Measure and test
CSAT calculator
Work out a customer satisfaction score from a count per rating on a 1 to 5, 1 to 7, 1 to 10 or 0 to 10 scale, with the average rating, the margin of error and whether a change is larger than the noise.
Any common scale
A count per rating on 1 to 5, 1 to 7, 1 to 10 or 0 to 10, with the satisfied cut you choose.
Margin of error
How precise your CSAT is for your number of responses.
Compare surveys
Whether a change since your last survey is larger than the noise.
Calculate your customer satisfaction score
Free, no sign-up. The result updates as you type. The calculation runs in your browser: nothing you type here is sent anywhere.
CSAT score
78.3%
188 of 240 responses rated 4 or 5 on the 1 to 5 scale, the answers counted as satisfied: 78.3%.
Average rating: 4.03 out of 5. A separate figure, not the CSAT score.
Margin of error: ±5.2 points at 95% confidence, which is 73.1% to 83.5%. This is the simple formula.
Wilson 95% confidence interval: 72.7% to 83.1%. Where the two differ, use this range: it keeps its confidence level better.
Read it as: ranges built this way contain the score of everyone you could have surveyed in about 95% of surveys. That covers sampling error only, not who chose to answer.
Formula
CSAT = satisfied responses / all responses * 100
average rating = sum of (rating * responses) / all responses
margin of error = z * sqrt(p * (1 - p) / n) * 100
p = CSAT as a fraction, n = all responses, z = 1.96 at 95% confidenceShort answer
A CSAT calculator turns satisfaction survey answers into a customer satisfaction score: the satisfied responses divided by all responses, times 100. On a 1 to 5 scale, satisfied usually means a 4 or 5. Convertica's free CSAT calculator takes a count per rating or two totals and adds the margin of error.
- CSAT = satisfied responses / all responses × 100. Neutral and unhappy answers count in the total.
- Satisfied is a cut point you choose. The usual one is the top two answers, and the score is only comparable with scores that used the same cut.
- The average rating is a separate figure. The example survey scores 78.3% CSAT with an average of 4.03 out of 5.
- Every CSAT has a margin of error: ±5.2 points at 95% confidence for the 240 responses in the example.
- A change between two surveys only means something when it is larger than the noise in both of them.
Compare two CSAT surveys
This quarter against last quarter, or one team against another. Tests whether the gap is larger than sampling noise.
Change in CSAT
+7.5 points
Current 78.3% from 240 responses, previous 70.8% from 240.
Not larger than the noise at 95% confidence: p = 0.0592. The change could be anywhere from -0.3 to +15.3 points.
Margin of error of the change: ±7.8 points. With samples of this size, a move smaller than that cannot be told apart from sampling noise.
z = 1.89, two-sided p-value 0.0592 (two-proportion z-test).
This assumes two separate random samples. If the same people answered both surveys, it does not apply.
Formula
pooled p = (satisfied1 + satisfied2) / (n1 + n2)
z = (p1 - p2) / sqrt(pooled p * (1 - pooled p) * (1 / n1 + 1 / n2))
interval = (p1 - p2) ± z * sqrt(p1 (1 - p1) / n1 + p2 (1 - p2) / n2)How to use the CSAT calculator
Pick the way in that matches the data you have. Both give the same CSAT score for the same survey. Only a count per rating also gives the average rating, because two totals do not say how the answers were spread.
- Choose the scale your question used: 1 to 5, 1 to 7, 1 to 10 or 0 to 10.
- Choose what counts as satisfied. The default is the top two answers. If your earlier reports used another cut, use that one, so the numbers stay comparable.
- Enter the number of responses at each rating, including the neutral and unhappy ones. Or switch to two totals and enter satisfied and all responses.
- Read the score, then the margin of error. The margin tells you how far the score could move if you asked a different sample of the same customers.
- Comparing with an earlier survey? Put both into the second calculator before you call the change a rise or a fall.
What is a CSAT score?
A CSAT score, short for customer satisfaction score, is the percentage of survey responses that chose a satisfied answer to a question such as "How satisfied were you with your order?". It runs from 0% to 100%. It is usually asked straight after one interaction, such as a purchase, a delivery or a support conversation, so it measures that moment rather than the whole relationship.
How do you calculate a CSAT score?
To calculate a CSAT score, count the responses at or above your satisfied cut point, divide by the number of all responses and multiply by 100. On a 1 to 5 scale the usual cut is the top two answers, 4 (satisfied) and 5 (very satisfied). Neutral and unhappy answers are not counted as satisfied but stay in the total.
