Customer Satisfaction Survey Template
A 12-question customer satisfaction survey you can send today. One question gives you the CSAT score, six drivers explain it, two are in the customer's own words, and it closes on a rating. Preview the questions below, lift the ones you want, or open the template in the editor and reword it before you send it.
- 12questions
- 2 to 3 minto complete
- 5-pointsatisfaction scale
- Anonymousby default
What a customer satisfaction survey measures
A customer satisfaction survey asks one person to rate one experience, on a fixed scale, soon after it happened. Do that often enough and the answers turn into a number you can watch move.
The score. Question one is the CSAT item. Your CSAT is the share of people who picked the top one or two answers on it, written as a percentage. Nothing else on this page changes that number.
The drivers. Six questions cover the things a customer can be pleased or annoyed about: the product, the people, the speed, how easy you were to deal with, the price, and how well you kept them informed. None goes into the score; their job is to say which one is holding it down.
Is CSAT the right measure for you? Use it for a number on one interaction, tracked over time. For the reasons in people's own words and no score at all, run a customer feedback survey instead. Both are built and ready to send in the free survey creator.
The 12 questions, and the wording to swap in for each moment
Every question, its scale, and what a low answer is telling you. Copy one block or the whole set, or use the template and skip the retyping. The last three blocks are alternative wordings for a specific moment.
Why every question here is a rating scale. On this platform, a rating scale question reached by a respondent is answered 96.2% of the time, in a median of 5.0 seconds, across 159,545 askings, the highest answer rate of any question type we measure.[3] A star rating manages 93.7% in 4.6 seconds, across 57,468 askings. That is the design case for this whole survey: twelve rating scales, no stars, and open text only where it is marked optional.
The score
The number, and the same thing from a different angle, so you can tell a good experience from one that merely beat a low bar.
- Overall, how satisfied were you with your most recent experience with us?The score. Everything else explains it.
- Compared with what you expected, how did it turn out?High satisfaction with a low expectation means the bar was low.
The six drivers
One question each for what a customer can be pleased or annoyed about. They stay out of the score; their job is to say what is holding it down.
- How satisfied are you with the product or service itself?The thing itself. If lowest, service will not move the score.
- How satisfied are you with the people you dealt with?The people. Usually the widest gap.
- How satisfied are you with how quickly it was done?Speed. Slow but simple often beats fast and confusing.
- How satisfied are you with how easy we were to deal with?Effort. Rarely volunteered, always felt.
- How satisfied are you with the value for what you paid?Price against what they got.
- How satisfied are you with how clearly we kept you informed?Communication. Cheapest to fix, most often missing.
In their own words
Two open questions, both optional, before the end. Never make either required; one mandatory text box loses the response.
- What is the one thing we could have done better?The most usable open prompt there is.
- What did we get right?Protects what works, easy to remove by accident.
The two closing questions
Ends on rating scales, the ending that loses the fewest answers. Both ask about the future rather than the event.
- How confident are you that we would get it right next time?Confidence falls before satisfaction does.
- How likely are you to use us again?Repeat intent. Moves before revenue does.
Swap in: right after a purchase
In place of questions one and three to six after a checkout. Same scale, so the score stays comparable.
- Overall, how satisfied were you with buying from us today?The purchase, not the product, which has not arrived yet.
- How clear was the price before you paid?Top complaint in post-purchase text.
- How clear was the delivery date?Split from price so each surprise is fixed separately.
- What nearly stopped you buying today?Names the moment, unlike "any other comments".
Swap in: after a support ticket
Fire once the fix has held, not the second the ticket closed.
- How satisfied are you with the way your issue was handled?The handling, not the outcome.
- How satisfied are you that your issue is now actually resolved?A closed ticket and a solved problem differ.
- What would have made this quicker to sort out?Names the fix, not the feeling.
Swap in: after a delivery or an install
One to three days after the van leaves: long enough to use it, short enough to recall who delivered it.
- How satisfied are you with your delivery?The delivery, not the goods.
- How satisfied are you that it arrived when you were told it would?The biggest driver in delivery feedback.
- What would you change about the delivery?Keeps the answer on delivery, not the product.
