What Is CSAT Score (Customer Satisfaction Score)?

CSAT asks a direct question — “How satisfied were you with [interaction]?” — usually on a five-point scale from Very dissatisfied to Very satisfied. The score is the percentage of respondents who chose a “satisfied” rating, out of everyone who answered.

The key difference from NPS: CSAT measures satisfaction with one specific moment, while NPS measures overall loyalty to the relationship. A customer can be highly satisfied with a single support interaction (high CSAT) while still being unlikely to recommend the company overall (low NPS). The two answer different questions and are usually tracked together, not as substitutes.

How to calculate CSAT

The formula is simple:

CSAT (%) = satisfied responses ÷ all responses × 100

What counts as “satisfied” is a choice, and there are two common ones. Top box counts only the highest rating (Very satisfied). Top two boxes counts the highest two (Satisfied and Very satisfied). Top two boxes is widely used, and it always gives a number at least as high as top box on the same responses.

Neither is wrong. What matters is picking one and keeping it, because a score is only comparable with another score calculated the same way. Some teams skip the percentage and report the average rating instead (3.9 out of 5, say). It uses every answer, but it is harder to explain to someone who was not in the room.

The second example above shows the effect. The same 500 answers give 76% on top two boxes, 34% on top box and 3.9 on average.

Choose a scale and label every point

A five-point scale is the usual choice, and it is what the template uses. The more important decision is the labels. A bare number leaves each respondent to decide what a 3 or a 4 means. Putting a word on every point (“Very dissatisfied” through “Very satisfied”) makes the same rating much more likely to mean the same thing to everyone, and it makes the top two boxes easy to define. Keep the same scale from one survey to the next, or your trend line will move for reasons that have nothing to do with your customers. See Likert scale for how labelled scales behave.

What is a good CSAT score?

There is no number that is good everywhere. A score depends on the industry, the interaction you ask about, the scale and whether you count top box or top two boxes. A percentage from one company’s survey tells you little about yours.

One public index gives some orientation, with a caveat: it is a score out of 100 built from several survey questions, not a percentage of satisfied customers, so it is not directly comparable with a CSAT percentage.

• In the United Kingdom, the July 2026 UK Customer Satisfaction Index (UKCSI) was 78.3, covering 277 organisations. See the Institute of Customer Service’s UKCSI summary.

The practical target is your own baseline. Measure the same interaction the same way for a quarter, then aim to improve on that. A rising line on a method you trust is worth more than a headline number from someone else’s.

When and how to ask

Ask right after the interaction, while it is fresh, and ask about that one interaction: right after a ticket closes, right after checkout. Keep it to the rating and one optional comment, because a longer survey lowers the share of people who answer (see response rate) and asking too often wears people out (survey fatigue).

CSAT, NPS and CES

	CSAT	NPS	CES
Asks about	One interaction	The overall relationship	How easy a task was
Typical question	“How satisfied were you with…?”	“How likely are you to recommend…?”	“The company made it easy for me to…”
Usual scale	1–5, labelled	0–10	Often 1–7, disagree to agree
Use it for	Support, purchase, onboarding moments	Loyalty over time	Finding friction in a process

They are complements. CSAT tells you how a moment went, NPS how people feel about you overall, and CES where the process is hard.

Mistakes that skew CSAT

• Leading wording. “How great was our support?” pushes people toward the top. Ask neutrally; see leading question.
• Two questions in one. “How satisfied were you with the speed and friendliness of our support?” cannot be answered with one number; see double-barreled question.
• Reading too much into a small sample. A few dozen answers can swing the score by many points from one week to the next. Check sample size before you act on a change.
• Counting only the people who answer. A score from 8% of customers may describe only the happiest and the angriest. Report how many people answered next to the score.

Frequently asked questions

How do you calculate CSAT?

Divide the number of satisfied responses by the total number of responses and multiply by 100. State whether “satisfied” means the top box or the top two boxes, because the two give different results from the same data.

What is a good CSAT score?

It depends on the industry, the scale and the counting method, so compare with your own earlier scores first. Public indexes such as the UKCSI are scores out of 100, not percentages of satisfied customers, and give only rough context.

Is CSAT the same as NPS?

No. CSAT measures satisfaction with a specific interaction, while NPS measures how likely someone is to recommend the company overall. Many teams track both.

How many responses do you need?

Enough that one unusual answer does not move the score. The smaller the sample, the wider the margin of error, so treat a swing in a small sample as noise until it repeats. See sample size and margin of error.