Customer research · 12 Nov 2026 · 6 min read
Customer surveys: how to run one and write questions people answer
A survey is worth running when it answers a decision you have to make. Here is how to plan one, write neutral questions and use what comes back.
The short answer
To run a customer survey, decide the one decision it should inform, choose who to ask and when, write short neutral questions with a mix of closed and open answers, test it on a few people, send it, then act on the results and tell customers what changed. Keep it as short as the decision allows.
Customer surveys start with one decision
Customer surveys are worth running only when the answers will change something. Before you write a question, finish this sentence: after the results come in, we will decide whether to do this. It might be which product to develop, what to fix at the checkout, which message to use or which customers to contact. If you cannot complete the sentence, you are collecting opinions with no use for them.
One survey, one decision. A survey that tries to settle pricing, product, service and brand questions at once is long, tiring to answer and hard to act on. Split it into separate surveys sent at different moments.
Who to ask and when
| Group | Best moment | What you can learn |
|---|---|---|
| New buyers | Soon after delivery or first use | How buying went and what nearly stopped them |
| Regular customers | After a normal repeat order, or at a steady rhythm | What keeps them coming back and what they would change |
| Lapsed customers | Some time after their last order | Why they stopped |
| People who enquired but did not buy | Soon after the decision | What they chose instead and why |
| Website visitors | While on a key page, with a short on-page prompt | What they came to do and whether they found it |
Ask close to the experience, while memory is fresh. A question about a delivery asked months later gets vague answers. Hold back from anyone in the middle of an unresolved complaint, and sort the complaint first.
Email is the usual route for existing customers, and you should check the rules on contacting people in each country you sell to. A short prompt on the site reaches visitors you have no contact details for. Sending the message automatically after an order, or after a support request closes, saves effort, and marketing automation explained shows how such triggers work. A poll on social media gives quick answers from people who already follow you, who may not look like your wider customers. Digital marketing versus social media marketing explains what each channel can and cannot do.
Question types and what each is good for
| Type | Good for | Watch for |
|---|---|---|
| Multiple choice | Counting answers and comparing groups | Missing options, so add an other box |
| Rating scale | Tracking satisfaction over time | People use scales differently, so keep the scale identical each time |
| Yes or no | Simple facts, such as whether they found what they wanted | It hides the reason, so follow with a why |
| Ranking | Finding priorities | It tires people if the list is long |
| Open text | Reasons, customer wording, surprises | It takes time to read, so use it sparingly and read every answer |
How many questions a survey should have
As few as your decision needs. There is no magic number, but there is a useful test. For every question, write down what you will do if the answers come back one way and what you will do if they come back the other. If both actions are the same, or you cannot name any, delete the question.
Most small business surveys end up as a handful of closed questions and one open box. A survey that needs several screens of scrolling is probably several surveys stuck together. Put the easy closed questions first and the open question last, because people drop out as effort rises. Tell them at the start roughly how long it will take, and keep that promise.
How to avoid biased questions
A biased or loaded question steers the answer. The common forms are wording that signals the right answer, questions that ask about two things at once, and statements that invite people to agree. Compare these pairs.
| Biased version | What goes wrong | Better version |
|---|---|---|
| How much did you enjoy our excellent service? | The word excellent signals the answer | How would you describe our service? |
| Was the checkout fast and easy? | Two questions in one | How easy or difficult was the checkout? |
| The delivery was on time, wasn't it? | It invites agreement | How did the delivery time compare with what you expected? |
| Which of our great features do you use most? | It assumes the features are great and that they use some | Which of these features, if any, do you use? |
| Do you agree that our prices are fair? | Agree or disagree statements encourage agreeing | How do you feel about our prices compared with what you received? |
A few habits help. Offer not sure or does not apply when it is true. Keep answer options in a neutral order, and rotate them if your survey tool allows. Avoid jargon the customer would not use. Ask about what people did, not what they might do, because predictions about future behaviour are unreliable.
Test it, then send it
- Read it aloud and remove anything that makes you stumble.
- Ask a few colleagues or friendly customers to complete it while you watch, and note where they hesitate.
- Check it on a mobile device, because many people will answer on one.
- Send it with a short, honest explanation of why you are asking and how long it takes.
- Send one polite reminder to people who have not replied, then stop.
- Close the survey on a date you chose in advance.
Reading the results
Read every open answer, and group them by theme before counting. Quote customers' own words in your notes, because they are also useful when you write copy. Treat small numbers cautiously. A theme raised by two people is a lead, not a finding. People who answer surveys differ from people who do not, often being happier or more annoyed than average, so do not assume the replies represent everyone.
Compare what customers say with what they do. If people say they struggled to find information, the entry-page and engagement reports described in GA4 reports small businesses should use can show where visitors leave. Findings that repeat across many answers also belong in your buyer profile, and how to create a buyer persona shows how to record them.
Privacy, incentives and fairness
Rules about contacting people, storing their answers and offering rewards differ by country and change over time. Ask a lawyer or the official regulator what applies in the markets you sell to. As a working habit, say who is asking and why, request only the personal details you need, state whether answers are anonymous and keep that promise, and store results securely.
If you offer a reward, think about whether it will attract people who only want the reward, and make clear that it does not depend on giving a positive answer. Never push people toward a good score.
Close the loop
Tell respondents what you learned and what you changed, even in one sentence in a newsletter or on the site. It shows the survey was not a formality and raises the response next time. Follow up directly with anyone who reported a serious problem and agreed to be contacted. Respondents who were happy are good people to invite to leave a public review, and how to ask for and use customer reviews explains how to do that without pressure.
If you would like help planning a survey and joining its results to your tracking, the digital marketing page describes the work Liftup Technologies does in this area.
Related guides
- Digital marketing
- Buyer persona how to create one
- Customer reviews how to ask and use them
- Digital marketing vs social media marketing
- GA4 Reports: The Ones Small Businesses Should Use
- Marketing automation explained
Questions and answers
How do I run a customer survey?
Decide the one decision the survey will inform, choose the group to ask and the moment, write a few neutral questions and test them on a handful of people. Send it with a short explanation, read every open answer, and tell customers what you changed.
How many questions should a customer survey have?
As few as your decision needs. For each question, write down what you would do with each possible answer, and delete any question where the action would be the same or unknown. Many small business surveys work with a handful of closed questions and one open box.
How do I avoid biased or loaded questions?
Use neutral wording, ask about one thing at a time, and avoid statements that invite agreement. Replace how much did you enjoy our great service with how would you describe our service. Offer a not sure option, keep answer order neutral, and test the survey on a few people first.
When is the best time to send a customer survey?
Soon after the experience you are asking about, while memory is fresh. For a purchase that means shortly after delivery or first use. For support, send it when the request is closed. Avoid people with an unresolved complaint until it has been dealt with.
Should I offer an incentive for completing a survey?
It can raise the response, but it may also attract people who only want the reward. If you offer one, make clear that it does not depend on a positive answer. Rules on rewards and on contacting people differ by country, so check with a lawyer or the official source.
What should I do with open text answers?
Read every one, group them by theme and count how often each theme appears. Keep a few quotes in the customers' own words. Treat themes raised by only a couple of people as leads to check, not as conclusions.
Tell us what you are building.
A short call, a written plan and a fixed quote. No obligation.