The most useful customer survey questions ask about one specific, recent moment: a purchase, a support case, the first weeks with a product. Below are 36 questions sorted by six such moments, each with its answer scale and what a poor result should make you do. Pick five to seven for one survey.
By the Wavefield Research team · Published Nov 13, 2026
Most weak customer surveys fail before the first question is written. They try to cover the whole relationship in one questionnaire, sent whenever someone remembered to send it. A customer cannot rate “your experience with us” in general. They can tell you exactly how last Tuesday's delivery went.
So decide which moment you are asking about. Each has its own timing, its own main measure and its own sensible length.
| Moment | When to send | What to measure | Length |
|---|---|---|---|
| After a purchase | Within a day of delivery or completion | Satisfaction with this order | 3 to 5 questions |
| After a support case | When the case is closed, the same day | Was it resolved, and how much effort it took | 3 to 6 questions |
| First weeks as a customer | Two to four weeks after signing up | Did they get to the result they paid for | 4 to 6 questions |
| The ongoing relationship | Once or twice a year, on a fixed schedule | Overall satisfaction and likelihood to recommend | 8 to 15 questions |
| Before you build or change something | When a decision is on the table | What customers need and what they do today | 6 to 10 questions |
| When a customer leaves | At cancellation, or within a week | The main reason, in their words | 3 to 5 questions |
Each question is written with its answer options, because a question without a scale is half a question. The note under each says what a poor result points to. Replace the square brackets with your own names.
Ask about this order only. The customer remembers it today and will not next month.
Overall, how satisfied are you with your purchase? (Very dissatisfied / Dissatisfied / Neither / Satisfied / Very satisfied)
The standard satisfaction question. Report the share who chose the top two answers. If it drops, the questions below tell you where.
How easy was it to find what you were looking for? (Very difficult / Difficult / Neither / Easy / Very easy)
A poor result here is a navigation or search problem, and no amount of work on checkout will fix it.
How easy was it to complete your order? (Very difficult to Very easy)
Read it beside your abandoned-cart rate. People who gave up never see this survey, so a good score means only that the people who finished found it easy.
Did your order arrive when we said it would? (Yes / No, later / No, earlier / It has not arrived)
A factual question. It sorts late-delivery complaints from everything else before you read a single comment.
Did anything nearly stop you from buying? (Open text)
The most useful open question at this moment. These customers bought anyway, which makes them the best witnesses to what stops other people.
What is one thing we could have done better? (Open text)
Asking for one thing gets a specific answer. Asking for 'any feedback' gets 'no'.
These are the customer service survey questions most teams need. Start with whether the problem is gone.
Was your issue resolved? (Yes / Partly / No)
Ask this first. A customer whose problem is still there will rate everything else through that, and the 'No' group needs a follow-up call more than a score.
How much effort did you personally have to put in to get your issue resolved? (Very low / Low / Moderate / High / Very high)
The customer effort question. High effort with a 'Yes' above means you fixed it the hard way: look at transfers, repeat contacts and how many times they had to explain.
How many times did you contact us about this issue? (Once / Twice / Three times or more)
Check it against your own ticket count. When customers say three and the system says one, contacts are being logged as new cases.
How satisfied are you with the help you received? (Very dissatisfied to Very satisfied)
Satisfaction with the service, kept separate from whether it was resolved. A kind agent who could not fix it shows up as a gap between the two.
How knowledgeable was the person who helped you? (Not at all / Slightly / Moderately / Very / Extremely)
A low result points to training or to documentation the agent could not find. Ask about one quality at a time, never 'knowledgeable and friendly'.
What should we have done differently? (Open text)
Send it to the team lead the same week. The value of this question decays faster than any other on the list.
The question underneath all of these: has the customer got what they paid for yet?
How easy was it to get started with [product]? (Very difficult to Very easy)
A setup score. If it is poor, find the step where people stall before rewriting the welcome emails.
