How to write good survey questions: 6 rules and 24 poor and better examples
Good survey questions ask about one thing, use neutral and plain words, and are specific. Each respondent can answer them, and their answer options are complete and do not overlap. This guide gives these 6 rules, with 4 poor and better pairs for each rule and a one-sentence reason for each pair. All 24 pairs come from marketing work: campaigns, messages, brands, content and events. The guide also gives 7 checks for a question that runs in several languages.
What makes a survey question good?
A good survey question gives you an answer that you can use, and each respondent understands it in the way that you intended. The wording has a large effect: Pew Research Center says that even small wording differences can substantially affect the answers people provide. Source: Pew Research Center: Writing Survey Questions.
Use these 6 rules to check each question:
- Ask about one thing.
- Keep the wording neutral.
- Use plain words.
- Be specific.
- Ask what each person can answer.
- Give complete answer options that do not overlap.
These rules are about wording. To choose the format of a question first, read 10 types of survey questions, with examples.
The 6 rules, with poor and better pairs
The examples are our own, and they do not come from real surveys. Change the words ‘our’ and ‘us’ to the name of your company, event or product where this makes the question clearer.
Rule 1: Ask about one thing
Pew Research Center says that questions about more than one concept, known as double-barreled questions, are difficult for respondents to answer and often lead to responses that are difficult to interpret. Source: Pew Research Center: Writing Survey Questions. To find this problem in your own question, use our double-barreled question checker.
Event
Poor: How satisfied were you with the speakers and the venue at our event?
Better: How satisfied were you with the speakers at our event?
Why: A person can like the speakers and dislike the venue, so ask a separate question about each.
Message
Poor: Was our last email clear and useful?
Better: How clear was our last email?
Why: An email can be clear and not useful, so one answer cannot describe both.
Content
Poor: How useful and easy to find are our guides?
Better: How useful was the last guide that you read?
Why: A low score on the poor version does not show if the problem is the guide or the search for it.
Brand
Poor: Is our brand modern and trustworthy?
Better: How much do you trust our brand?
Why: Modern and trustworthy are two different ideas, so one answer cannot separate them.
Rule 2: Keep the wording neutral
SurveyMonkey tells writers to avoid leading questions that encourage respondents to give a specific answer. Source: SurveyMonkey: Good survey questions. Agree-disagree questions have a different problem. Pew Research Center says that less educated and less informed respondents have a greater tendency to agree with the statement, and that a better practice is to offer a choice between alternative statements. Source: Pew Research Center: Writing Survey Questions.
Campaign
Poor: How much did you love our new campaign?
Better: How much did you like or dislike our new campaign?
Why: The word ‘love’ shows the answer that you hope for, and the better version gives both directions.
Brand
Poor: Why is our brand better than other brands?
Better: How does our brand compare with other brands that you know?
Why: The poor version assumes that the respondent prefers your brand.
Message
Poor: Do you agree or disagree: Our new slogan is clear.
Better: Which statement is closer to your opinion? ‘Our new slogan is clear.’ or ‘Our new slogan is confusing.’
Why: Some respondents tend to agree with a statement, so a choice between two statements gives a more useful result.
Event
Poor: How would you rate our award-winning conference?
Better: How would you rate our conference?
Why: The word ‘award-winning’ adds praise that can change the answer.
Rule 3: Use plain words
Pew Research Center says that simple and concrete language is more easily understood, and that double negatives, unfamiliar abbreviations and jargon can confuse respondents. Source: Pew Research Center: Writing Survey Questions. SurveyMonkey adds that if you must use industry-specific language, you should define it for respondents. Source: SurveyMonkey: Good survey questions.
Campaign
Poor: How would you rate the CTA on our retargeting ads?
Better: How clear was the button on the ad that you saw?
Why: ‘CTA’ and ‘retargeting’ are terms that marketers use, and many customers do not know them.
Brand
Poor: How strong is our brand salience in your category?
Better: When you think of email tools, which brands do you think of first?
Why: A respondent cannot rate ‘brand salience’, but can tell you which names come to mind.
Content
Poor: Would you say that our newsletter is not unhelpful?
Better: How helpful is our newsletter?
Why: The double negative makes the respondent stop and work out which answer means what.
Event
Poor: How was the onboarding flow for our hybrid summit?
Better: How easy was it to join the online part of our event?
Why: The better version replaces internal terms with the task that the respondent did.
Rule 4: Be specific
Pew Research Center says that it is important to ask questions that are clear and specific. Source: Pew Research Center: Writing Survey Questions. A vague word such as ‘recently’ or ‘regularly’ can mean different things to different people, so replace it with a period or a number. This is our own advice.
Message
Poor: What do you think of our emails?
