Conduct a survey design review before you send

Surveys Fail in Design, Not in Send

Most surveys collect data that cannot drive decisions because they were designed to capture opinion rather than inform action. Typeform's survey design guidance and Dovetail's research writing both document the common failures: leading questions, double-barreled questions, ambiguous scale anchors, too many open-text fields, and no explicit link between the survey and a decision.

How the Conduct a survey design review before you send Prompt Works

The prompt audits the survey across five dimensions (purpose, length, question-level, analyzability, sampling) and forces a decision connection — what will we do differently based on the most common answer? The "one question most likely to produce unusable data" output is the honesty check; every survey has one.

When to Use It

• A survey is scheduled to go to a large customer base.
• A previous survey produced unusable data.
• NPS or CSAT programs are being designed or revised.
• A new PM is standing up their first survey program.
• You want a customer feedback survey to inform a launch decision.

Common Pitfalls

• No decision connection. Surveys without a linked decision collect data nobody uses. Name the decision.
• Too many open-text fields. Open text produces stories but not metrics. Cap at 2 fields per survey.
• Leading questions. "How much do you love X" pre-biases the answer. Rewrite to neutral language.

Sources

• Dovetail Blog — Dovetail
• User Interviews Blog — User Interviews
• Maze Blog — Maze
• Which UX Research Methods — Nielsen Norman Group