You learn a practical length rule, the cost of every extra question, and an objective criterion for cutting.
For experience surveys, five to seven questions are usually enough: one metric, such as NPS or CSAT, the reason behind the score, one or two context questions, and at most one contact request. Long questionnaires increase abandonment and rushed answers on the final questions. The inclusion rule is objective: if the answer changes no decision, the question stays out. Deep-dive studies can run longer — but then state the duration right in the invitation.
Post-purchase in four questions: a 0–10 score (NPS), “what is the main reason for your score?” (open), “what influenced your experience the most?” (single choice), and “may we contact you?” (yes or no). Under two minutes, and every answer feeds an action.
Five to seven questions cover most decisions; every extra one costs completions.
Start from the decision: if the answer changes no action, the question is out.
One well-placed open question beats three generic closed ones.
The recommendations follow the Pew Research Center guidelines on writing survey questions and the practice behind NPSLab templates: one metric, one reason, context, and at most one contact request. There is no magic number — only the minimum that supports the decision.
Fewer questions reduce abandonment but do not fix badly written ones. Clarity, order, and answer options weigh as much as quantity.
It works for tracking a number, but without the reason you cannot tell what to fix. The metric + open question pair is the smallest format that produces action.
Up to two or three minutes for experience surveys. State the duration in the invitation and honor it: a broken promise depresses the next wave’s participation.
Make only the main metric required. Forcing everything increases abandonment and produces random answers, which are worse than blanks.
Right after the metric, while the reason is fresh. At the end of the questionnaire, fatigue shortens comments and cuts their usefulness.
The references support the definitions and assumptions. The formula used is shown on this page and covered by automated tests.
Put the learning into a real survey and track every response in NPSLab.