Survey Design: How to write questions that get honest answers
Learn how to write questions that avoid leading language, choose the right scale for the right metric, keep surveys short enough to complete, and use logic to adapt to each respondent.
The Psychology of Survey Design
Writing survey questions seems simple: you want to know something, so you ask. But in practice, human psychology makes survey design a minefield. The way you frame a question can easily bias the response, leading to junk data and incorrect business decisions.
1. Avoid Leading and Biased Questions
A leading question nudges the respondent toward a specific answer. For example:
❌ "How satisfied were you with our friendly customer service team?"
The word "friendly" pre-conditions the respondent's attitude. Instead, keep it completely neutral:
✅ "How would you rate the customer service you received?"
2. Say No to Double-Barreled Questions
A double-barreled question asks about two different things but only allows a single response. For example:
❌ "How satisfied are you with our product quality and customer support?"
If a customer is satisfied with the product but hates the support, they cannot answer this accurately. Split these into two distinct questions.
3. Select the Right Scale
Choose a scale that fits your goal. For transactional satisfaction, a 1-5 scale (CSAT) works best. For relationship measurement, a 0-10 scale (NPS) is the industry standard. Always label the endpoints clearly so respondents know which end is positive.
4. Optimize for Completion Rates
The best survey is a completed survey. Keep it under 2 minutes, use skip logic to hide irrelevant questions, and design for mobile-first layouts where taps are easy and text entry is minimized.