silence after the first answer.

Use follow-up loops consistently. “Tell me more about that” and “What did you mean by that?” are two of the most productive phrases in user research. Never accept the first answer as complete; it is almost always a summary.

Practise laddering for motivational depth. When a participant states something they value or find difficult, ask “Why does that matter to you?” Follow the thread tactfully, two or three levels deep. The surface complaint rarely explains the underlying behaviour.

Managing participants who talk extensively is a common challenge. Avoid abrupt interruptions, which damage rapport. Instead, use a natural section transition as a redirect: “That’s really useful context — I want to make sure we cover [next topic] before we run out of time, so let me ask you about…”

For note-takers: capture verbatim quotes and non-verbal cues, and note timestamps for key moments. Do not paraphrase or summarise in real time — that interpretation belongs in synthesis, not in the raw record.

Always record sessions where legally permitted and consent has been given. Explain clearly how recordings will be used and stored before the session begins, not buried in a consent form. Good recording practice protects both participants and the validity of your data. For more on moving this material towards impact, see our piece on turning research into impact.

---

Recruiting the right participants

Recruitment shapes everything. Interviews with the wrong participants produce coherent, confident, entirely misleading insight.

Write a screener before you look for participants. A screener defines inclusion and exclusion criteria — and the strongest screeners use behavioural criteria, not demographics alone. “Has made a purchase decision involving three or more stakeholders in the past six months” is more useful than “works in procurement.” Demographics describe people; behaviours describe relevant experience.

Common sourcing channels each carry trade-offs:

• Your own CRM or customer list — high relevance, but risks participants who are either unusually loyal or unusually dissatisfied
• Panel providers — fast and scalable, but participants may be professional respondents whose answers are rehearsed
• Online communities and social groups — good access to specific contexts; requires careful screening
• Guerrilla recruitment (intercepting people in relevant locations) — quick for early discovery; less control over criteria

Over-recruit by roughly 20% to account for no-shows. Send reminder messages at 48 hours and again at 2 hours before the session.

Build purposive variation into your sample. Seek difference in experience level, context of use, and life situation. This is not the same as demographic quotas — it is about ensuring your sample is not accidentally homogeneous in ways that blind you to important variation. More on logistics is covered in our participant recruitment and panels guidance.

---

Common mistakes and how to avoid them

Most errors in user interview practice fall into a small number of recurring patterns.

Asking hypothetical questions. “Would you use a feature that did X?” sounds useful. It produces unreliable data. People are poor predictors of their own future behaviour. Replace every hypothetical with a past-behaviour equivalent: “Tell me about the last time you needed to do X.”

Leading questions. “How frustrating was it when the system crashed?” embeds the assumption that the participant found it frustrating. A neutral framing — “What happened when the system crashed?” — leaves room for a more accurate answer.

Skipping the pilot. Even one internal run-through with a colleague surfaces questions that are confusing when spoken aloud, sections that run over time, and transitions that feel abrupt. It takes 45 minutes and is rarely wasted.

Delaying synthesis. The further you get from a session, the more detail fades. Run a short debrief — even bullet notes — within 24 hours while observations are fresh.

Observer interference. Observers watching via video or through a viewing window can send messages mid-session that reach the facilitator. Establish a clear protocol before fieldwork: observers observe; they do not direct.

Reporting small samples as percentages. Writing “50% of participants said X” when you interviewed six people implies a precision that qualitative data cannot support. Use frequencies and qualitative framing: “Three of six participants described…” or “Several participants mentioned…”

---

What to do after the interviews: a brief synthesis primer

Synthesis begins before the last interview ends. After each session, the facilitator and note-taker should spend ten to fifteen minutes aligning on their top observations, surprises, and open questions. This immediate debrief preserves detail and begins to surface patterns early.

Across the full set of sessions, affinity mapping is the most common approach to thematic clustering. You move raw observations — typically quotes and paraphrased notes — into emergent groupings, then label those groupings as themes. The goal is not to confirm what you expected to find but to let the data organise itself.

Moving from raw notes to actionable insight typically involves three artefacts:

• Quote banks — organised collections of verbatim quotes by theme, useful for evidencing findings with product teams
• Opportunity statements — framed as “How might we…” or “People need a way to…”, these translate observations into design-relevant problems
• Journey sketches — rough maps of the sequence of events, decisions, and emotional states participants described

Match your output format to your audience. An executive summary should fit a single page with a clear “so what.” A product team benefits more from annotated quotes and specific opportunity statements than from polished slides.

Close the loop explicitly. Return to the research question you defined at the outset and answer it directly. If the research changed or extended that question — which is common — note why.

For detailed guidance on synthesis methods and workflows, see our pieces on synthesis and analysis workflows and product discovery.

---

Frequently asked questions

How many user interviews do I need to conduct?

For a single, well-defined user segment, 5–8 interviews typically reach thematic saturation in qualitative research. Add another round of 5 per additional distinct segment. The practical signal is repetition: when new sessions surface the same themes rather than genuinely new information, you have enough. More sessions beyond that point rarely change the findings but always increase the time to synthesis.

What is a discussion guide in user research?

A discussion guide is a structured but flexible document that outlines the topics, questions, and timing for a user interview session. Unlike a survey, it is a scaffold for conversation rather than a rigid script. It keeps sessions comparable across participants while giving the facilitator space to follow threads that emerge naturally. A good guide is short enough to complete comfortably within the session time and written in the spoken register you will actually use.

How long should a user interview be?

Should I record user interviews?

Yes, where legally permitted and where the participant has given explicit informed consent before the session begins. Recordings allow you to retrieve verbatim quotes accurately and catch detail the note-taker missed in the moment. Store recordings securely, anonymise where required by your data-handling obligations, and delete them according to the retention policy you stated when you collected consent.

What is the difference between a user interview and a usability test?

User interviews explore attitudes, motivations, and past behaviours through conversation. Usability tests observe participants attempting specific tasks with a product or prototype to evaluate whether the design functions as intended. Both are valuable, but they answer different questions. Interviews tell you what people think, feel, and do in their own context. Usability tests tell you whether people can successfully complete a task with a specific interface. Choosing between them — or combining them — depends on what you need to learn.