How to Write Survey Questions: Examples, Rewrites and Rules

Write a survey question by asking about one thing, in neutral words, for a specific time frame, with answer options that do not overlap. If two different people could read the question two different ways, rewrite it. Below are the rules, then bad-to-good rewrites you can copy straight into a free Qwenvo form.

Start from the decision, not the question

Before you type anything, finish this sentence: 'After this survey I will decide whether to ___.' Every question must help with that decision. If a question only satisfies curiosity, cut it. Short surveys also get finished more often, so every extra question has a cost.

Five rules that fix most bad questions

1. One idea per question. 'How satisfied are you with our prices and support?' asks two things. Split it.
2. Neutral wording. Do not hint at the right answer. 'How helpful was our excellent support team?' leads the witness.
3. A clear time frame. 'How often do you shop with us?' means different things to different people. 'In the last 30 days, how many times did you order from us?' does not.
4. Answer options that cover everything once. Ranges like 0-5 and 5-10 overlap. Use 0-5 and 6-10, and add 'None of these' or 'Other' when you cannot list every case.
5. Plain words. Say 'use', not 'utilize'. Avoid jargon your respondents may not know.

Rewrites you can copy

Each pair shows the common mistake first and the fix second.

• Double-barreled. Before: 'Was the checkout fast and easy?' After: 'How easy was checkout?' and, separately, 'How fast did checkout feel?'
• Leading. Before: 'Don't you agree our new layout is cleaner?' After: 'How would you describe the new layout compared with the old one?'
• Vague. Before: 'Do you use the product regularly?' After: 'In the past 7 days, on how many days did you open the product?'
• Loaded assumption. Before: 'What do you dislike about our onboarding?' After: 'How was your onboarding?' with a follow-up asking what, if anything, could be better.
• Absolute words. Before: 'Do you always read the instructions?' After: 'How often do you read the instructions?' with options from Never to Always.
• Jargon. Before: 'Rate the UX of the dashboard.' After: 'How easy was it to find what you needed on the dashboard?'

Match the question to the answer type

• Multiple choice or Dropdown when you can list the answers. Easy to count.
• Checkboxes or Multi-select when more than one answer can be true. Say 'Select all that apply'.
• Linear scale or Rating for opinions you want to average. Label both ends, such as '1 = Very hard, 5 = Very easy'.
• Matrix when several statements share one scale. Keep it short; long grids tire people.
• Long answer for the one or two things you cannot predict. See open-ended questions.

Build the survey in Qwenvo

1. Click New form, type a title, and press Enter.
2. Type / and pick a block for each question. The title and input are added together.
3. Make only the essential questions required. Forced answers produce random clicks.
4. Order the questions from easy to personal: opinions first, demographics last.
5. Click Preview and complete it yourself on your phone. Then ask two or three real people to read the questions aloud and tell you what they think each one means.
6. Click Publish and share the link from the Share tab.

Short on time? Describe the survey to the AI builder on the empty-form screen, or press Ctrl J inside a form to open Ask Qwenvo. Treat what it writes as a draft and run it through the five rules above. Ready-made starting points are in the templates, for example the customer feedback form and the product feedback survey.

Mistakes to avoid

• Making every question required. People who cannot answer will pick anything to move on.
• Skipping the pilot. Five readers will find problems you cannot see.
• Mixing scales. If 5 means 'great' in one question and 'bad' in the next, answers get muddled.
• Asking what you already know. If you can fill it in from your records, use a hidden field instead of a question. See hidden fields.

An honest limit: Qwenvo checks the form works, not that your wording is fair. No tool can. A pilot with a few real people is the only reliable test.

Questions people ask

How many questions should a survey have?

As few as the decision needs. Test the length on your own audience: watch where people drop off in the form's drop-off analytics and trim there.

What is a double-barreled question?

A question that asks two things at once, such as 'Was the staff friendly and knowledgeable?' A person can say yes to one and no to the other. Split it into two.

Should I use a scale of 5 or 10?

Pick one and keep it the same across the survey. Five points are easier on small phone screens. A 0-10 scale is the standard for NPS. Whichever you choose, label the ends.

Can AI write my survey questions?

It can draft them fast. You still need to check for leading words, double ideas and vague time frames.