UX Researcher interview questions, with worked answers

How to use these questions

Most research interviews have a portfolio talk at their centre: you walk through two or three studies, and the panel asks why you chose what you chose. Prepare each one the same way: the decision the team faced, the question, the method and why, the participants, what you found, and what happened next. The answers below follow that order.

What a UX researcher is expected to know

• The craft: user researchPlan, run and report research with real users that answers a question the team has to decide. from the question to the finding.
• The methods: interviewsRun interviews and field visits that go beyond opinions to what people actually do and why., usability testsWatch people use a product or prototype, see where they stumble, and say what to change first. and surveysWrite questions that do not lead, pick a sample that fits, and read the answers honestly., each for its own kind of question.
• The numbers: quantitative workSize a finding with surveys, product data or experiments, and know what a sample can show. and enough statisticsKnow when a difference is real, how big a sample must be, and what a survey cannot prove. to size a finding honestly.
• The people: recruiting, consent, incentives and care for the data you collect.

Tools are named in ads side by side and none is a must: Dovetail for notes and synthesis, Maze and UserTesting for remote and unmoderated tests, Lookback for live and recorded sessions, Optimal Workshop for card sorts and tree tests, and FigmaRead and test prototypes in Figma, and add findings where designers will see them. prototypes to test. Panels care more about how you chose a method than about the tool you ran it in.

Questions and methods

How do you turn a vague request into a research question?

What they check: that research starts from a decision.

Short answer: "I ask what the team will decide and when, and what it believes today. A research question is the one thing we need to learn to make that decision, written so a study can answer it: not 'do users like the app', but 'why do new users stop before their first transfer'. I agree it with the product manager and the designer before choosing a method."

The full answer: A request for "some interviews" is a method looking for a question; turn it round before you book anyone.

A weak answer: "I do whatever research the team asks for." Follow-up: "The team wants an answer next week." Say what you would still do and what you would drop.

When do you run user interviews, and when a usability test?

What they check: methods matched to questions.

Short answer: "A user interview explores what people need, do and think, before there is much to show. A usability test watches people try to complete real tasks with a product or prototype, to see where it fails them. Early, open questions get interviews; a design that exists gets tested. Often a session does a little of both."

The full answer: People are poor at predicting their own behaviour, so ask about the past in interviews and watch the present in tests.

A weak answer: "I always do interviews; they are the richest." Follow-up: "You have a finished prototype and two days." Say which you run.

How would you design a survey that does not lead people?

What they check: survey designWrite questions that do not lead, pick a sample that fits, and read the answers honestly. without traps.

Short answer: "I start from what we need to size, usually something interviews found. I keep questions neutral and one idea each, offer balanced answer scales and an honest 'does not apply', put sensitive questions late, and test the draft with a few people before it goes out. Then I check who answered against who we wanted."

The full answer: A survey cannot find what nobody thought to ask; it measures how common a known thing is.

A weak answer: "I ask how much they love the new feature." Follow-up: "Most answers came from your keenest customers." Say what that does to the result.

How many people do you need in a study, and how do you choose the sample?

What they check: honesty about what a sample can show.

Short answer: "It depends on the question. A handful of people in each user group finds most large usability problems; interviews continue until new sessions stop adding new themes. To say how common something is, I need a survey or product data with a sample sized for it. I choose participants for the spread of behaviour we care about, not for convenience."

The full answer: Small studies find problems and reasons; large ones measure. Do not quote a share from eight interviews.

A weak answer: "Five users is always enough." Follow-up: "Your product has three very different user groups." Say how the plan changes.

Recruiting, consent and care

How do you recruit participants who are not just your keenest customers?

What they check: a fair picture of the users.

Short answer: "I write a short screener on behaviour, not on opinions, and recruit from more than one channel: the product itself, a panel or recruiting service, and customer support for people who had problems. I track who we have talked to over the year so the same voices do not decide everything, and I include people who left."

The full answer: The users who never reply to emails are often the ones the product is losing.

A weak answer: "I post in our community forum." Follow-up: "You need people who tried the product and gave up." Say where you find them.

What does informed consent look like in your studies?

What they check: care for participants and their data.

Short answer: "Informed consent means the participant knows, before we start, what the study is for, what we record, who sees it, how long we keep it and that they can stop at any time, and agrees to it. Under the EU GDPR that consent must be freely given and as easy to withdraw as to give. I keep recordings only as long as the study needs and remove names from notes."

The full answer: Each country has its own rules; the privacy team says which apply and how long data may be kept.

A weak answer: "They tick a box in the booking tool." Follow-up: "A participant asks you to delete their recording a week later." Say what you do.

How do you set an incentive, and when do you not pay one?

What they check: fairness without bias.

Short answer: "An incentive thanks people for their time, so I size it to the session's length and the group's usual rates, and pay it whether or not the session went as planned. I avoid amounts so high that people say what they think we want. Some groups, such as employees or public servants, may not be allowed to accept one; then I offer a donation or nothing and say so."

The full answer: Ask finance early how incentives are paid and taxed; it decides how fast you can recruit.

A weak answer: "Only people who finish the full test get paid." Follow-up: "A participant leaves after ten minutes because the prototype broke." Say what you pay.

