Guide · 8 min read
Survey design: writing questions people actually answer
Scale choice, matrix fatigue, NPS placement and the order effects that bias results.
One idea per question
Double-barrelled questions ('Was support fast and friendly?') produce unusable data. Split them.
Scales
Use a consistent direction and labelled endpoints. 5 points is plenty for most attitudes; reserve 0–10 for NPS.
Matrix questions are efficient but fatigue respondents — cap rows and keep columns short.
Order effects
Ask sensitive or open questions after rapport-building ones. Put NPS early if it's the priority metric, before opinions drift.