CSAT = (responses at or above the cut) / (all responses) * 100Worked example: a 1 to 5 survey
A survey gets 240 answers. These are the default values in the calculator, made up to show the arithmetic. They are not a benchmark.
| Rating | Answer | Responses | Share | Counted as satisfied |
|---|---|---|---|---|
| 1 | Very dissatisfied | 9 | 3.8% | No |
| 2 | Dissatisfied | 14 | 5.8% | No |
| 3 | Neutral | 29 | 12.1% | No |
| 4 | Satisfied | 96 | 40.0% | Yes |
| 5 | Very satisfied | 92 | 38.3% | Yes |
- Satisfied: 96 fours + 92 fives = 188.
- All responses: 240, including the 52 answers of 1 to 3.
- CSAT: 188 / 240 × 100 = 78.3%.
Is CSAT the same as the average rating?
No. The CSAT score is the share of responses at or above the satisfied cut, as a percentage. The average rating is the mean of every rating, on the scale itself. They answer different questions: how many customers were satisfied, and how satisfied the typical answer was. The calculator shows both, and a report should say which one it uses.
For the example: (1 × 9 + 2 × 14 + 3 × 29 + 4 × 96 + 5 × 92) / 240 = 968 / 240 = 4.03 out of 5, beside a CSAT of 78.3%. An average can move while the CSAT stays put, for example when unhappy customers become neutral: both answers sit below the cut, so the score does not change.
How do you calculate CSAT on a 10-point scale?
On a 1 to 10 or 0 to 10 scale, CSAT is calculated the same way: choose which ratings count as satisfied, count them and divide by all responses. There is no single fixed cut on a 10-point scale, and the choice moves the score a long way, so state it every time you report the number.
Take a made-up survey of 100 answers on a 1 to 10 scale, with 20 tens, 24 nines, 20 eights and 14 sevens:
- Counting 9 and 10 as satisfied (the top two): 44 / 100 = 44.0%.
- Counting 8 to 10: 64 / 100 = 64.0%.
- Counting 7 to 10: 78 / 100 = 78.0%.
Same answers, three scores. The average rating, 7.73, does not depend on the cut. Pick the cut once, write it down beside the score, and keep it for every survey you want to compare. The calculator's "Counts as satisfied" menu does this for 1 to 5, 1 to 7, 1 to 10 and 0 to 10.
What is the margin of error of a CSAT score?
The margin of error of a CSAT score is how far the score could sit from the true figure for all your customers, through the luck of who answered. A CSAT is one percentage, so it uses the standard formula for a proportion. The example survey of 240 responses scores 78.3% with a margin of error of ±5.2 points at 95% confidence.
standard error = sqrt(p * (1 - p) / n)
margin of error = z * standard error * 100Worked example: the margin of error on 240 responses
- Share: p = 188 / 240 = 0.7833, so 1 - p = 0.2167.
- p × (1 - p) = 0.1697, and divided by 240 that is 0.000707.
- Standard error: √0.000707 = 0.0266.
- Margin of error at 95%: 1.96 × 0.0266 = 0.0521, or ±5.2 points.
- So the CSAT is 78.3% ± 5.2: anywhere from 73.1% to 83.5%.
| Responses | At the example score (78.3%) | Widest case (a score of 50%) |
|---|---|---|
| 50 | ±11.4 | ±13.9 |
| 100 | ±8.1 | ±9.8 |
| 200 | ±5.7 | ±6.9 |
| 400 | ±4.0 | ±4.9 |
| 1,000 | ±2.6 | ±3.1 |
| 2,500 | ±1.6 | ±2.0 |
Read across the 100 row: a CSAT near 78.3% from 100 responses is good to about ±8.1 points. Two such surveys could land several points apart with nothing having changed. The margin of error calculator does the same arithmetic for any survey percentage, with a finite population correction for small customer bases.
Why the calculator also shows a Wilson interval
The simple formula is a normal approximation. With few responses, or a score close to 100%, it is too confident, and it can even run past 100%: 17 satisfied out of 20 is 85.0% ± 15.6, an upper end of 100.6%. The Wilson score interval, from Edwin B. Wilson's 1927 paper in the Journal of the American Statistical Association, stays inside 0% to 100% and keeps close to its stated confidence level. For the same 20 answers it gives 64.0% to 94.8%.
