®
Untold Opinion
®
Untold Opinion
Back to Blog
Survey DesignWriting Guide 11 min read

How to Write Survey Questions: The Complete Guide to Better Questions

Bad survey questions produce bad data — no matter how many responses you collect. This guide covers every aspect of writing effective survey questions: choosing the right type, avoiding bias, writing clear answer options, and 25+ real examples of good vs bad questions.

25+
Good vs bad examples
8
Bias types to avoid
6
Question types covered
Free
AI question generator
By Sudhaman (Founder - Untold Opinion)
April 7, 2025
8 min read
General

Why Question Wording Is the Most Important Part of Survey Design

Survey design has many components — question types, survey length, distribution method, incentives — but none matters more than the wording of individual questions. A poorly worded question produces unreliable data regardless of how many people respond. A well-worded question produces actionable insights even from a small sample.

The challenge is that bad questions often look fine on the surface. "How satisfied are you with our service?" seems reasonable — but it's vague (which service?), potentially leading (it assumes there's something to be satisfied with), and doesn't specify a timeframe. Small changes in wording can shift response distributions by 10–20 percentage points.

This guide gives you a systematic framework for writing questions that produce accurate, unbiased, and actionable data.

The 5 Principles of Good Survey Questions

One question, one concept

Each question should ask about exactly one thing. "How satisfied are you with our product quality and delivery speed?" is two questions. Split them. Respondents who feel differently about each aspect can't answer accurately.

Specific and unambiguous

Avoid vague terms like "often", "sometimes", "good", or "satisfied" without defining them. "How often do you use our product?" is vague. "How many times per week do you use our product?" is specific.

Neutral and unbiased

Questions should not suggest a "correct" answer. "How much did you enjoy our excellent customer service?" leads respondents toward positive ratings. "How would you rate your customer service experience?" is neutral.

Mutually exclusive and exhaustive options

Answer options should not overlap (mutually exclusive) and should cover all possible answers (exhaustive). Always include "Other" or "None of the above" when the list might not be complete.

Appropriate for the respondent

Don't ask about things respondents haven't experienced. Don't use jargon they won't understand. Don't assume knowledge they may not have. Always include "Not applicable" or "Haven't used this" options.

Quick Poll — Vote & See Results

What's the most common survey question mistake you've seen?

213 votes so far · Click an option to vote

8 Types of Survey Question Bias (With Examples)

Leading questions

✗ Biased

"How much did you enjoy our award-winning product?"

✓ Better

"How would you rate your experience with our product?"

Describes the product positively before asking for a rating, pushing respondents toward positive answers.

Double-barrelled questions

✗ Biased

"How satisfied are you with our product quality and customer support?"

✓ Better

Split into two separate questions — one for product quality, one for support.

Respondents may feel differently about each aspect. Combined questions produce uninterpretable data.

Loaded questions

✗ Biased

"When did you stop finding our product useful?"

✓ Better

"How useful do you currently find our product? (1–5)"

Assumes the respondent has stopped finding the product useful — a false premise that biases the response.

Acquiescence bias

✗ Biased

"Do you agree that our service is excellent?"

✓ Better

"How would you rate the quality of our service? (1–5)"

People tend to agree with statements regardless of their true opinion. Avoid yes/no agreement questions.

Social desirability bias

✗ Biased

"Do you recycle regularly?"

✓ Better

"How often do you recycle? (Never / Rarely / Sometimes / Often / Always)"

Yes/no questions on socially sensitive topics produce inflated positive responses. Scale questions are more honest.

Order bias

✗ Biased

Listing your brand first in a list of competitors

✓ Better

Randomise the order of options, especially for brand or competitor questions.

The first option in a list receives disproportionately more selections. Randomise to eliminate this effect.

Recency bias

✗ Biased

"How was your experience with us?" (asked 3 months after the interaction)

✓ Better

Send feedback surveys within 24–48 hours of the interaction.

Memory fades and becomes distorted over time. Recent experiences are rated more accurately.

Vague quantifiers

✗ Biased

"Do you often use our product?"

✓ Better

"How many times per week do you use our product? (0 / 1–2 / 3–5 / Daily)"

"Often" means different things to different people. Specific frequency options produce comparable data.

Choosing the Right Question Type

Question type selection by research goal

Multiple choice — quantifiable preferences95%
Rating scale — intensity of opinion90%
Likert scale — attitude measurement85%
Open text — qualitative insight70%
Ranking — relative priority75%
NPS — loyalty measurement88%

Multiple choice

When there are a fixed set of possible answers. Best for demographic questions, preference questions, and categorical data.

Always include "Other (please specify)" if the list might not be exhaustive.

Rating scale (1–5 or 1–10)

When you want to measure intensity — satisfaction, importance, likelihood. Easy to benchmark over time.

Label both ends of the scale clearly. "1 = Very Dissatisfied, 5 = Very Satisfied" removes ambiguity.

Likert scale

When measuring agreement with a statement. Best for attitude and opinion research.

Use 5 or 7 points. Keep all statements in the same direction (all positive or all negative).

Open text

When you need qualitative insight — the "why" behind a rating. Use sparingly (1–2 per survey).

Place open text questions after closed questions on the same topic. Context improves response quality.

Ranking

When you need to understand relative priority. Ask respondents to rank options from most to least important.

Keep ranking lists to 5 items or fewer. Longer lists cause fatigue and random ordering.

NPS (0–10)

When measuring overall loyalty or likelihood to recommend. Best used as a consistent benchmark over time.

Always follow NPS with an open text question asking why they gave that score.

Quick Poll — Vote & See Results

Which question type do you find hardest to write well?

207 votes so far · Click an option to vote

10-Point Checklist: Before You Publish Your Survey

Every question connects to a specific decision or insight you need

No double-barrelled questions (each question asks about one thing only)

No leading or loaded language that suggests a "correct" answer

All answer options are mutually exclusive (no overlap)

All answer options are exhaustive (cover all likely answers, with "Other" if needed)

Vague terms like "often" or "good" have been replaced with specific language

Sensitive questions include an "N/A" or "Prefer not to say" option

The survey has been tested on mobile — all questions display correctly

A colleague has read every question and confirmed they're unambiguous

The survey takes under 5 minutes to complete (time it yourself)

Quick Poll — Vote & See Results

Do you review your survey questions for bias before publishing?

261 votes so far · Click an option to vote

Let AI write your survey questions

Describe your research goal and Untold Opinion's AI generates unbiased, well-structured questions instantly. Free forever.

Create with AIExplore Surveys