Prompt library
Draft unbiased survey questions
Making the model audit its own draft for leading and double-barreled questions catches most survey-design errors in one pass, and showing the rewrite teaches you the pattern. The left-out-questions section forces scope discipline, which is what keeps completion rates up.
Last reviewed July 17, 2026
The prompt
Draft a survey to learn {{goal}} from {{audience}}.
Requirements:
1. 8-12 questions maximum, ordered from easy to sensitive.
2. For each question, state the type (single choice, multi-select, Likert, open text) and why that type fits.
3. No leading, loaded, or double-barreled questions. After drafting, audit your own questions and rewrite any that fail; show the before and after.
4. Include one attention check and place demographic questions last.
5. End with the 2 questions you deliberately left out and the risk each would have introduced (priming, fatigue, social desirability bias).
Keep every question under 20 words.Run in idaptOpens a new chat with the prompt prefilled. Nothing sends until you press send.
Fill in the variables
| Variable | What it is | Example |
|---|---|---|
| {{goal}} | What you want to learn | why trial users do not activate in the first week |
| {{audience}} | Who will answer | people who signed up but sent fewer than 3 messages |