How to measure customer satisfaction (CSAT)
CSAT is the fastest way to measure how satisfied customers are with a specific experience. Here is the question, the formula, a worked example, and when to send it.
Customer satisfaction (CSAT) is a metric that measures how satisfied a customer is with a specific product, service, or interaction. You measure it by asking one question, usually “How satisfied were you with [the experience]?”, collecting answers on a five-point scale, and then calculating the percentage of respondents who chose the top two ratings. The formula is simple: CSAT = (number of satisfied responses / total responses) x 100. If 160 of 200 people rate you 4 or 5 out of 5, your CSAT is 80 percent.
That is the whole metric in one paragraph. The rest of this guide covers the question wording and scale, the calculation with a worked example, when and where to send the survey, what a good score looks like, and how CSAT differs from NPS and CES so you know which one to reach for.
What CSAT actually measures
CSAT captures satisfaction with a moment, not with your brand as a whole. It answers a narrow question: did this specific experience meet the customer’s expectations? That could be a resolved support ticket, a delivered order, a completed onboarding call, or the first week using a new feature.
Because it is tied to a moment, CSAT is a transactional metric. You fire it right after the event while the experience is still fresh, and you read it as a signal about that touchpoint rather than about the overall relationship. That is what makes it useful for operational teams: a support manager can watch CSAT on ticket resolutions and know within days whether a process change helped or hurt.
It also means CSAT has limits. A customer can rate a single support interaction 5 out of 5 and still be planning to churn. Satisfaction with one moment does not guarantee loyalty, which is why CSAT works best alongside a relational metric rather than on its own. If you are building a broader feedback system, our voice of customer research program guide shows how CSAT fits next to interviews, win-loss, and churn signal.
The CSAT question and scale
The standard CSAT question is a single line:
How satisfied were you with [your recent experience]?
Fill the bracket with the specific touchpoint. “How satisfied were you with the support you received today?” is far better than a vague “How satisfied are you with our company?” The more specific the moment, the more actionable the answer.
The most common answer scale is a five-point scale:
| Rating | Label | Counts as satisfied? |
|---|---|---|
| 5 | Very satisfied | Yes |
| 4 | Satisfied | Yes |
| 3 | Neutral | No |
| 2 | Dissatisfied | No |
| 1 | Very dissatisfied | No |
The 4s and 5s are your “top two boxes,” and only those count as satisfied in the standard formula. Some teams use a three-point or seven-point scale, and others replace numbers with stars or emoji, but the five-point scale is the safest default because it is quick to answer and maps cleanly to the top-two-box calculation.
One rule matters more than the scale you pick: keep it consistent. If you switch from a five-point to a seven-point scale mid-year, your trend line breaks and you lose the ability to compare periods. Pick a scale, document it, and hold it steady.
Always pair the rating with an optional open-text follow-up: “What is the main reason for your score?” The number tells you what is happening; the comment tells you why. For teams that want to act on those comments at scale, the discipline is the same as message testing with real buyers, where the verbatim response is the signal, not the score.
How to calculate CSAT: a worked example
The formula is the number of satisfied responses divided by the total number of responses, times 100. Satisfied responses are the top two boxes.
Imagine you sent a CSAT survey after 200 resolved support tickets and got responses from all 200 customers. Here is how they rated the experience:
| Rating | Label | Responses |
|---|---|---|
| 5 | Very satisfied | 95 |
| 4 | Satisfied | 65 |
| 3 | Neutral | 20 |
| 2 | Dissatisfied | 12 |
| 1 | Very dissatisfied | 8 |
| Total | 200 |
Satisfied responses (the 4s and 5s) add up to 95 + 65 = 160.
CSAT = (160 / 200) x 100 = 80 percent.
That means 80 percent of customers who responded were satisfied with their support experience. The 20 percent who chose neutral or below are your improvement pool, and their open-text comments are where you look first.
A common mistake is to average the raw ratings instead of using top-two-box. Averaging the scores above would give you (95x5 + 65x4 + 20x3 + 12x2 + 8x1) / 200 = 4.16 out of 5. That is a valid mean satisfaction score, but it is not CSAT, and mixing the two methods across reports is how teams end up arguing about numbers that measure different things. Decide on top-two-box percentage, and use it everywhere.
When and where to send a CSAT survey
CSAT is only as good as its timing. Satisfaction fades fast, so you want the survey to reach the customer while the experience is still vivid, ideally within minutes to a few hours of the event.
The strongest CSAT programs trigger the survey off the event, not off a calendar. Good moments to fire a transactional CSAT include:
- Right after a support ticket or live chat is resolved
- After an order ships or a delivery is completed
- At the end of an onboarding or implementation milestone
- After a customer uses a specific feature for the first time
- Following a training session, webinar, or QBR
Channel matters as much as timing. Embed the first rating directly in the medium the customer is already in: a one-click rating inside the support email, a star tap in the app, or a single question on the order-confirmation page. Every extra click costs you responses. Make the first rating a single tap, then send the optional comment field only after they have engaged.
There is a second, slower use of CSAT: a periodic relationship survey sent quarterly to gauge overall satisfaction with your product. That version behaves more like a relational metric and is worth running alongside your transactional CSAT, not instead of it.
CSAT benchmarks: what is a good score?
A CSAT score of 75 to 85 percent is broadly considered good, and anything above 85 percent is strong. But benchmarks are slippery, because the number swings with your industry, your channel, and even how you word the question. According to the long-running American Customer Satisfaction Index, national averages hover around the high 70s on a 100-point scale, and they vary meaningfully by sector year to year.
