How to Calculate CSAT: Formula, Score, and Examples

How to calculate CSAT: divide the number of valid ratings that meet your stated satisfaction cutoff by the total number of valid rating responses, then multiply by 100. This customer satisfaction score formula gives you a CSAT percentage you can explain: the question asked, the ratings counted as satisfied, the response count, and the period or touchpoint covered. You’ll finish with a percentage you can explain: the question asked, the ratings counted as satisfied, the response count, and the period or touchpoint covered.

Customer satisfaction score formula: count satisfied responses, not average points

CSAT (%) = satisfied responses ÷ valid rating responses × 100. “Satisfied responses” means the ratings that meet the positive cutoff you have chosen. For a common 1–5 approach, that means ratings of 4 and 5. This result is a percentage, not the average of all the ratings.

For example, imagine a survey uses a 1–5 scale and your rule counts 4s and 5s as satisfied. Count those positive ratings, divide by all valid ratings submitted, then multiply by 100. If you use a different scale or cutoff, write that rule down before doing the calculation.

The distinction from an average matters. An average combines the points customers selected and answers a different question: what was the mean rating? CSAT’s positive-response percentage answers how many valid responses met the threshold you defined. Two surveys can have the same average but different shares of positive ratings, or the same CSAT percentage but different averages.

Neither number explains everything customers felt. CSAT captures satisfaction with a particular experience or touchpoint. Use it alongside the context of the interaction, and avoid treating the score as a complete measure of customer loyalty or overall relationship health.

Step 1: Choose one question, scale, and scoring rule

Choose a clear question about one identifiable experience, such as a purchase or a support interaction. Select the response scale and decide which ratings count as satisfied before collecting results. A focused question and a consistent rule make the score easier to interpret and compare.

A practical question is: “How satisfied are you with your recent support interaction?” For a purchase survey, ask about the purchase experience instead. Avoid combining several topics in one question. If a customer rates the purchase, delivery, product quality, and support all at once, a low score does not tell you which part needs attention.

For a 1–5 scale, label every option so customers know what each number means. One set of labels is:

• 1 — Very Unsatisfied
• 2 — Unsatisfied
• 3 — Neutral
• 4 — Satisfied
• 5 — Very Satisfied

A common scoring rule counts 4s and 5s as satisfied. Some teams choose a stricter rule and count only 5s. Either can be useful for a particular purpose, but they produce different percentages from the same responses. State the cutoff beside the result and keep it consistent when comparing periods or teams.

Also record the exact wording of the question. Changing “How satisfied are you with your support interaction?” to “How satisfied are you with our company?” changes what the answer is about. Keep the question stable when you want to track the same experience over time. If you need to change it, note when the wording changed rather than treating the results as a perfectly continuous series.

Step 2: Set the denominator and clean the response set

Use completed, valid answers to the rating question as the denominator—not the number of survey invitations sent. Exclude unanswered rating questions from the calculation, and decide how to handle partial submissions before reviewing the result. Keep that rule consistent and report the valid response count alongside the percentage.

An invitation is an opportunity to respond, not a rating. If you sent a survey to customers and only some submitted a score, the denominator for CSAT is the set of valid ratings received. Dividing by invitations instead would mix response behavior with customer satisfaction and could make the result hard to interpret.

For a survey with a rating question and optional follow-up text, a response with a valid rating can count even if the person leaves the text blank. The text answer is not required to calculate the rating percentage. If the rating itself is blank or unusable, it cannot be classified as satisfied or unsatisfied, so exclude it from the denominator under a clearly documented rule.

Partial submissions need a consistent policy. If someone answers the rating question but skips a later optional question, the rating remains usable for CSAT. If the person starts the survey but does not provide a rating, there is no rating to include. State how you treat incomplete responses in your reporting notes, especially if your survey form changes.

Always show the count of valid ratings. A percentage without a count hides how much feedback it represents. Also remember that response patterns can be uneven: people who choose to answer may not represent everyone who received the survey. A low response count or a group of respondents drawn mostly from one customer segment can limit what you should infer from the result.

