Calculate Customer Satisfaction Score (CSAT) by dividing satisfied responses by all valid rated responses, then multiplying by 100. On a five-point satisfaction scale, ratings of 4 and 5 commonly count as satisfied.
CSAT (%) = satisfied responses ÷ total rated responses × 100 Example: 170 satisfied responses ÷ 200 rated responses × 100 = 85%
The result describes the people who answered that survey question. It does not mean that 85% of every customer is satisfied. This guide covers the calculation, a worked example, a spreadsheet formula, combined scores, and common reporting mistakes.
Define what counts as a satisfied response
Use the meaning of your survey’s answer labels. Qualtrics describes the common five-point approach using satisfied and very satisfied responses, represented by 4 and 5. Document your threshold before reporting results and keep it consistent.
| Rating | Example label | Included in rated responses? | Counted as satisfied? |
|---|---|---|---|
| 1 | Very dissatisfied | Yes | No |
| 2 | Dissatisfied | Yes | No |
| 3 | Neither satisfied nor dissatisfied | Yes | No |
| 4 | Satisfied | Yes | Yes |
| 5 | Very satisfied | Yes | Yes |
For a binary survey, use the positive answer as the numerator and both positive and negative rated answers as the denominator. Do not apply a five-point threshold to a different scale without defining what its answers mean. Check your survey tool’s metric definition; Zendesk’s CSAT guide explains percentage-based scoring and survey scales.
Calculate CSAT in four steps
- Choose the scope: identify the survey question, interaction, period, and population being reported.
- Count valid rated responses: include every eligible rating for that question, including neutral and negative ratings.
- Count satisfied responses: apply the documented positive threshold to the same response set.
- Divide and convert: divide satisfied by rated responses and multiply by 100 for a percentage.
Keep unanswered invitations separate. Excluding a blank answer from the CSAT denominator does not mean excluding a dissatisfied customer who supplied a valid rating. Define how you handle duplicate or revised answers, and apply that rule consistently.
Worked five-point CSAT example
Suppose a support team receives these 100 valid ratings. The figures are illustrative.
| Rating | Number of responses |
|---|---|
| 1 | 5 |
| 2 | 10 |
| 3 | 15 |
| 4 | 30 |
| 5 | 40 |
Satisfied responses = 30 + 40 = 70 Total rated responses = 5 + 10 + 15 + 30 + 40 = 100 CSAT = 70 ÷ 100 × 100 = 70%
If 500 invitations were delivered for this same survey and period, the rated-response rate would be 100 ÷ 500 × 100 = 20%. That is a separate measure. The CSAT score remains 70%, not 14%.
Excel or Google Sheets CSAT formula
For a cleaned column of numeric ratings from 1 to 5 in cells B2:B101, use:
=IFERROR(COUNTIFS(B2:B101,">=4",B2:B101,"<=5")/COUNTIFS(B2:B101,">=1",B2:B101,"<=5"),"No rated responses")
Format the result cell as a percentage. The formula returns a fraction such as 0.70, which percentage formatting displays as 70%. Do not also multiply it by 100 when using percentage formatting.
For function syntax, see the official Excel COUNTIFS and Google Sheets COUNTIFS references.
Use one rating per row in this range. Validate that ratings are integers from 1 to 5, convert imported text ratings to numbers, remove ineligible records according to your documented rules, and keep other fields outside the rating column. Expand the range for your data. Some spreadsheet locales use semicolons instead of commas between arguments.
If there are no valid ratings, report “No rated responses” rather than 0% satisfaction. A 0% score means there were rated responses and none met the satisfied threshold.
CSAT percentage is different from an average rating
In the worked example, the mean rating is:
Mean rating = (1×5 + 2×10 + 3×15 + 4×30 + 5×40) ÷ 100 Mean rating = 390 ÷ 100 = 3.9 out of 5
The same survey therefore has 70% CSAT and a 3.9/5 mean rating. These describe different aspects of the responses. Dividing the mean by five gives 78%, but that is a percentage of the maximum rating, not the percentage of satisfied respondents.
Label each measure clearly. Do not switch formulas between reporting periods and treat the results as one continuous CSAT trend.
Combine scores using response counts
When combining compatible surveys, add their satisfied counts and rated-response counts, then calculate the percentage. Do not take a simple average of group percentages when the groups have different numbers of responses.
| Group | Satisfied | Rated responses | CSAT |
|---|---|---|---|
| A | 80 | 100 | 80% |
| B | 5 | 10 | 50% |
| Combined | 85 | 110 | 77.3% rounded |
A simple average of 80% and 50% would be 65%, giving the small group the same weight as the larger one. The pooled response score is 85 ÷ 110 × 100, or approximately 77.3%.
Only pool results when the questions, scale, positive threshold, and scope are compatible. Otherwise report them separately and explain the difference.
Common CSAT calculation mistakes
- Using invitations as the denominator: this mixes satisfaction with participation.
- Removing neutral or negative ratings: this inflates the score and changes the meaning.
- Calling a mean rating percentage CSAT: label the measure and formula accurately.
- Averaging unequal groups: combine the underlying response counts when appropriate.
- Changing the question or survey timing unnoticed: this can break comparability.
- Reporting a percentage without counts: 9/10 and 900/1,000 both produce 90%, but have very different amounts of evidence.
- Treating nonresponse as proof of satisfaction: people who did not answer may differ from respondents.
Report CSAT with enough context to use it
Survey question and interaction: Reporting period and population: Scale and satisfied threshold: Satisfied responses: Total valid rated responses: CSAT percentage: Delivered invitations / rated-response rate, if available: Duplicate and exclusion rules: Changes in survey method or customer mix: Main feedback themes: Action owner and review date:
Compare like-for-like periods and segments, and show the response counts. A move from 80% to 85% is an increase of five percentage points. It does not by itself establish the cause of the change or prove that all customers had a better experience.
Read comments and relevant support cases to understand the score. Record the problem, choose a specific improvement, and check the outcome after making it. For a wider measurement plan, see our customer-service metrics guide.
Turn feedback into a support improvement
If comments repeatedly describe confusing setup instructions, review the actual steps customers struggled with. A clearer explanation or a short tutorial video may be worth testing. If the issue is hard to reproduce, a customer screen recording in Zendesk can provide context for investigation.
Use CSAT alongside operational evidence and customer feedback. A tool or workflow change should be evaluated on its observed results, not assumed to raise satisfaction automatically.
Frequently asked questions
Do neutral responses count in CSAT?
In the five-point example above, neutral ratings are included in all valid rated responses but do not count as satisfied.
What is a good CSAT score?
There is no universal target for every survey and industry. Use comparable benchmarks cautiously, document their method, and compare your own consistent results with comments and response counts.
Should unanswered surveys count as dissatisfied?
No. A missing rating does not tell you the customer’s satisfaction. Keep nonresponse separate and report participation where available.
Can a high CSAT score hide a problem?
Yes. A pooled score can hide a struggling segment, and respondents may not represent everyone. Review the rating distribution, relevant segments, and written feedback.