Have you been able to do what you bought [product] to do? (Yes / Partly / Not yet)
The most important onboarding question, and the one most surveys skip. 'Not yet' after four weeks is an early warning of cancellation.
How well does [product] match what you expected when you bought it? (Much worse / Worse / As expected / Better / Much better)
A gap here is often a sales or marketing problem. The product may be fine and the promise wrong.
Which of these have you used so far? (Select all that apply: your main features)
Compare with your usage data. Features people say they have not found are a discovery problem.
What, if anything, was confusing in your first week? (Open text)
Code the answers into themes and count them. Three people naming the same screen is a finding.
What nearly made you give up? (Open text)
Answered only by people who did not give up, so treat it as the mildest version of what the others felt.
Customer experience survey questions for a scheduled check-in, asked the same way every time so the trend means something.
Overall, how satisfied are you with [company]? (Very dissatisfied to Very satisfied)
The headline number for the year. Keep the wording and the scale identical from wave to wave, or the comparison is lost.
How likely are you to recommend [company] to a friend or colleague? (0 = Not at all likely, 10 = Extremely likely)
The Net Promoter question. It belongs here, in the relationship survey, and not after every support ticket.
What is the main reason for your score? (Open text)
Without this, the score is a thermometer with no diagnosis. Coded into themes, it is the most quoted page of the report.
How would you rate [product] on value for money? (Very poor / Poor / Fair / Good / Excellent)
Watch this one before a price change and after it.
Compared with a year ago, has your experience with us got better, stayed the same, or got worse?
A direct trend question. Useful in the first year of tracking, before you have a previous wave to compare with.
If you could change one thing about [company], what would it be? (Open text)
The same 'one thing' device as before, aimed at the whole relationship.
A customer needs survey asks what people are trying to get done. It is the one survey on this page about the future.
What were you trying to get done when you chose [product]? (Open text)
Customers describe the job in their own words, which are usually not the words on your website.
Which of these matter most to you when choosing a [category]? (Select up to three)
Forcing a limit of three makes people choose. Asked as separate importance ratings, everything comes back 'very important'.
What do you still use another product or a workaround for? (Open text)
Existing workarounds are the strongest evidence of an unmet need, because the customer is already paying for them in time or money.
How well does [product] meet your needs? (Not at all / Slightly / Moderately / Very well / Extremely well)
Split the results by customer type. An average of 'moderately' is often one segment served well and another served badly.
How disappointed would you be if you could no longer use [product]? (Very / Somewhat / Not disappointed)
The share who say 'very' are the customers the product is really for. Read their other answers first.
Which one of these should we do next? (Single choice from a short list, plus 'None of these')
Offer real candidates with real trade-offs. Always include 'None of these', or the winner is only the least unpopular.
Short, and sent while the decision is fresh. You are collecting the reason, and it is too late to save the sale.
What is the main reason you are leaving? (Single choice: Too expensive / Missing something I need / Switched to another provider / No longer need it / Problems with service / Other)
Force one main reason. With 'select all', price gets ticked by everyone and tells you nothing.
Was there a particular moment that made you decide? (Open text)
Looks for the trigger, which is often an event you can find in your own records.
What are you using instead? (Another provider / Doing it ourselves / Nothing / Prefer not to say)
Tells you whether you lost to a competitor or to the customer no longer needing the thing at all. The responses to those are different.
What would have kept you as a customer? (Open text)
Many will say nothing would. The ones who name something specific are your list of fixable losses.
How likely are you to use [company] again in the future? (Very unlikely to Very likely)
Separates customers who are gone for good from those who left over circumstances.
May we contact you about your answers? (Yes / No)
Ask, and respect a 'No'. A short call with five willing leavers teaches more than fifty more survey responses.
Related banks go deeper on single topics: NPS survey questions, product survey questions and exit survey questions. How to run the study around the questions is covered in the customer feedback survey guide.
