By the Wavefield Research team · Published Sep 30, 2026
Good survey questions are mostly reused, not invented: a satisfaction rating, a recommend scale, an ease question, a reason follow-up, and a use-case block around them. Below are 100+ survey question examples organized by study type, each set with the craft note that separates a working question from a lookalike that produces junk.
Where a use case deserves a full treatment — option sets, analysis rules, worked examples — its section links to the deep bank. The wording discipline underneath all of them is catalogued in bad survey questions and the format menu in survey question types; Pew Research Center's Writing Survey Questions is the standing reference on how much wording moves answers.
Ask the headline satisfaction measure before the detail questions so their wording cannot contaminate it, and always pair a rating with one open follow-up for the reason. Full bank with option sets: customer feedback surveys.
Behavioral questions (what people did) outperform speculative ones (what they would do); when a study must be speculative — concept tests, price checks — keep arms randomized and read differences, not absolutes. Full bank with option sets: app survey questions.
Employee surveys live or die on perceived anonymity: state the reporting threshold (no group smaller than five, for example) and honor it, or the honest answers stop coming. Full bank with option sets: employee engagement surveys.
Field within 48 hours while memories are specific, keep it under ten questions, and make the most-valuable-session question a closed list — open recall of session names produces uncodeable soup.
The question that predicts real impact is confidence applying the skill, not satisfaction with the day — satisfaction measures catering; application confidence measures training.
The task-completion pair — what did you come to do, and did you manage it — beats any satisfaction scale for diagnosing a website; run it on-site with one or two questions, not as a long form.
Unaided questions must come before aided ones — once you show a brand list, spontaneous recall is contaminated for the rest of the survey and cannot be recovered. Full bank with option sets: brand awareness surveys.
Every option set here is secretly a weighting target: keep bands census-aligned, give every sensitive item a Prefer not to say, and ask only what the analysis plan will actually use. Full bank with option sets: demographic survey questions.
Candidate and policy questions need balanced stems and rotated orders, favorability asked for every candidate rather than just yours, and screeners fielded before a single opinion question. Full bank with option sets: political polling.
Keep open ends optional, tie each to a specific episode rather than a general opinion, and plan the codeframe before fielding — an open end nobody will code is a question that should not exist. Full bank with option sets: open-ended survey questions.
Swap the subject, keep the structure. These sample survey questions work because of their bones — one idea per question, neutral stems, options that give every respondent exactly one honest home. Change the bracketed subject freely; the moment you also change the structure (“How satisfied are you with our excellent new dashboard?”), you have written a leading question wearing a good question's clothes.
Keep scales consistent within a survey. If satisfaction runs Very dissatisfied to Very satisfied in question 3, every later satisfaction item uses the same five points in the same direction. Mixed scale lengths and flipped directions are how clean-looking data quietly stops being comparable.
Order by contamination risk. Unaided before aided, headline measures before diagnostics, demographics last, screeners first. The sequence rules and the instrument thinking behind them are worked through in what is a questionnaire.
A versatile starter set: 1. How satisfied were you with [experience]? 2. How likely are you to recommend us? 3. What was the main reason for your score? 4. How easy was it to [complete the task]? 5. Which of the following have you used? 6. What almost stopped you? 7. How does it compare to what you expected? 8. What should we improve first? 9. Which age group are you in? 10. Anything else you'd like us to know? Swap the bracketed pieces for your context and you have a working feedback survey.
The most common across all study types: a satisfaction rating, a likelihood-to-recommend scale, an ease or effort question, a frequency-of-use question, a most-important-factor question, one or two open-ended reason questions, and a short demographic block of age, gender, and region. Most real surveys are built from this same core with use-case-specific questions around it.
One idea per question, words every respondent reads the same way, response options that cover every honest answer exactly once, and a neutral stem that does not lean toward any answer. The craft is mostly about what to avoid — leading wording, double-barreled constructions, overlapping ranges, and unbalanced scales — and every one of those faults has a worked fix in our bad survey questions guide.
As few as the decision needs — for most feedback studies that is 8 to 12 questions and about five minutes. Completion quality decays measurably past ten minutes, and every marginal question taxes the answers after it. The working test: name the report table each question feeds; a question that feeds none gets cut.
Mostly closed, with a few deliberate open ends. Closed questions produce countable, comparable data and respect respondent time; open-ended questions collect reasons and language you could not anticipate. The standard pattern is closed questions for everything you can enumerate, plus one open follow-up behind the headline rating and one optional catch-all at the end.
The specialty banks: multiple choice · open-ended · ranking · qualitative · dichotomous · gender · income · screening
Describe your study and the agent drafts the whole instrument — questions, options, logic, quotas — for you to review and field. From $99 per study.
Start a study