The table shows how often each "95%" interval really contains the true score. Nothing in it is simulated: for each row, every possible survey result of that size is listed, and the probabilities of the ones whose interval contains the true share are added up.
| True CSAT | Responses | Simple interval | Wilson interval |
|---|---|---|---|
| 80% | 10 | 88.6% | 96.7% |
| 80% | 20 | 92.1% | 95.6% |
| 80% | 50 | 93.8% | 95.1% |
| 80% | 100 | 93.3% | 94.1% |
| 80% | 200 | 94.1% | 95.8% |
| 95% | 10 | 40.0% | 91.4% |
| 95% | 20 | 63.9% | 92.5% |
| 95% | 50 | 92.0% | 96.2% |
| 95% | 100 | 87.7% | 96.6% |
| 95% | 200 | 92.6% | 96.7% |
When the true CSAT is 80%, with 20 responses the simple interval is right 92.1% of the time and the Wilson interval 95.6%. Near the top of the scale the simple one gets worse: 92.0% at 50 responses when the true score is 95%. By 200 responses at 80% the two are close (94.1% against 95.8%). The confidence interval calculator covers both methods for any proportion.
Either interval covers sampling error only. It assumes the people who answered are a random sample of the customers you care about. It cannot correct for who chose to reply, or for a survey shown only after a support ticket was closed. Those gaps can be larger than the margin of error, and no formula on this page measures them.
Is a change in CSAT real, or noise?
A change in CSAT is more than noise when a two-proportion test on the two surveys gives a p-value below your threshold, 0.05 at 95% confidence. The margin of error of a change is wider than either survey's own margin, because both scores carry sampling noise. In the example, a rise from 70.8% to 78.3% on 240 responses each is not larger than the noise.
Worked example: 70.8% last quarter, 78.3% this quarter
Both surveys have 240 responses: 170 satisfied last quarter and 188 this quarter. These are the default values in the second calculator, made up to show the arithmetic.
- Difference: 78.3% - 70.8% = 7.5 points (0.0750).
- Pooled share, both surveys together: (188 + 170) / (240 + 240) = 0.7458.
- Standard error under no change: √(0.7458 × (1 - 0.7458) × (1/240 + 1/240)) = 0.0397.
- z = 0.0750 / 0.0397 = 1.89, a two-sided p-value of 0.0592. That is above 0.05.
- Interval for the change, from each survey's own spread (standard error 0.0396): 1.96 × 0.0396 = ±7.8 points, so -0.3 to +15.3 points. It includes 0.
This is the same test an A/B test uses on two conversion rates, from the same code as the statistical significance calculator. The table keeps the same two scores and only raises the number of responses.
| Responses in each survey | Margin of error of the change | Interval for the change (points) | p-value | Verdict |
|---|---|---|---|---|
| 240 | ±7.8 | -0.3 to +15.3 | 0.0592 | Inside the noise |
| 480 | ±5.5 | +2.0 to +13.0 | 0.0076 | Larger than the noise |
| 960 | ±3.9 | +3.6 to +11.4 | 0.0002 | Larger than the noise |
The same move clears the noise at 480 responses in each survey. Before a change in CSAT goes into a report, check that it is larger than the noise, that both surveys asked the same question on the same scale with the same cut, at the same point in the customer's journey, and that they are separate samples. If the same people answered both times, this test does not apply.
How many responses do you need for a CSAT survey?
The number of responses a CSAT survey needs depends on the margin of error you can accept. In the widest case, a score near 50%, a margin of ±10 points at 95% confidence takes 97 responses and ±5 points takes 385. A score further from 50% needs fewer. Halving the margin takes four times as many.
responses needed = z^2 * p * (1 - p) / (margin of error / 100)^2 rounded up
use p = 0.5 when you have no earlier survey to go on| Margin of error | Widest case (a score of 50%) | At the example score (78.3%) |
|---|---|---|
| ±10 points | 97 | 66 |
| ±5 points | 385 | 261 |
| ±3 points | 1,068 | 725 |
| ±2 points | 2,401 | 1,630 |
| ±1 point | 9,604 | 6,520 |
These are completed responses, not surveys sent. With fewer, the score is still worth reading: report it with its interval and read the comments.
What is a good CSAT score?
A good CSAT score is one that is higher than your own last reading by more than the margin of error, measured the same way. There is no universal pass mark, and this page does not print one. A score depends on the question, the scale, the cut point, who was asked and when, so a figure from another company or an industry table is rarely a fair comparison.
- Your own baseline is the comparison you control: the same question, scale and cut, sent at the same moment to the same kind of customer.
- The margin of error tells you whether a step in that trend is a step at all. Use the comparison calculator, not two margins read by eye.
- The cut point has to match. A top-two-box score and a top-box score of the same survey can be far apart.
- Segments can differ where the total hides it: one product against another, or chat against email support. Segments are smaller samples, so test each gap.
CSAT vs NPS vs CES: what is the difference?
