Spot double-barrelled, leading, vague and forced-choice wording, then make one neutral question at a time. These examples include wording, response choices, use notes and a bias to watch for.
Free to useExamples are editorial guidance, not validationUpdated
Double-barrelled: delivery and product quality
Avoid: ‘How satisfied were you with the delivery speed and product quality?’ Rewrite as two questions: ‘How satisfied were you with the delivery speed?’ and ‘How satisfied were you with the product quality?’
Response choices or scale
Use the same satisfaction scale for each item
One idea per question.
When to use it
When a question contains ‘and’ or asks about more than one experience that may receive different answers.
Bias to avoid
A respondent may be satisfied with one part and dissatisfied with the other, leaving no valid single response.
Source
Illustrative rewrite demonstrating a standard questionnaire-design principle. This is illustrative common practice, not a validated standard.
Leading: praise embedded in the question
Avoid: ‘How excellent was our new support experience?’ Rewrite as: ‘How would you rate your most recent support experience?’
Response choices or scale
Very poor
Poor
Neither poor nor good
Good
Very good
I did not contact support
Balanced, labelled choices with a not-applicable option.
When to use it
When adjectives or assumptions in the question may push answers in one direction.
Bias to avoid
The word ‘excellent’ presumes a positive experience and can inflate ratings.
Source
Illustrative rewrite demonstrating neutral wording and balanced response choices. This is illustrative common practice, not a validated standard.
How to adapt these examples
Use bracketed text to name the product, experience or time period. Keep each question focused on one idea, offer an honest way to answer or skip, and pretest it with people from the intended audience. A change to wording or choices can change what is measured.
When should I use bad survey question examples and rewrites?
Spot double-barrelled, leading, vague and forced-choice wording, then make one neutral question at a time.
Should I copy these examples word for word?
Treat them as starting points. Replace bracketed text for your audience, pretest the wording and keep the answer choices appropriate to the decision.
Are these questions validated or benchmarked?
The examples labelled common practice are not psychometrically validated standards and this library does not provide benchmarks. Test questions with the intended audience before relying on results.
PulseLake · Research Intelligence OS.
Run research end to end. Keep the knowledge working.
One AI-native operating system for market research and insight professionals — from study design and evidence generation to agents, institutional knowledge, delivery and action.
