A survey is only as good as its questions. Respondents can’t tell you what you meant to ask; they can only answer what you wrote. Survey researchers have studied for decades how wording and structure shift answers, and their findings translate directly into practical rules.
Start from the decision
Before writing a single question, write down the decision the survey will inform and what you would do differently depending on the answer. Every question should earn its place against that. Each extra question costs you respondents who give up before the end.
Key takeaway
If no answer to a question would change what you do, cut the question.
Ask one thing at a time
A double-barrelled question asks about two things at once, such as “How satisfied are you with our pricing and support?” Someone who likes the support but not the pricing has no honest answer, and you can’t interpret the result. Split it into two questions.
Avoid questions that lead
Wording that signals the “right” answer pushes people towards it. Compare “How much did you enjoy our new onboarding?” with “How would you describe your experience of onboarding?” The second leaves room for a negative answer.
Agree–disagree statements carry a known bias: some respondents tend to agree with whatever is put to them. Pew Research Center recommends asking people to choose between balanced alternative statements instead.
| Instead of | Ask |
|---|---|
| How satisfied are you with our pricing and support? | Two questions: one about pricing, one about support |
| How much did you love the new dashboard? | How easy or difficult was the new dashboard to use? |
| Agree or disagree: “The website is easy to use.” | Which comes closer to your view: the website is easy to use, or the website is hard to use? |
| Do you usually read our newsletter? | How many of our last four newsletters did you open? |
Ask about behaviour, recently
People are poor at predicting what they will do and only moderately good at remembering what they did long ago. Questions about specific, recent behaviour — “the last time you…”, “in the past month…” — produce more reliable answers than hypotheticals such as “would you use…?”
Design the answer options carefully
- Make options cover every reasonable answer and not overlap, adding “other” or “don’t know” where appropriate.
- Keep lists short. Pew suggests closed questions generally have no more than four or five options, except for factual items such as age ranges.
- Order matters. In self-completed surveys, people are more likely to pick options near the top, so randomise lists that have no natural order. Keep ordered scales, such as “very easy” to “very difficult”, in order.
- Use balanced scales with as many positive as negative points.
Mind the order of questions
Earlier questions colour later ones. Pew’s experiments show that the same question can produce noticeably different results depending on what was asked just before it. Put general questions before specific ones on the same topic, start with something easy and relevant, and keep demographic questions for the end.
Respondents can only answer what you wrote, not what you meant.
Use open questions deliberately
Open questions capture things you didn’t think to ask, but they take longer to answer and to analyse. A good pattern is to run a small open-ended pilot first, then build closed options from the most common answers, keeping an “other” box for the rest.
Always pretest
Ask three to five people from your audience to complete the survey while thinking aloud. You will find confusing words, missing options and questions that mean different things to different people within minutes. Pew describes pretesting as an essential step, especially for new questions.
Checklist
- One idea per question.
- Neutral wording; avoid agree–disagree.
- Recent, specific behaviour over predictions.
- Short, complete, well-ordered options.
- Pretest before you send.
Related service
Market research and data analytics
Research designed around the decision it needs to support.
