CSAT Calculator

CSAT = satisfied responses ÷ total responses × 100.

On a 5-point scale, "satisfied" conventionally means the top two boxes — 4 and 5. Counting only 5s gives a much harsher number, and counting 3s gives a flattering one. Pick one definition and never change it, because the comparison over time is the whole value.

Out of 5: count the 4s and 5s as satisfied. 120 satisfied out of 150 responses = 80% CSAT. Out of 7: count 6s and 7s. Out of 10: count 9s and 10s (or 8–10 if your team has always done so — consistency matters more than the cut-off).

CSAT is a percentage of satisfied respondents, not an average rating. "4.2 out of 5" is a mean score and a different metric; it hides how many people were unhappy. If you want one number to track, report the percentage and show the distribution beside it.

CSAT is usually measured with one question asked right after an interaction, such as "How satisfied were you with your experience today?", answered on a labeled scale from "Very dissatisfied" to "Very satisfied". The formula is: CSAT = satisfied responses ÷ total responses × 100.

Worked example (illustrative numbers): 420 customers rate a support conversation on a 1–5 scale. The counts are 12 ones, 18 twos, 40 threes, 150 fours and 200 fives. The satisfied responses are the 4s and 5s: 150 + 200 = 350. CSAT = 350 ÷ 420 × 100 = 83.3%.

Total means everyone who answered the satisfaction question, including the neutral 3s and the unhappy 1s and 2s. It does not include people who were sent the survey and ignored it; that is your response rate, a separate number. This page is about customer satisfaction scores, not the CSAT paper in India's civil services exam.

The percentage method above is called top-2-box: you count responses in the two highest boxes. It is the most common way to report CSAT and the one this calculator uses. The alternative is to report the average rating, such as "4.2 out of 5", sometimes rescaled to 0–100 as (mean − 1) ÷ (scale max − 1) × 100.

The two can tell different stories. Illustration: team A gets 100 ratings, all 4s. Team B gets 70 fives, 10 threes and 20 ones. Both average exactly 4.0. Team A's CSAT is 100%; team B's is 70%, and a fifth of its customers gave the lowest possible rating. The mean hides the unhappy tail; the percentage shows it.

Report the percentage as the headline and keep the full distribution beside it. If your team has always reported a mean, keep it as a secondary figure rather than switching mid-series.

The American Customer Satisfaction Index is often quoted as if it were a CSAT benchmark. It is a different measurement. ACSI asks three questions on a 1–10 scale (overall satisfaction, how the experience compared with expectations, and how close it came to an ideal), weights them using its own statistical model, and rescales the result to 0–100. The national ACSI score was 76.9 out of 100 for the fourth quarter of 2025, from a sample of about 200,000 customers.

That 76.9 is an index, not a percentage of satisfied customers, so a CSAT of 80% is not "above the national average". The two numbers answer different questions with different arithmetic.

We could not find a cross-industry CSAT percentage benchmark that publishes how it was collected: which question, which scale, which cut-off, and at what moment. Without those, a table of "average CSAT by industry" is not something you can hold your own number against, so we do not print one.

What moves CSAT in practice is the moment you ask. A survey sent after a resolved password reset will score higher than one sent after a billing dispute, for the same team. The useful benchmarks are internal: the same question, the same scale, the same trigger, compared over time and across teams or channels.

CSAT is a simple proportion, so its 95% margin of error is 1.96 × √(p × (1 − p) ÷ n) × 100, with p as the satisfied share in decimals. For the worked example, p = 0.833 and n = 420, which gives about ±3.6 points: the true figure is probably between 79.8% and 86.9%.

Small samples are where teams get fooled. At 80% CSAT, 30 responses give a margin of about ±14 points and 100 responses about ±8. An agent whose CSAT drops from 85% to 75% over 30 tickets has not measurably got worse. This calculator shows the percentage only; paste the same counts into our margin of error calculator, or use the formula, before acting on a change.

With 1–5 ratings in B2:B421, the top-2-box CSAT is =COUNTIF(B2:B421,">=4")/COUNT(B2:B421), formatted as a percentage. The same formula works in Google Sheets. For a 7-point scale change ">=4" to ">=6". DSAT is =COUNTIF(B2:B421,"<=2")/COUNT(B2:B421).

Use COUNT, not COUNTA, so blank and text cells are ignored. If your export stores labels ("Very satisfied") instead of numbers, use COUNTIF on the label text for each satisfied option and add them.

CSAT is built for a specific, recent interaction. It is a weak measure of the overall relationship (NPS is designed for that), and it says little about how hard something was to do (Customer Effort Score is designed for that). It is also easy to game: agents who choose which customers get the survey, or who ask for a "5" at the end of a call, will lift the number without lifting satisfaction. Send the survey automatically after every eligible interaction.

Our comparison of NPS vs CSAT vs CES covers which metric belongs at which moment.

Keep the form to the satisfaction question plus one open follow-up ("What is the main reason for your rating?"). In Zunoform, the rating question supports scales of 5 to 10 points with labels at each end, it is available on every plan, and the form's summary tab shows the distribution of answers, which gives you the satisfied and total counts to enter above. Logic can send low ratings to a "What went wrong?" question.

The Customer Satisfaction (CSAT) template is a short conversational survey built on a 1–5 rating. The service feedback, patient satisfaction and course evaluation templates follow the same pattern for other settings.