How to write user research questions that produce useful answers
The hardest part of writing a study is often deciding what you are actually asking. “What should we improve?” invites a wish list. “Tell us about the last time you shared a report with someone outside your team” gives the participant an experience to describe. The second question creates room for details that a product team can investigate.
Good user research questions connect a research objective to something a participant can answer from their own experience. They use plain language, make few assumptions, and ask for one thing at a time. This guide shows how to move from a broad product concern to a short, coherent study.
Separate the team’s question from the participant’s question
A research question might be: what makes account administrators uncertain about inviting a colleague? That is the question your team wants to answer after analysis. It is not necessarily the sentence you should put in front of a participant. Asking it directly presumes uncertainty and asks the respondent to explain a pattern that may not apply to them.
Instead, begin with an event: “Think about the most recent person you invited to your workspace. What did you do before sending the invitation?” Follow with “What, if anything, did you need to check?” Participants who had no concerns can say so, while participants who checked permissions can explain why.
For each study question, write a private note about how the answer will help your research objective. If you cannot explain the connection, remove the question or clarify the objective. This simple edit keeps a focused study from accumulating unrelated requests from across the team.
Replace leading and double questions
| Weak wording | A clearer alternative | Why it helps |
|---|---|---|
| How frustrating was the confusing setup? | How would you describe setting up your workspace? | Does not supply an emotion or judgment |
| Was the invitation flow fast and easy? | How easy or difficult was it to send an invitation? | Asks about one dimension |
| Would you use a better export tool? | Tell us about the last time you exported a report. | Starts with an actual experience |
| Why did you fail to invite someone? | What happened when you tried to invite someone? | Avoids blaming the participant |
Neutral wording does not require vague wording. Name the experience, relevant time period, and action. “What happened the last time you tried to find a report from a previous month?” is both specific and open. It lets the participant describe success, difficulty, or an unexpected workaround without steering them toward the explanation you prefer.
Avoid internal names unless the audience uses them. A team may call something a workspace provisioning flow; a customer may call it setting up an account. If a term is part of the research question, test what the person thinks it means rather than defining it in advance and then asking whether it is clear.
Choose a response type for the information you need
Written responses let participants explain in their own words without recording. Recorded responses are useful when spoken explanation is a better fit for the study. Single select works when one answer should apply; multi select works when several may apply. Scales capture an assessment along a defined dimension, while ranking asks people to order a set of options.
Palette user studies support these formats, including recorded responses. Microstudies support written, single-select, multi-select, scale, and ranking questions. Choose the format that serves the question and the setting. A long ranking exercise can be an awkward interruption when someone is trying to finish a task inside your product.
- Make options distinguishable and include an appropriate way to say none apply.
- Label scale endpoints so participants know what the numbers mean.
- Keep ranking lists short enough to compare thoughtfully.
- Use a written follow-up when a selected option needs an explanation.
A scale score can show how respondents rated an experience. It cannot explain the reason by itself. Pairing a clear rating question with “What most influenced your answer?” can help, provided the follow-up is worth the extra effort you are asking of the participant.
Create a sequence that follows an experience
Start with enough context to understand the participant’s role and recent experience. Move into the event or task that matters to the study. Ask about the choices, constraints, and outcome. Save broader reflections for later so that an early opinion question does not frame every answer that follows.
For a hypothetical invitation study, a compact sequence could ask what prompted the invitation, what the person checked, what happened while sending it, and what they would need to do differently next time. That sequence stays close to one event. A separate question about long-term pricing priorities would dilute the study unless pricing is directly relevant.
GOV.UK’s interview guidance recommends a logical topic flow, starter questions, follow-ups, and a trial of the guide. The same preparation is useful when participants complete a study independently.
Pilot the complete study before collecting responses
Read the study as a participant. Check whether a question assumes experience that the screener did not establish. Check whether a selected answer makes a later question impossible to answer. Ask a pilot participant to explain what they think each question means, rather than simply asking if the wording looks fine.
In Palette, you can work with the agent to draft and revise study questions. Review the wording yourself before launch, especially claims embedded in a prompt and response options that might be missing. Faster drafting gives you more room to improve the study; the draft still needs your understanding of the audience.
For the audience side of study design, read how to recruit user research participants. For a short study inside your product, see the microstudy guide. Explore Palette user studies when your question needs a fuller sequence of answers.