UX Research · Step 3 · 6 min read
Asking good questions
The words of a question decide what kind of answer you can get. Why opinions and predictions are weak data, how to ask about specific moments, the six weak question patterns and their rewrites, and how to stop pitching your idea in the middle of a research interview.
People will answer any question you ask. That's the problem. Ask whether they'd use a feature and they'll say yes, kindly. Ask whether a task was frustrating and they'll agree it was. None of it is a lie, and none of it is data you can design on. A good question doesn't ask people to judge, predict or agree. It asks them to remember something that really happened, and then it gets out of the way.
What you'll be able to do after reading:
- Tell weak data (opinions, predictions, generalisations) from strong data (specific events)
- Spot six weak question patterns and rewrite each one
- Use open questions to explore and closed ones only for details
- Ask about a product idea without pitching it
1. What an interview can and can't tell you
Jakob Nielsen is blunt about it: the critical failing of interviews is asking people to remember the past or predict the future, and both kinds of answer are "extraordinarily weak and often misleading". People can't remember the details of how they used something, and they tidy what they do remember so it sounds more logical. And they can't predict how they'd use something they haven't seen (NN/g, Nielsen, Interviewing Users, 2010).
He gives an uncomfortable example: in many studies, facilitators asked users about features that didn't exist, slipped in on purpose, and the users gave plenty of feedback on them.
So interviews aren't for deciding between a dropdown and radio buttons (that's a usability test). They're for understanding people: their goals, their problems, how they think about the task. And within that, some answers are much stronger than others.
| Weaker | Stronger |
|---|---|
| Opinion: "I'd love that" | What they did: "Last month I…" |
| Prediction: "I'd use it weekly" | What they already pay, in money or time, to solve it |
| Generalisation: "I usually…" | One specific time, with the details |
| Your words, agreed with | Their words, unprompted |
2. Ask about one specific moment
The fix is the one you met in Short user interviews: ask about a specific event, because remembering a particular incident nudges memory and brings back real details (NN/g, User Interviews 101, 2023).
You can make it sharper. Nielsen recommends the critical incident approach for exploratory interviews: ask people to recall a specific time when something was particularly difficult, or when it worked particularly well. These extreme cases are more vivid in memory and carry the details you need (Nielsen, 2010).
- "Tell me about the last time you…" (the most recent, so the freshest)
- "Tell me about a time when it went wrong." (the critical incident)
- "Tell me about a time it worked really well." (the other extreme, often forgotten)
Then stay on that one event. When a participant drifts back to "usually I…", bring them back gently: "And that time in March, what happened?"
3. Six weak patterns, rewritten
- Futureasks for a predictionWould you use an app that splits bills for you?Tell me about the last time you split a bill. What did you use?
- Genericgets the tidy versionHow do you usually handle shared bills?Think of the electricity bill last month. What happened, step by step?
- Leadingputs the answer inWasn't it frustrating to chase your flatmates?How did it go when you asked them to pay?
- Closeda yes or a no, and silenceDid you use a spreadsheet?How do you keep track of who owes what?
- Two in onewhich half did they answer?How do you split the bills and remind people to pay?How do you split them? (Then, later:) And when someone forgets?
- Pitchasks for a complimentWe're building a reminder feature. What do you think?What have you tried so far to get people to pay on time?
Open and closed. Open questions let people choose what to share and in how much detail; closed ones limit them to a short answer. Closed questions aren't banned: they're useful for clarifying a detail. But NN/g notes they eliminate surprises: "what you expect is what you get". A handy test: leading closed questions often start with "did", "was" or "is"; open ones with "how" or "what" (NN/g, Rosala, Open-Ended vs. Closed Questions, 2024).
Leading. A leading question puts the answer inside it, and people tend to repeat the interviewer's words back. Amy Schade's example from a usability session: "What do you think this button does?" made the user realise that the text she was pointing at was a working link, which changed how she behaved for the rest of the session (NN/g, Schade, Leading Questions, 2017). Neutral wording names the topic without naming a feeling: "How did it go when…?" instead of "Wasn't it frustrating when…?".
Two in one. "How do you split bills and remind people?" gets an answer to whichever half was easier. Ask one thing, then the next.
4. Don't pitch
The hardest habit to break is talking about your idea. The moment you describe it, the interview becomes a sales meeting, and kind people respond to sales meetings with compliments. Rob Fitzpatrick's The Mom Test is a whole short book about this: how to get honest learning from conversations even when everyone is being nice to you (momtestbook.com).
What to do instead:
- Ask about the problem, not the solution. "What have you tried so far?" tells you whether the problem is real enough that they already spend time or money on it.
- Treat compliments as noise. "That sounds great" isn't a finding. Steer back: "When did this last come up for you?"
- Save the concept for later. If you must show an idea, do it at the end, after you've heard their story, and then watch what they do with it rather than what they say about it (that's usability testing).
5. Common mistakes
- "Would you…?" and "How likely are you to…?" in an interview.
- Asking "usually" and accepting the tidy, average version.
- Putting a feeling into the question: frustrating, easy, annoying.
- Describing the solution, then asking what they think of it.
- Firing closed questions in a row, like a form read aloud.
- Asking people what features they want, and treating the list as requirements.
6. Vocabulary
| Term | Meaning |
|---|---|
| Open question | Lets the participant choose what to share and in how much detail |
| Closed question | Limits the answer to a short reply or a set of options |
| Leading question | Contains or suggests the answer you expect |
| Critical incident | A specific, memorable time when something went very wrong or very right |
| Double-barrelled question | Two questions in one sentence |
| Pitching | Describing your idea during research, which turns answers into politeness |
7. Try it (no tools needed)
- Find five questions someone on your team wants to ask users (or write the five you'd ask about a food-delivery app).
- Label each one: opinion, prediction, generalisation, leading, closed, two in one, or fine.
- Rewrite every one that isn't fine as a question about one specific past moment.
Reflection questions: Which rewrite lost something the original was trying to learn? How would you learn that part without asking for an opinion?
8. If you already practise
- Listen back to one of your recordings and count your questions that start with "did", "is" or "would". Then count the ones that start with "how", "what" or "tell me".
- When you must ask about a concept, plant a feature that doesn't exist alongside the real ones. If it gets as much praise, so does the praise.
- Write a "weak question" bank with your team's real examples and their rewrites. It trains new interviewers faster than any rule.
Checklist
Learn more
- NN/g: Interviewing Users (Nielsen, 2010) · Open-Ended vs. Closed Questions in User Research (Rosala, 2024) · Avoid Leading Questions to Get Better Insights from Participants (Schade, 2017) · User Interviews 101 (Rosala and Pernice, 2023)
Books, courses and videos to go deeper
- The Mom TestRob FitzpatrickBook · 2013 (external site)
- Interviewing UsersSteve PortigalBook · 2nd edition (Rosenfeld Media, 2023) (Amazon)
Book links go to Amazon. As an Amazon Associate I earn from qualifying purchases, at no extra cost to you.
Related
Short user interviews · The interview guide · Running the session and listening · Bias in surveys · Writing survey questions · User needs and jobs to be done · Interview guide