Rather than chase an external number, treat your own history as the benchmark:
- Watch the trend, not the snapshot. A CSAT moving from 78 to 82 percent over two quarters tells you more than a one-time 80.
- Segment before you judge. Blended CSAT hides problems. Slice by product, region, channel, and team, and a healthy-looking 80 percent often conceals a 60 percent pocket that needs attention.
- Set a minimum sample. Aim for at least 100 responses per segment before reading the number as reliable. On small samples, a swing of several points is usually noise, not signal.
If you want defensible benchmarks for a whole market rather than just your own customers, you need a representative sample of the target segment, which is a recruitment problem more than a survey problem. Our guide to B2B market research processes walks through how to design that kind of study.
CSAT vs NPS vs CES
CSAT is one of three satisfaction and loyalty metrics that teams often confuse. They measure different things and answer different questions, so the right move is usually to run more than one.
| CSAT | NPS | CES | |
|---|---|---|---|
| Measures | Satisfaction with a specific experience | Loyalty and likelihood to recommend | Effort required to get something done |
| Question | How satisfied were you with [X]? | How likely are you to recommend us? | How easy was it to [do X]? |
| Scale | 1 to 5 (top two boxes) | 0 to 10 (promoters minus detractors) | 1 to 5 or 1 to 7 (agreement) |
| Timing | Right after an interaction | Periodic, relationship-level | Right after an effortful task |
| Best for | Operational touchpoints | Long-term brand health | Reducing friction in a journey |
| Type | Transactional | Relational | Transactional |
The short version: use CSAT to check whether a specific moment landed well, use NPS to track overall loyalty over time, and use CES when your goal is to remove friction from a task like onboarding, checkout, or getting help. None of them replaces the others. A customer can find a task easy (high CES), be satisfied with support (high CSAT), and still not recommend you (low NPS), and each of those gaps points somewhere different.
For a fuller treatment of how these metrics fit into a research plan, see our basics of market research primer and the market research methodology guide.
Turning CSAT into decisions
A CSAT score that nobody acts on is a vanity metric. The value is in the loop: read the low scores, mine the comments for themes, ship a change, and watch whether the number moves. That cycle is what separates a satisfaction dashboard from a satisfaction program.
Two habits make the loop work. First, route detractor comments to an owner within a day, so an unhappy customer hears back while the issue is fresh. Second, tie CSAT movement to specific changes you made, so you can prove which fixes worked. This is the same evidence discipline covered in turning product research into better product decisions: the metric is only useful when it is connected to an action.
Running CSAT with the right people
Transactional CSAT reaches your existing customers, which is exactly who you want for operational feedback. But two situations need people beyond your own list. The first is benchmarking, where you want to know how satisfied a whole market segment is, not just your customers. The second is testing satisfaction with a concept, competitor, or category before you have customers of your own.
Both are recruitment problems. You need verified respondents who actually match the segment, and you need them fast enough to keep a research cadence. This is where CleverX fits in: it is a B2B research platform with more than 8 million verified professionals plus B2C reach across 150+ countries, with pay-as-you-go pricing and roughly 2 to 5 day delivery. You can field a CSAT or satisfaction study against a defined target segment and know the respondents are who they claim to be, verified by work email and LinkedIn rather than self-reported.
For teams that struggle to reach the right roles, our guide on how to recruit B2B research participants covers the recruitment mechanics in depth. When you are ready to field a study, you can recruit verified participants on CleverX and reach a matched segment in days.
Frequently asked questions
What is a good CSAT score?
A CSAT score between 75 and 85 percent is generally considered good across most industries, and scores above 85 percent are strong. Benchmarks vary widely by sector, channel, and how you word the question, so the most useful comparison is your own trend over time rather than a single industry number. A score that climbs quarter over quarter matters more than hitting an arbitrary threshold.
How is CSAT calculated?
CSAT is the number of satisfied responses divided by the total number of responses, multiplied by 100. Satisfied responses are usually the top two ratings on a five-point scale, meaning the 4s and 5s, or satisfied and very satisfied. For example, if 160 of 200 respondents rate you 4 or 5, your CSAT is 160 divided by 200, times 100, which equals 80 percent.
What is the difference between CSAT and NPS?
CSAT measures satisfaction with a specific experience or interaction and is usually asked right after that moment. NPS measures overall loyalty and likelihood to recommend the brand as a whole, on a 0 to 10 scale, and is asked periodically rather than after a single touchpoint. CSAT is transactional and immediate, while NPS is relational and long-term. Most mature programs run both.
When should you send a CSAT survey?
Send a CSAT survey immediately after a specific interaction while it is fresh, such as after a support ticket is resolved, an order is delivered, an onboarding call ends, or a feature is used. Timing matters because satisfaction fades and details blur within hours. Transactional CSAT should trigger automatically off the event rather than on a fixed calendar.
What scale should a CSAT survey use?
The most common CSAT scale is a five-point scale running from very dissatisfied to very satisfied, and the top two boxes count as satisfied. Some teams use a three-point or seven-point scale, and others swap numbers for emoji or stars. A five-point scale is the safest default because it is easy for respondents to answer quickly and it maps cleanly to the top-two-box formula.
How many responses do you need for a reliable CSAT score?
For a stable transactional CSAT you generally want at least 100 responses per segment or period before reading the number as reliable, and more if you plan to slice it by product, region, or team. Small samples swing wildly, so a jump from 78 to 85 percent on 20 responses is usually noise. Set a minimum response threshold before you act on any change.