A survey is a measurement instrument, not a communication exercise. When it is designed well, it produces insight you can act on. When it is not, it produces noise, survey fatigue, and cynicism.
Start with the decision, not the questions
Before writing a single item, decide what decisions the results will inform and who will act on them. If no one can name the decision, the survey should not run.
Design principles that protect data quality
- Use validated instruments where they exist rather than inventing new scales
- Keep it short: every extra item lowers completion and quality
- Avoid double-barrelled and leading questions
- Offer a balanced response scale with a clear neutral point
- Protect anonymity so people answer honestly
Response rates are a design outcome
Low response rates are rarely a mystery. They follow long surveys, weak communication, and a history of running surveys whose results went nowhere. Visible follow-through is the most reliable way to raise participation next time.
Analysis and follow-through
Segment results to find meaningful differences, report them honestly including the uncomfortable findings, and commit to a small number of actions with owners and dates. Communicate what changed as a direct result.