CSAT Score: Formula, Benchmarks and How to Improve It

CSAT (Customer Satisfaction Score) measures how satisfied customers are with a specific product, interaction or experience. You ask a satisfaction question, respondents choose a rating, and the CSAT score is the percentage of respondents who chose one of the top ratings on the scale.

The key word is specific. CSAT is a transactional metric: it works best right after a concrete moment such as a support conversation, a purchase, an onboarding step or a delivery. That focus is what makes it actionable — a low score points at a particular experience you can go and fix. For how CSAT sits alongside other measures, see customer satisfaction metrics.

On the common 1–5 satisfaction scale, the top two ratings (4 and 5) count as 'satisfied.' The CSAT score is the share of all responses that fall into those top boxes:

CSAT = (number of 4 and 5 ratings ÷ total number of responses) × 100

The result is a percentage from 0 to 100. A CSAT of 80 means 80% of respondents rated their experience a 4 or 5. The exact wording of the top ratings varies (for example, 'Satisfied' and 'Very satisfied'), but the principle is the same: count the positive responses and divide by the total.

Suppose you sent a post-support CSAT survey and received 200 responses on a 1–5 scale. The ratings broke down like this:

• Rating 5 (Very satisfied): 96 responses
• Rating 4 (Satisfied): 44 responses
• Rating 3 (Neutral): 30 responses
• Rating 2 (Dissatisfied): 18 responses
• Rating 1 (Very dissatisfied): 12 responses

Follow the steps:

1. 1Add the top-box responses (4 and 5): 96 + 44 = 140 satisfied responses.
2. 2Count total responses: 96 + 44 + 30 + 18 + 12 = 200.
3. 3Divide satisfied by total: 140 ÷ 200 = 0.70.
4. 4Multiply by 100: 0.70 × 100 = 70.

The CSAT score is 70. You can skip the arithmetic and enter your rating counts directly into the CSAT calculator to get the same result instantly.

There are two common ways to report satisfaction, and they answer different questions. The top-box CSAT (the percentage choosing the top ratings) is what most people mean by 'CSAT score.' The average rating is simply the mean of every rating.

In the example above, the average rating is (5×96 + 4×44 + 3×30 + 2×18 + 1×12) ÷ 200 = 3.97. That single figure hides the shape of the data: the average looks fine, but 30 customers gave a 1 or 2. The top-box CSAT of 70 is more honest about how many people you disappointed.

Neither number is wrong, but the top-box percentage is usually more actionable because it focuses attention on the customers who were not satisfied. Whichever you use, always look at the full distribution rather than a single summary figure.

The most common question about CSAT is 'what is a good score?' It is tempting to reach for a universal benchmark, but there is no single reliable number that applies across companies — and quoting invented figures would only mislead you. What matters is understanding why benchmarks vary so much.

• Scale and wording: a 1–5 top-two-box score is not comparable to a 1–7 or 1–10 score, or to one that counts only the single highest rating.
• Touchpoint: CSAT after a smooth purchase tends to run higher than CSAT after a support escalation, even at the same company.
• Industry and expectations: customers hold different baseline expectations across sectors, so 'good' shifts with what people are used to.
• Culture and geography: rating tendencies differ across regions, which shifts absolute scores independent of real satisfaction.
• Response bias: who replies (only the delighted or only the angry) can move a score more than any real change in the experience.

In practical terms, a high CSAT means most of your customers had a positive experience, and a low CSAT means a meaningful share did not. Rather than fixating on a magic threshold, interpret your score in context:

• A 'good' score is one that is stable or improving over time using a consistent methodology, with a distribution that is not hiding a cluster of unhappy customers.
• A 'bad' score is one that is declining, or one where a healthy-looking average masks a significant tail of 1s and 2s.
• A sudden drop matters more than the absolute level — it usually signals a specific, recent problem you can trace to a change in the experience.

Once you know where you stand, the next step is turning the score into action. Read the open-text comments behind the ratings, and follow the playbook for improving customer satisfaction to prioritize fixes.

CSAT is one of three widely used experience metrics. They are complementary, not competing — each answers a different question.

For a deeper comparison, see customer satisfaction vs NPS and customer satisfaction vs CES.

The CSAT number tells you where you stand; the comments tell you what to do. Always pair the rating question with an open-text 'why' follow-up, then group the responses into themes so you can see what is driving satisfaction or frustration.

AI analysis makes reading those comments cheap: it can summarize hundreds of open-text answers into recurring themes and sentiment in minutes. And when you need to build the survey itself, you can describe your goal and create a CSAT survey with AI rather than writing every question by hand.