Step 3: Calculate CSAT percentage on a 1–5 scale with an example

For a 1–5 scale using the 4–5 cutoff, count every valid rating of 4 or 5, divide that count by all valid ratings, and multiply by 100. For example, 85 positive responses among 100 valid ratings gives a CSAT score of 85%: 85 ÷ 100 × 100.

Write out the pieces before calculating:

• Question: How satisfied were you with your support interaction?
• Scale: 1–5, with 4 and 5 counted as satisfied.
• Satisfied responses: 85.
• Valid rating responses: 100.
• Calculation: 85 ÷ 100 × 100.
• Result: 85% CSAT.

The example makes the denominator explicit: it is the number of valid ratings, not the number of customers contacted. The cutoff is explicit too, so another person can reproduce the calculation and understand what the percentage represents.

If you use a spreadsheet, count the valid rating responses and the responses meeting your cutoff in separate cells. Divide the positive count by the valid count and format the result as a percentage. Check the underlying counts as well as the displayed percentage: a spreadsheet can format a decimal incorrectly or include blank cells in a formula you did not intend to use.

Keep the numerator and denominator in the report, not just in a working file. A score written as “85% CSAT (n=100; 4–5 ratings)” is more informative than “85%.” Here, “n” indicates the number of valid rating responses. Include the timeframe and touchpoint too, such as the survey period and whether the question followed support or a purchase.

Step 4: Calculate 1–10 CSAT only after defining positive ratings

A 1–10 scale does not define its own CSAT cutoff. Decide which ratings count as positive before comparing results; one approach counts 9s and 10s, while another may include 8s. Then divide the ratings meeting your chosen rule by all valid ratings and multiply by 100.

The highest scores on a 10-point scale indicate very satisfied customers, but that does not settle which ratings belong in the numerator for your business. A stricter cutoff produces a smaller positive-response group than a more inclusive one. Neither should be presented as a universal rule without explaining the choice.

Write the scoring definition next to the question and use it consistently. For instance, a report could state that it counts 9–10 as positive. If you change to an 8–10 cutoff, mark that change clearly. Otherwise, a change in the percentage might reflect a scoring-rule change rather than a change in customer sentiment.

Also consider whether a 10-point scale serves the decision you need to make. A longer scale offers more response options, but the calculation still depends on a clear threshold. If customers or teammates interpret the options differently, clarify the labels and keep the scoring rule visible in reports.

Step 5: Report sample size, timing, and what the score covers

A useful CSAT report pairs the percentage with the valid response count, survey window, question, scoring cutoff, and touchpoint measured. Send transactional surveys while the experience is still fresh, and keep feedback about a specific interaction separate from a survey about the broader customer relationship.

A report might say: “Support-interaction CSAT: 85% (n=100 valid ratings; 4–5 counted as satisfied; survey period stated).” For a repeatable process, a ticketing system for small teams can help keep support interactions organized, while the survey and CSAT calculation remain separate. Replace the general period wording with the actual dates or reporting window used by your team. The percentage alone does not tell a reader whether it reflects a support conversation, a purchase, onboarding, or an overall view of the business.

For transactional feedback, send the question after the interaction you want to measure. A survey after a support exchange asks about that exchange; one after a purchase asks about the purchase experience. An event-based trigger can help connect the survey to the relevant moment. The timing should make sense for the experience, so customers can recall what they are rating.

Relationship surveys serve a different purpose. They ask about broader satisfaction with the business rather than one recent touchpoint. Keep those results separate from transactional CSAT so a strong support interaction does not get confused with an overall relationship measure, or vice versa.

Look at trends only when the question, scale, cutoff, and survey timing are comparable. If one period asks immediately after support and another asks much later, the results may not describe the same conditions. If the invitation audience changes, that can affect who responds. Note such changes alongside the score rather than quietly treating the figures as directly comparable.