Better: In our last email, how clear was the offer?
Why: A general question gives a general answer, and you cannot see what to change.
Campaign
Poor: Have you seen our ads recently?
Better: Have you seen an ad from us in the past 30 days?
Why: ‘Recently’ can mean one week to one person and six months to another.
Content
Poor: Do you read our blog regularly?
Better: How many of our blog articles did you read in the past 3 months?
Why: ‘Regularly’ has no fixed meaning, but a number from a fixed period is easy to compare.
Event
Poor: How was the event?
Better: Which session at the event was the most useful for your work?
Why: The better version tells the respondent what to judge and gives you a result that you can use.
Rule 5: Ask what each person can answer
Pew Research Center says that it is important to ask questions that each respondent will be able to answer. Source: Pew Research Center: Writing Survey Questions. Check that the question does not assume an action, a memory or a measurement that some respondents do not have. This is our own advice.
Event
Poor: How useful was the closing talk?
Better: How useful was the closing talk? Options: Very useful, somewhat useful, not useful, I did not see it.
Why: Some people leave early, and the poor version gives them no correct answer.
Campaign
Poor: Which of our ads made you buy last year?
Better: Where did you first hear about us?
Why: A respondent may not remember one ad, and the poor version assumes that an ad caused the purchase.
Brand
Poor: Why did you choose us and not our competitors?
Better: Which other brands did you consider before you chose us? Options include ‘None’.
Why: Some customers do not compare brands, so the poor version assumes a decision that they did not make.
Content
Poor: How much did our last guide increase your revenue?
Better: What did you do after you read our last guide?
Why: A reader can report an action more easily than the revenue effect of one guide.
Rule 6: Give complete answer options that do not overlap
Pew Research Center says that closed-ended questions should include all reasonable responses, and that the response categories should not overlap. It also says that, in most circumstances, the number of choices should be small: just four or perhaps five at most, especially in telephone surveys. Source: Pew Research Center: Writing Survey Questions. For the labels of an agreement scale, read Likert scale: definition, examples and how to read the results.
Event
Poor: How many of our events did you go to last year? Options: 1–2, 2–3, 3–5.
Better: How many of our events did you go to last year? Options: 0, 1–2, 3–5, 6 or more.
Why: In the poor version, a person who went to 2 events has two correct options, and a person who went to 0 or 6 has none.
Campaign
Poor: Where did you first see our campaign? Options: Facebook, Instagram.
Better: Where did you first see our campaign? Options: Social media, email, a website, a friend or colleague, somewhere else.
Why: The poor version has no correct option for a person who saw the campaign anywhere else.
Brand
Poor: How would you rate our brand? Options: Excellent, very good, good, okay.
Better: How would you rate our brand? Options: Very good, good, neither good nor bad, bad, very bad.
Why: The poor version has no negative option, so it cannot show dislike.
Message
Poor: How often do you want emails from us? Options: Often, sometimes, rarely.
Better: How often do you want emails from us? Options: Every week, every 2 weeks, every month, less often.
Why: ‘Often’ can mean a different period to each person, so the counts are hard to read.
7 checks for a question in several languages
A question can pass the 6 rules in the language that you wrote it in and fail in a translation. Do these 7 checks for each question, in each language, before you send the form. Most of the facts below come from the translation guidelines of the European Social Survey, a large international survey, and from a study of one questionnaire translation. A small team can do a short version of this work.
- Does the translation keep the meaning? The European Social Survey asks for translations that are functionally equivalent to the source, not word-for-word translations. Source: European Social Survey: ESS Round 11 Translation Guidelines (2022). In a German translation of a questionnaire, the word ‘feel’ could not be translated literally, because the German question implied an evaluation or an emotion as the answer. Source: Walde and Völlm: The TRAPD approach as a method for questionnaire translation, Frontiers in Psychiatry (2023). Ask the translator to say in their own words what the question asks, and compare this with your intention.
- Does the question have an idiom or a word with two meanings? The European Social Survey adds notes where it feels that translators need the intended sense of an idiomatic expression. Its guidelines also say that an English word can have different meanings: ‘fair’ can mean ‘just’ in one context and ‘equal’ in another. Source: European Social Survey: ESS Round 11 Translation Guidelines (2022). So a question such as ‘Is our price fair?’ needs a one-line note for the translator, or a simpler word.
- Does the language need a word form for each gender? Some languages require masculine and feminine versions of a word where English has one form. The European Social Survey guidelines give the word ‘satisfied’, which can be ‘satisfait’ or ‘satisfaite’ in French. They say that it is good to clarify in advance how to deal with this, and they recommend that you keep to one approach across the questionnaire, as far as possible. Source: European Social Survey: ESS Round 11 Translation Guidelines (2022).