How do you keep your own bias out of a study?

What they check: awareness of how research goes wrong.

Short answer: "I write down what I expect before the study, use a guide with open questions, never explain the design to the participant, and let silences run. In analysis I look for evidence against my expectation as hard as for it, and ask a second person to code part of the notes. Confirmation bias is the one I watch for most."

The full answer: The team's bias matters too: invite them to watch, so they see the problem with their own eyes.

A weak answer: "I am objective." Follow-up: "The designer who made the prototype wants to moderate." Say how you handle it.

Synthesis and outputs

Walk us through your synthesis after ten interviews.

What they check: turning notes into findings.

Short answer: "Synthesis is turning what people said and did into a few findings with evidence. I tag notes and quotes as I go, cluster them into themes with a colleague, check each theme against how many participants it comes from, and write each finding as an insight: what happens, why, and what it means for the decision. Then I pick the clips that show it best."

The full answer: A finding without a 'so what' gets nodded at and forgotten.

A weak answer: "I write up what each user said." Follow-up: "Two themes contradict each other." Say how you report it.

When is a persona useful, and when does it mislead?

What they check: artefacts used with judgement.

Short answer: "A persona is a short portrait of a group of users built from research: their goals, their situation and how they behave. It helps when it is based on real studies and the team uses it to choose between options. It misleads when it is invented, describes age and hobbies instead of behaviour, or stays the same for years while the users change."

The full answer: If nobody can say which study a persona came from, retire it.

A weak answer: "We make personas at the start of every project." Follow-up: "Marketing has segments that do not match your personas." Say what you do.

How do you build a journey map that a team actually uses?

What they check: maps that lead to work.

Short answer: "A journey map shows the steps a person takes to reach a goal, across channels, with what they do, think and feel at each step and where it goes wrong. I build it from research, with the people who own each step in the room, and end with the few moments to fix first and who owns them."

The full answer: A map on the wall that names no owner becomes decoration within a month.

A weak answer: "I make it look good and share it." Follow-up: "The worst step belongs to another department." Say how you get it fixed.

How do you combine quantitative data with qualitative findings?

What they check: quantitativeSize a finding with surveys, product data or experiments, and know what a sample can show. and qualitative work together.

Short answer: "Product data and surveys tell me what happens and how often; interviews and tests tell me why. I start from whichever raises the question: a drop in the funnel sends me to interviews, an interview theme sends me to the data to size it. An A/B test then checks whether the fix works for many people, with the analyst setting it up."

The full answer: Agree the success measure with the data analyst before the test starts, not after the result.

A weak answer: "Numbers are not my area." Follow-up: "The A/B test shows no difference, but users loved the change in tests." Say what you conclude.

Influence, tools and AI

A stakeholder disagrees with your findings. What do you do?

What they check: influenceAgree research questions with product, design and engineering leads before a study starts. without a fight.

Short answer: "I first find out what they disagree with: the method, the sample or the conclusion. I show the evidence, clips rather than slides, and say plainly what the study can and cannot show. If their doubt is fair, I propose a small follow-up to settle it. I involve stakeholders before the study next time, so the question is theirs too."

The full answer: Watching one real user struggle convinces faster than any report.

A weak answer: "The data speaks for itself." Follow-up: "The stakeholder is the head of product." Say what changes.

How would you set up a research repository so findings are not lost?

What they check: research that builds up over years.

Short answer: "A research repository keeps studies, findings and evidence in one searchable place, tagged by product area, user group and theme. I start small, with the last year's studies and a few agreed tags, store findings rather than raw notes, and remove personal data. Teams check it before asking for a new study. Tools such as Dovetail help, but the habit matters more."

The full answer: Name an owner and review the tags twice a year, or the repository fills with noise.

A weak answer: "Reports are on the shared drive." Follow-up: "A team repeats a study you ran last year." Say how you prevent it.

Where does AI help in research, and where do you not trust it?

What they check: sober use of new tools.

Short answer: "AI tools save time on transcripts, first-pass tagging and finding quotes across many sessions, and several research platforms now include them. I check every summary against the recording, because they smooth over contradictions and miss what people did rather than said. I never feed participant data into a tool the privacy team has not approved."

The full answer: Synthetic users invented by a model are no substitute for real ones; use them, if at all, to draft a guide.

A weak answer: "AI can write the findings now." Follow-up: "The AI summary says users liked the feature; the clips show confusion." Say which you report.

Senior rounds

As a senior UX researcher, how would you help a junior UX researcher run their first study alone?

What they check: building the research team.

Short answer: "I let them own a small, clear study: we agree the question together, I review their guide and screener, sit in on the first session and then step back. We do the synthesis side by side and they present the findings themselves. I share why I would choose differently, not only what to change."

The full answer: A senior UX researcher or UX research manager is judged by how many good studies the team runs without them in the room.

A weak answer: "Juniors take notes in my sessions." Follow-up: "Their findings overstate what eight interviews show." Say how you give the feedback.

Your CV for these questions

Every study on your CV invites the questions above: why that method, who took part, what you found and what changed. Give each study a line a research lead would ask about:

Hold your CV against a UX researcher job with the free score; rehearse first the study that answers the area it finds thin, from the question to what the team did next.