Use the comments and operational context to decide what to investigate next. A percentage can flag a shift, but it will not identify the cause on its own. When support conversations are part of the process you are measuring, a support desk such as momo may fit the wider workflow: it answers from a business’s own content, cites answers when confident, and opens a ticket for a human when it is not sure. Teams comparing live chat customer service workflows can consider how the interaction connects to their feedback process. It is not a CSAT calculator, so the survey and score still need to be handled separately.

Troubleshoot distorted scores and compare the right metric

When CSAT changes, check the question, cutoff, denominator, timing, and respondent mix before deciding that satisfaction itself changed. The percentage is a useful summary, but it cannot explain why a customer selected a rating. Add a short optional follow-up question when you need that context.

The cutoff changed. Check whether the positive-rating rule stayed the same. Moving from a 4–5 rule to a 5-only rule will change the numerator even if no customer’s rating changes. Label the new rule and avoid presenting the results as a like-for-like comparison without noting the change.

The denominator includes invitations or blanks. Recalculate using valid ratings. Review whether unanswered rating questions or incomplete forms entered the total. Keep the inclusion policy the same across periods and show the response count.

The survey timing changed. Compare when customers received the question. A survey sent soon after support is measuring a fresh interaction; a survey sent later may be answered with other experiences in mind. Note timing changes and avoid merging separate touchpoints into one score.

Few people responded, or the respondents differ. Look at who answered and who did not. Responses may over-represent customers who are especially engaged, whether positively or negatively. Report the count, and be careful about drawing broad conclusions from a small or uneven response group.

The percentage has no explanation. Add an optional open-text question such as “What is the main reason for your rating?” Keep it short and optional so the rating remains easy to submit. Review the responses for recurring problems, then connect those themes to the relevant purchase or support process.

The team is comparing different metrics. CSAT measures satisfaction with a company, product, service, or specific experience. NPS measures willingness to recommend. Customer Effort Score measures how easy it was to complete a task or get an answer. An average rating is a separate summary of the scale points, not the same calculation as the share of satisfied responses. To measure customer satisfaction, define the experience and survey question first, then calculate CSAT from valid ratings using a stated cutoff.

These measures can complement one another, but they are not interchangeable. Use CSAT to understand satisfaction with a defined interaction or experience; use NPS when the question is recommendation likelihood; and use CES when you want to examine effort. Pair the metric with the question and context it was designed to capture.

Frequently asked questions

Should unanswered CSAT surveys count in the denominator?

No. The denominator is the number of valid rating responses, not the number of survey invitations. Exclude unanswered rating questions. If a survey includes optional follow-up questions, a valid rating can still count when the customer skips the optional text. Decide how to handle partial submissions and apply that rule consistently.

What is a good CSAT score, and does 80% mean the same thing in every industry?

A score’s meaning depends on the experience measured, the survey audience, and the scoring rule. An 80% result means that 80% of the valid ratings met your defined positive cutoff; it does not automatically mean the same thing across different businesses or touchpoints. Compare like with like over time, and interpret the score alongside customer comments and relevant operational context.

How do you calculate CSAT and NPS, and is CSAT the same as average customer rating?

CSAT and NPS are calculated differently: CSAT is the percentage of valid responses that meet a chosen positive-rating cutoff, while NPS measures willingness to recommend. An average rating is the mean of the points customers selected. These answer different questions, so label the figure you report and do not call an average a CSAT percentage.

How often should a business send a relationship CSAT survey?

A relationship survey asks about overall satisfaction rather than one recent interaction. Choose a cadence that fits your customer relationship and the decisions you need to make, and avoid confusing it with transactional feedback after support or a purchase. Keep the timing consistent enough that you can interpret changes in responses.

Can a CSAT score explain why customers are satisfied or dissatisfied?

Not by itself. The percentage shows how many ratings met the cutoff, but it does not reveal the reasons behind them. Add an optional open-text question, review comments for recurring themes, and connect those themes to the experience being measured.

Put the calculation into practice

Write down the question, scale, positive cutoff, valid-response rule, survey period, and touchpoint before sharing a score. If you are also reviewing how support conversations are handled, you can try momo free.