A bad survey does not fail loudly. It returns tidy percentages built on questions that led the respondent, asked two things at once, or measured a construct nobody defined. These prompts catch that at the drafting stage, when it costs nothing to fix.
What's Included
- Construct per question: what each item is actually measuring.
- Scale selection matched to the construct rather than to habit.
- Bias screen for leading, double-barreled, and ambiguous wording.
- Length discipline so completion does not collapse halfway.
How To Use
- State what decision the results will inform.
- Draft the instrument and run the bias screen on every item.
- Pilot with five people and fix what they misread.
- Collect responses in one place and analyze as they land.
Ground it first in user interview prompts, then structure the analysis with the research summary report.