How to write research questions participants actually want to answer

Bad questions cause bad data. Not "slightly fuzzy" data — data that actively misleads, because participants answer the question they think you asked, not the one you wrote down. The good news: the fix is mechanical. Most broken questions share four failure modes, and rewriting them takes under fifteen minutes.

The four failure modes

Leading. "Don't you think the new dashboard is easier?" The respondent now knows which answer earns a nod. They'll give it.

Double-barreled. "Do you like the speed and the layout?" Two questions crammed into one. A "yes" tells you nothing about either.

Vague. "How do you feel about the product?" Feel in what sense? Compared to what? The answer you get is whatever's top-of-mind in that specific second.

Abstract. "What's your ideal onboarding experience?" You're asking a designer's question. Participants aren't designers. They'll either guess or go silent.

The three rules

Past-tense specificity. Ask about something that actually happened, on a specific day, for a specific reason. "When you last exported a report, what were you trying to do with it?" beats "How do you use exports?"

Single focus. One question, one answer. If you're tempted to join two ideas with "and," make them two questions.

Answerable in a sentence. If the honest answer requires a paragraph, the question is too open. You can always follow up. You can't recover a rambling non-answer.

Eight questions, rewritten

1. Bad: "Do you find the new homepage easy to use?" Good: "The last time you opened the homepage, what did you do first?"

2. Bad: "What do you think about our pricing?" Good: "When you saw the price, what did you think it included?"

3. Bad: "Is the product faster now?" Good: "When you ran your last report, how long did it feel like it took?"

4. Bad: "Do you like the new filters and the new sort?" Good: "Which filter did you use most this week, and what were you trying to find?"

5. Bad: "How would you describe your ideal workflow?" Good: "Walk me through how you ran your last export, step by step."

6. Bad: "Would you recommend this to a colleague?" Good: "Has a colleague ever asked you what tool you use for this? What did you say?"

7. Bad: "What features do you want?" Good: "What's the most recent thing you tried to do and couldn't?"

8. Bad: "How do you feel about the onboarding?" Good: "On your first day using this, what was the first thing you got stuck on?"

Run the rewrite

Pull your current interview guide. Mark every question that falls into the four failure modes. Then run each one through the three rules. You'll be surprised how much of your guide shrinks — and how much sharper the remaining questions feel. Fifteen minutes, better data for the next six months.