If you need sample customer service survey questions today, this set is complete. It starts with whether the problem was resolved, measures effort and repeat contacts, and ends with one open question. It takes a customer about a minute.
Customer service survey (send when the case is closed) 1. Was your issue resolved? Yes / Partly / No 2. How much effort did you personally have to put in to get your issue resolved? Very low / Low / Moderate / High / Very high 3. How many times did you contact us about this issue? Once / Twice / Three times or more 4. How satisfied are you with the help you received? Very dissatisfied / Dissatisfied / Neither / Satisfied / Very satisfied 5. How knowledgeable was the person who helped you? Not at all / Slightly / Moderately / Very / Extremely 6. What should we have done differently? (Open text, optional)
The order is deliberate. Resolution comes first because it colors every later answer, and you will want to read the other results separately for the “Yes” and “No” groups. The two questions about the agent come after the questions about the outcome, so that a pleasant conversation does not paper over a problem that is still there.
The same customer service survey questionnaire works for an internal help desk. As customer survey questions for IT service, change two things. Replace “contact us” with “contact the help desk”, and add one question: “How long were you unable to work because of this issue?” with time bands as answers. For an IT team, lost working time is the number that gets budget.
A relationship survey covers the whole experience, so it runs longer, and it is sent on a schedule so that the results form a trend. Nine questions is enough. The last one, about who the customer is, matters as much as the first: without it you cannot compare segments.
Customer experience survey (yearly or twice a year) 1. Overall, how satisfied are you with [company]? Very dissatisfied / Dissatisfied / Neither / Satisfied / Very satisfied 2. How likely are you to recommend [company] to a friend or colleague? 0 (Not at all likely) to 10 (Extremely likely) 3. What is the main reason for your score? (Open text) 4. How would you rate [product] on value for money? Very poor / Poor / Fair / Good / Excellent 5. How well does [product] meet your needs? Not at all / Slightly / Moderately / Very well / Extremely well 6. In the past 12 months, have you contacted our support team? Yes / No If yes: How satisfied were you with the help you received? 7. Compared with a year ago, has your experience with us got better, stayed the same, or got worse? 8. If you could change one thing about [company], what would it be? (Open text, optional) 9. Which best describes you? (your own segments: plan, size, region)
Two rules keep a yearly survey useful. Do not change the wording or the scales between waves, because a rewritten question starts a new trend from zero. And send it to a random sample of customers if your list is large, so the same people are not asked every time. The wording of the email that carries it is in the survey reminder email templates.
Three standard questions sit inside the bank above. Each produces a score that can be tracked, and each fits a different moment. Using one of them everywhere is the most common mistake in customer measurement.
| Score | The question | Use it | How it is reported |
|---|---|---|---|
| CSAT (customer satisfaction) | How satisfied are you with [the order, the call, the product]? | Right after a specific event | Share choosing the top two answers on a five-point scale |
| CES (customer effort) | How much effort did you have to put in to get your issue resolved? | After support cases and tasks such as returns or setup | Share reporting low or very low effort |
| NPS (Net Promoter Score) | How likely are you to recommend [company] to a friend or colleague? | In the relationship survey, once or twice a year | Share giving 9 or 10 minus share giving 0 to 6 |
A customer effort survey is the least familiar of the three. The idea took hold after a 2010 Harvard Business Review article argued that companies win loyalty by solving customers' problems, and that going beyond that to delight them adds little. Effort is the right measure wherever the customer came to get something done: a return, a billing fix, a password reset. Nobody wants a delightful password reset. They want a short one.
Each score has its own page with the detail: the CSAT survey guide and the NPS calculator. The scales themselves are compared in the Likert scale guide.
A bank of 36 questions is a menu. Nobody should be served all of it. Every extra question costs responses, and there is evidence for how much. A Cochrane review of randomized trials found that a shorter electronic questionnaire raised the odds of response by about half (an odds ratio of 1.51). For a survey that would otherwise get a 20% response rate, that is roughly 27%. The trials covered many kinds of questionnaire and their results varied, so treat the figure as the direction and rough size of the effect.
