How to write good survey questions (with examples)
A good survey question asks about one thing, in plain language, with answer options that are exhaustive and mutually exclusive — and without nudging the respondent toward any particular answer. The most common failures are double-barreled questions (two things at once), leading wording, vague quantifiers like "often," and answer scales that overlap or leave gaps. This guide gives you ten rules, before/after rewrites for each, and a checklist to run before you hit publish.
What makes a survey question good?
A good survey question produces data you can trust. That means every respondent interprets it the same way, every respondent can find an answer option that fits them, and nothing in the wording pushes them toward one answer over another. If two reasonable people could read your question differently, the data it produces is noise.
Write for the busiest, least-motivated person on your list. If they can answer accurately in five seconds, the question is good.
The 10 rules of good survey questions
1. Ask about one thing at a time
Double-barreled questions are the most common mistake. "How satisfied are you with our pricing and support?" forces someone who loves your support but hates your pricing to pick a meaningless middle answer.
- Bad: How satisfied are you with our pricing and support?
- Good: How satisfied are you with our pricing? (then a separate question for support)
2. Use plain, concrete language
Jargon, acronyms, and abstract nouns make respondents guess. Replace internal vocabulary with the words your audience actually uses.
- Bad: How would you rate the efficacy of our omnichannel engagement touchpoints?
- Good: How easy was it to get help from us by email, chat, or phone?
3. Avoid leading and loaded wording
Any phrase that signals the "right" answer — "our award-winning support," "don't you agree" — inflates positive responses and quietly destroys your data. Strip out adjectives and assumptions.
- Bad: How much did you enjoy our beautiful new dashboard?
- Good: How would you rate the new dashboard?
4. Replace vague quantifiers with concrete ranges
"Often," "regularly," and "sometimes" mean different things to different people. Anchor frequency questions in real numbers and time periods.
- Bad: Do you use the app often?
- Good: In the past 7 days, how many days did you use the app? (0, 1–2, 3–4, 5–6, 7)
5. Make answer options exhaustive and mutually exclusive
Every respondent must have exactly one option that fits. Ranges like "1–5" and "5–10" overlap at 5; a list of departments with no "Other" leaves some people stranded. Check both directions: no gaps, no overlaps.
6. Don't assume behavior or knowledge
"Which feature of the mobile app do you like most?" assumes the person has used the mobile app. Add a screening question first, or include a "I haven't used it" option — otherwise you collect fabricated answers.
7. Keep scales balanced
A scale with three positive options and one negative option (Excellent / Great / Good / Poor) biases results upward. Offer the same number of positive and negative points around a neutral midpoint.
8. Ask about recent, specific experiences
Memory fades fast. "How was your support experience in the past year?" produces vague impressions; "Think about your most recent support conversation — how well did we resolve your issue?" produces usable data.
9. Make sensitive questions optional and explain why you ask
Demographics, income, and health questions deserve a short justification ("to make sure we hear from a range of customers"), a "Prefer not to say" option, and a non-required setting. Forced sensitive questions cause drop-off and false answers.
10. Pilot the survey before launch
Send the draft to three to five people who resemble your audience and ask them to think out loud while answering. You will catch ambiguous wording, missing options, and broken logic that you are too close to see.
Common mistakes: before and after
| Bad question | What's wrong | Better version |
|---|---|---|
| How satisfied are you with our fast, friendly support? | Leading — asserts support is fast and friendly | How satisfied are you with our support? |
| Do you exercise and eat healthy? | Double-barreled | How many days per week do you exercise? |
| Don't you think the checkout is easy? | Leading — invites agreement | How easy or difficult was checkout? (Very difficult → Very easy) |
| How often do you shop online? (Rarely / Sometimes / Often) | Vague quantifiers | In the past 30 days, how many online purchases did you make? (0 / 1–2 / 3–5 / 6+) |
| What is your age? (18–25 / 25–35 / 35–50) | Overlapping ranges, missing under 18 and over 50 | What is your age? (Under 18 / 18–24 / 25–34 / 35–49 / 50+ / Prefer not to say) |
| Which of our integrations do you use? | Assumes usage, no opt-out | Do you use any integrations? If yes, which ones? (include None) |
How should you order survey questions?
- 1Start with an easy, engaging question that confirms the survey is relevant — never demographics.
- 2Group related questions together so respondents stay in one mental context.
- 3Put the questions you most need answered in the first half, before fatigue sets in.
- 4Place open-ended questions sparingly and after the related closed question ("Why did you choose that score?").
- 5End with demographics and optional sensitive questions, each with a Prefer-not-to-say option.
Read every question aloud and ask: could someone answer this dishonestly without noticing? If a question contains "and," an adjective, or a word like "often," rewrite it before launch.
Pre-launch checklist
- Every question asks exactly one thing
- No adjectives or assumptions baked into the wording
- Frequency questions use concrete numbers and time frames
- Answer options cover everyone, with no overlaps
- Scales are balanced around a neutral midpoint
- Sensitive questions are optional with Prefer not to say
- The survey was piloted with 3–5 real people
- Total time to complete is under 10 minutes
Once your questions are clean, the tooling part is easy. In a builder like Formkii you can mix multiple choice, rating scales, NPS, ranking, and open text in one survey, add skip logic so people only see relevant questions, and pilot with a shareable link before the real launch.
Frequently asked questions
What is a double-barreled survey question?
A double-barreled question asks about two things at once, such as "How satisfied are you with our price and quality?" Respondents who feel differently about each part cannot answer accurately, so the data is unusable. The fix is to split it into two separate questions.
How many questions should a survey have?
Most surveys work best with 5 to 10 focused questions, completable in under 5 minutes. Longer surveys are fine for engaged audiences like employees, but completion drops as length grows, so cut any question you don't have a plan to act on.
Should survey questions be required or optional?
Require only the questions that are essential to your goal, and always make sensitive questions like demographics optional with a "Prefer not to say" choice. Forcing answers increases abandonment and produces fabricated responses.
What is the best first question for a survey?
Start with an easy, relevant, closed-ended question tied to the survey's stated topic — for example, an overall satisfaction rating. Opening with demographics or open text increases early drop-off.