- Does each scale keep its strength and its intervals? The European Social Survey guidelines ask translators to mirror the intensity of each scale point. For example, ‘extremely’ should not be translated with a word that is equal to ‘very’, because ‘very’ has less intensity. They also say that if the source intervals have no overlap or gaps, the translated intervals should not have them either. Source: European Social Survey: ESS Round 11 Translation Guidelines (2022).
- Does each repeated term stay the same? The European Social Survey guidelines say that key terms with the same meaning should be translated consistently, because the respondent may otherwise think of a new context. They also suggest that you collect standard translations of repeated instructions in advance. Source: European Social Survey: ESS Round 11 Translation Guidelines (2022). If your form uses the word ‘campaign’ in three questions, give the translator one approved word for it, and check that it fits each question.
- Does each answer option exist in that market? The European Social Survey guidelines discuss a question about a campaign badge or sticker. For a country where these are not usual political symbols, but flags or scarves are, the guidance is to use those symbols. They say that such adaptations should always be discussed with the translation team. Source: European Social Survey: ESS Round 11 Translation Guidelines (2022). For your form, check each channel, job title and price range in the option lists, and write down each change, so that you know what differs when you compare the results.
- Do people who speak the language understand it? The European Social Survey uses five steps: translation, review, adjudication, pre-testing and documentation. Its guidelines say that this team approach is qualitatively better than others, such as the back translation approach. Source: European Social Survey: ESS Round 11 Translation Guidelines (2022). A study of a German questionnaire translation says that a forward translation with no further quality checks is prone to errors and should be avoided. In its pretest, patients said that the middle label ‘unentschieden’ (undecided) did not match their thoughts, so the researchers changed it to ‘teils teils’ (partly). Source: Walde and Völlm: The TRAPD approach as a method for questionnaire translation, Frontiers in Psychiatry (2023). Ask a few people who speak the language to answer the form and tell you what each question means.
Test the questions before you send the form
Pew Research Center calls pretesting an essential step in the questionnaire design process. Source: Pew Research Center: Writing Survey Questions. SurveyMonkey says that a test with a small group helps you find issues or biases in your questions before the survey goes to a larger audience. Source: SurveyMonkey: Good survey questions. Our advice is to ask a few people from your audience to read each question aloud and say what it asks, and to change any question that they understand in a different way.
After you send the form, you can measure the result with our survey response rate calculator.
Questions and answers
What is a leading question?
A leading question encourages respondents to give a specific answer. SurveyMonkey gives this example: ‘Since you love our product so much, how likely are you to recommend it to a friend?’ Source: SurveyMonkey: Good survey questions.
What is a double-barreled question?
It is a question that asks for an opinion on two things at one time. SurveyMonkey gives this example: ‘How would you rate our customer service and product reliability?’ Source: SurveyMonkey: Good survey questions. To check your own question, use our double-barreled question checker.
How many questions should a survey have?
The sources for this article give no fixed number. Our advice is to keep a question only if its answer can change a decision that you have to make.
Does the order of the questions matter?
Yes. Pew Research Center says that the order of questions can influence how people respond, because earlier questions can unintentionally provide context for the questions that follow. Source: Pew Research Center: Writing Survey Questions. Read the full form in order, and check whether a question changes the meaning of the next one.
Should a good question be open-ended or closed-ended?
It depends on what you need to know. To count answers, use a closed-ended question. To find answers that you did not expect, use an open-ended question. Read The difference between qualitative and quantitative survey questions for paired examples.
Where Tickbox fits
Tickbox is a form builder with voice and video. Forms receive text, audio and video responses, and AI analyzes the responses so they are easy to read. AI analysis is on all plans, including Free. Forms support multiple languages. You can use this guide with any form tool, or start from a free form template.
Sources for the facts in this article
Each external fact above links to its source, and the list below shows what each source supports. We checked these sources on 4 November 2026. The example questions and the advice with no source link are our own.
- Pew Research Center: Writing Survey Questions: the effect of wording, double-barreled questions, agree-disagree questions, plain language, double negatives and jargon, clear and specific questions, complete answer options that do not overlap, the number of choices, pretesting and question order.
- SurveyMonkey: Good survey questions: leading questions, double-barreled questions, industry-specific language and pretesting.
- European Social Survey: ESS Round 11 Translation Guidelines (2022): functional equivalence, idiomatic expressions and words with several meanings, gender, scale intensity and intervals, consistent terms, adaptations, and the five translation steps.
- Walde and Völlm: The TRAPD approach as a method for questionnaire translation, Frontiers in Psychiatry (2023): the German translation of ‘feel’, the risk of a forward translation with no further checks, and the change of a scale label after the pretest.