A workable way to cut: for each question, write down what you would do if the result were bad and what you would do if it were good. If the two answers are the same, the question is decoration. This removes most “how did you hear about us” questions from surveys where the answer is already in your records.
Keep one open question, and at most two. Open answers are where the reasons live, and each one is also the point where people on phones give up. Wording options for them are in the guide to open-ended survey questions.
People who are delighted or furious respond. People who are mildly content mostly do not. A score from a 10% response rate describes the 10%, and the direction of the gap is unknown until you check who replied against who was asked.
Asked by the company itself, soon after a person helped them, customers round up, and more so when the agent who handled the case is the one asking. Have surveys sent by the system, never by hand.
A question about last quarter's delivery gets an answer about the customer's general feeling toward you. Ask about one recent event, close to when it happened.
A satisfaction score of 86% from 120 responses has a 95% confidence interval of about 79% to 91%. A fall to 82% next month is inside that range and may be nothing at all.
The last point deserves arithmetic, because it is where most dashboards mislead. Before reacting to a change in a score, put a range around it with the confidence interval calculator. Before announcing that one region beats another, test the difference. Two regions at 82% and 74% satisfaction, with 150 responses each, look eight points apart. The statistical significance calculator puts the p-value at 0.09, which does not pass at 95% confidence. With 250 responses each, the same gap would pass.
Keep asking, and read the answers beside what customers do: repeat orders, renewals, support contacts. When the survey and the behavior agree, you can act with confidence. When they disagree, that disagreement is the next thing to investigate.
Start from one moment and one decision. After a purchase, ask how satisfied the customer is with the order, how easy it was, and what one thing could have been better. After a support case, ask whether the issue was resolved, how much effort it took and what should have been done differently. For a yearly check-in, ask overall satisfaction, likelihood to recommend and the main reason for the score. Five to seven questions about one moment beat thirty about everything.
Three to six for a survey sent after a purchase or a support case, and eight to fifteen for a yearly relationship survey. Shorter surveys get more responses: a Cochrane review of randomized trials found that a shorter electronic questionnaire raised the odds of response by about half. A practical test is to ask, for each question, what you would do differently depending on the answer. Questions that fail the test come out.
The best customer satisfaction survey questions name one specific thing and use a balanced scale. 'How satisfied are you with your purchase?' on a five-point scale from very dissatisfied to very satisfied is the standard. Add 'How easy was it to complete your order?' and one open question, 'What is one thing we could have done better?'. Avoid questions that ask about two things at once, such as 'How satisfied are you with our prices and service?'.
CSAT measures satisfaction with a specific event, such as an order or a call. CES, the customer effort score, measures how hard the customer had to work to get something done, and suits support and returns. NPS measures willingness to recommend the company overall and belongs in a relationship survey once or twice a year. They answer different questions, so most programs use CSAT or CES after events and NPS on a schedule.
As close to the event as you can. Send a support survey the day the case closes and a purchase survey within a day of delivery, while the details are still remembered. Send onboarding questions two to four weeks after signup, when the customer has had time to try the product. Relationship surveys go out on a fixed schedule so that results can be compared from one wave to the next.
It depends on what you will do with the answers. Surveys tied to a customer record let you follow up on a bad experience and compare answers with behavior, which is usually worth more than anonymity. Say clearly which kind it is. If the survey is linked to the account, tell customers who will see their answers. If you promise anonymity, do not ask questions that identify a person in a small group.
Related: customer feedback surveys · CSAT survey · NPS survey questions · survey question examples · customer satisfaction survey software
Paste a set into Wavefield or describe the moment you want to measure. The agent builds the survey with scales and skip logic for you to edit. Building and previewing is free; fielding starts at $99 per study.
Build a customer survey free