Where these questions come from. The wording is ours. The two score questions follow the shape of the American Customer Satisfaction Index's published method: an overall rating, and a comparison against what the customer expected.[1] Nobody owns the words "how satisfied were you", so edit, translate and reuse every line.
How to calculate CSAT, worked through
The formula is one line, and almost every page carries it. What they leave out is the arithmetic, and the three different numbers hiding in the same 63 answers.
The formula. Count the people who picked the top one or two answers, divide by the number who answered the question, multiply by 100, or let the survey tool total it for you.
Worked, with real numbers. Say 63 people answered question one: twenty-one picked 5, twenty-seven picked 4, nine picked 3, four picked 2, two picked 1. The 5s and 4s add to 48. 48 divided by 63 is 0.762. Your CSAT is 76 percent.
The same answers, three ways. Top two boxes gives 76 percent. Top box only gives 33 percent. The plain average gives 3.97 out of 5. All three are honest; pick one and never switch, because a switch reads exactly like a change in satisfaction.
Dealerships usually score this as CSAT; the car dealership survey has the wording.
For one help request rather than a purchase, the ticket, the repair or the visit, the customer service survey asks whether it was fixed and how hard it was to get there.
This scores one interaction.
What your CSAT means when only 20 people answer
This is the question every small business has, and none of the pages ranking for it answers. Here is the arithmetic, and the point below which you should stop quoting a percentage at all.
| Answers you got | One person is worth | Where your true score sits | What to report |
|---|---|---|---|
| 20 | 5 points | 57% to 95% | The counts: 15 of 20 picked 4 or 5. |
| 50 | 2 points | 64% to 88% | A percentage, with the count next to it. |
| 100 | 1 point | 68% to 84% | A percentage. Compare quarters, not weeks. |
| 400 | 0.25 points | 72% to 80% | A percentage, splittable by team or product. |
How that table is built. The third column is the ordinary 95 percent interval for a proportion, worked at a score of 76 percent, rough below about 30 answers.
Below about 30 answers, report the count. "Fifteen of twenty people picked 4 or 5" is clear. "75 percent" invites a comparison with last month's 80 that the data cannot support.
Two numbers, one honest comparison. Last quarter 76 percent on 63 answers, this quarter 81 percent on 58: those ranges overlap almost completely. You have not improved.
Ask everyone if you are small. Send it to everybody who had the experience, whether or not you expect them to be kind. Strong opinions self-select into answering more than mild ones, confirmed across more than 280 million online reviews,[4] so a survey fired only where you do well produces a good score rather than a correct one.
Never put a percentage on fewer than about ten answers. At eight responses, one person is worth 12.5 points. A dashboard tile reading 88 percent is a description of one customer.
When to send it, and how to get answers back
How the timing and the ending of a survey change whether it gets answered, from what we can measure across 2.27 million responses on this platform.[3]
Timing windows, and why ours are convention rather than evidence
Nearly every page says to send while the experience is fresh, but no published window is behind a study.[5] Ask too early and the customer cannot judge the thing yet; ask too late and they answer about your business in general. The table below is convention, not evidence.
| The moment | Our window | Why |
|---|---|---|
| After a purchase, in a shop or online | Same day, up to 48 hours | They remember checkout; the product is not yet used. |
| After a support ticket closes | Within a day, after the fix has held | Wait for the resolution, not the reply. |
| After a delivery or an install | 1 to 3 days | Long enough to use it, short enough to recall the driver. |
| After onboarding or a first month | 2 to 4 weeks in | They have a real opinion by then. |
| A periodic check-in | Twice a year, the same two months | A trend needs a fixed interval. |
This survey scores a single interaction. For one comparable number across the whole relationship instead, that is the net promoter score survey.
Four rules that belong to this survey
The send mechanics every survey shares, one link, one reminder, a stated length and a closing date, are in the survey research guide. These four are the ones a satisfaction score depends on.
Fire it from the event, to everyone who had it
Trigger it from the event record, the order, the closed ticket, never from staff picking who gets asked. Everyone who had it is eligible.
Swap in the block for that moment; the scale is unchanged.
Ask each customer once a quarter at most
Cap it in the trigger: no second survey to the same person inside 90 days. Weekly regulars are the happiest, so counting them weekly flatters the score.
Keep the score off the bonus sheet
Once a rating decides pay, staff ask for fives and customers give them; the score rises while complaints hold steady. Reward response count: nobody can coach it.
Freeze the scale, and file the count with the score
Same labels, same order, same formula every time; any change reads as a change in satisfaction. Print the count beside every percentage: 76 percent of 63 is not 81 percent of 12.
From a score to something you actually changed
The survey is the cheap part. Asking and then doing nothing is what people say they resent, and what stops them answering the next one.
Find the driver holding the score down. Split answers in two, people who rated a driver 4 or 5 and everybody else, and compare the CSAT of each group. The biggest gap is your lever, even if its own average is not the lowest.
Tag every open answer once. Eight to fifteen tags covers any business. The tag that comes up most is usually cheaper to fix than the score suggests.
Reply to the unhappy answers by hand. A person, a name, one sentence about what you will do. Finish the job.
Change one thing, and say what it was. One change, announced, beats five nobody hears about.
Customer satisfaction survey FAQ
What is a customer satisfaction survey?
A short questionnaire that asks a customer to rate one experience on a fixed scale. The answers become a single percentage, the CSAT score, tracked over time.
How do you calculate CSAT?
Count the people who picked the top one or two answers, divide by everyone who answered, multiply by 100. 48 of 63 picking 4 or 5 gives a CSAT of 76 percent. Pick top-one or top-two and never switch.
What is a good CSAT score?
No published number is worth aiming at. National benchmarks are modelled 0-100 indexes, not a percentage of satisfied customers, so they are not directly comparable to your CSAT. Compare your score with your own last quarter.
Should I use a 5-point or a 10-point satisfaction scale?
Five, unless you have a specific reason. A study of 149 people found reliability rising to about seven categories and preference highest for ten,[2] but five fits a phone screen and carries the top-two-box percentage everyone reports. Never change the scale once started.
Should I include demographic questions?
Only if you will act differently depending on the answer. Every one costs completions and can identify a person in a small business. Attach product, location or team in your own records instead.
How do I improve a low CSAT score?
Split answers by each driver question, compare the satisfaction of people who rated that driver well against those who did not, and fix the driver with the biggest gap. Reply personally to everyone who scored you 1 or 2, then tell customers what changed.
What is a good customer satisfaction survey response rate?
There is no reliable published number. The response-rate benchmarks that vendors publish come from their own platforms with no disclosed sample and no shared way of counting, so they are not comparable with yours. Track your own rate quarter to quarter instead: a fall usually means the survey is arriving at the wrong moment or too often.
Methods and sources
What this template is based on
The two score items follow the constructs ACSI publishes;[1] timing and completion figures come from the SuperSurvey Response Benchmark.[3]
Question licensing
No licensed instrument is reproduced; ACSI is cited with attribution and a link only.[1] Copy, edit or translate freely.
Limits and disclosure
Confidence ranges are normal binomial approximations, rough below 30 answers; the timing table is convention;[5] ACSI's national score is a modelled 0-100 index, not a top-two-box percentage or your norm.
SuperSurvey makes the editor this template opens in. The template is free either way.
References
- ACSI. US Overall Customer Satisfaction. American Customer Satisfaction Index, 2026. theacsi.com
- Preston, C.C. and Colman, A.M. "Optimal number of response categories in rating scales." Acta Psychologica 104 (2000): 1-15. doi.org
- SuperSurvey Response Benchmark, 16 September 2023 to 25 August 2026: 25,112 analysed surveys with 998,864 completed responses, out of 2,274,520 responses platform-wide. Method note.
- Schoenmueller, V., Netzer, O. and Stahl, F. "The Polarity of Online Reviews." Journal of Marketing Research 57, no. 5 (2020): 853-877. doi.org
- CustomerGauge. "NPS survey best practices." CustomerGauge blog. customergauge.com
Send the customer satisfaction survey this week
Twelve questions, about two minutes, one 5-point scale from the first question to the last. Open it in the editor, change anything, and share one link.
Use this template