CSAT, NPS and CES are three survey measures that ask different questions. CSAT asks how satisfied someone was with one experience. NPS asks how likely they are to recommend you. CES, customer effort score, asks how easy something was to get done. Each is calculated differently, so the numbers cannot be swapped or averaged together.
| Measure | Question it asks | Usual scale | How it is calculated | Reported as |
|---|---|---|---|---|
| CSAT | How satisfied were you with this? | 1 to 5, sometimes 1 to 7 or 1 to 10 | Satisfied responses / all responses × 100 | A percentage, 0% to 100% |
| NPS | How likely are you to recommend us? | 0 to 10 | % promoters (9 or 10) - % detractors (0 to 6) | A whole number, -100 to 100 |
| CES | How easy was it to do this? | Varies; 1 to 7 is common | The average rating, or the share who agreed it was easy | An average or a percentage |
For a Net Promoter Score, with its own margin of error and comparison test, use the NPS calculator. An NPS can need up to four times the responses of a single percentage such as CSAT for the same margin of error in points, because it is the difference between two shares of the same sample.
How to calculate CSAT in Excel or Google Sheets
To calculate CSAT in Excel or Google Sheets, count the ratings at or above your cut, divide by the count of all ratings and multiply by 100. With one rating per row in A2 to A241 on a 1 to 5 scale, these formulas work in both.
| You want | Formula |
|---|---|
| Satisfied | =COUNTIF(A2:A241,">=4") |
| All responses | =COUNT(A2:A241) |
| CSAT | =100*COUNTIF(A2:A241,">=4")/COUNT(A2:A241) |
| Average rating | =AVERAGE(A2:A241) |
| Margin of error at 95% | =100*NORM.S.INV(0.975)*SQRT((D1/D2)*(1-D1/D2)/D2) |
Use COUNT for the total, not the number of rows: it skips blank cells and text, so unanswered rows do not lower the score. For another cut, change the 4. For 90% confidence replace 0.975 with 0.95, and for 99% use 0.995.
What does CSAT have to do with conversion rate?
CSAT and conversion rate are built the same way: a count of one outcome divided by a total. That means the same statistics apply. A conversion rate has a margin of error, and a change in it needs the same two-proportion test as a change in CSAT. What they measure is different. Conversion rate counts what visitors did before buying; CSAT records what customers said afterwards.
- Satisfied customers tell you what to say. The comments behind a 5 are benefits in the customer's own words, the raw material for a headline or a product page. The conversion copywriting guide covers how to use them.
- Unhappy answers tell you what to fix or test. A complaint that repeats, such as a delivery cost that came as a surprise, is a hypothesis for an A/B test on the page where it started.
A CSAT survey only reaches people who became customers, so it says nothing about the visitors who left before buying. Convertica's free CRO audit looks at the page they left: eight checks, each scored out of 100, and three fixes to make first.
Related calculators
CSAT is one survey statistic among several. These tools cover the ones next to it.
See all free tools and how they fit together.
CSAT calculator FAQ
What does a CSAT calculator do?
A CSAT calculator turns answers to a customer satisfaction question into a CSAT score: the percentage of responses that chose a satisfied answer. This one takes a count for each rating on a 1 to 5, 1 to 7, 1 to 10 or 0 to 10 scale, or two totals, and also gives the average rating, the margin of error and a test for change.
How do you calculate a CSAT score?
Divide the number of satisfied responses by the total number of responses and multiply by 100. On a 1 to 5 scale, satisfied usually means a 4 or 5. With 188 satisfied answers out of 240, the CSAT score is 188 / 240 × 100 = 78.3%.
How do you calculate a CSAT score on a 10-point scale?
Choose which ratings count as satisfied, count them, and divide by all responses. With the top two answers, 9 and 10, a survey of 100 responses with 44 of them at 9 or 10 scores 44.0%. Counting 8 to 10 as satisfied, the same survey scores 64.0%, so always say which cut you used.
Is CSAT the same as the average rating?
No. CSAT is the share of responses at or above the satisfied cut point, reported as a percentage. The average rating is the mean of all the ratings, reported on the scale itself. The example survey has a CSAT of 78.3% and an average rating of 4.03 out of 5. The calculator shows both.
Do neutral responses count in CSAT?
Yes, in the total. A neutral answer, a 3 on a 1 to 5 scale, is not counted as satisfied, but it is part of the number of responses the satisfied count is divided by. Leaving neutral answers out of the total makes the score look higher than it is.
What is a good CSAT score?
There is no universal number, and this page does not give one. A score depends on the question, the scale, the cut point, who was asked and when, so another company's figure is rarely comparable with yours. The useful comparison is your own previous score, measured the same way, and whether the change is larger than the margin of error.
Is 80% CSAT good?
It depends on what you compare it with. 80% means four in five responses chose a satisfied answer. Whether that is better or worse than before depends on your own earlier surveys run the same way, and on the margin of error: with 100 responses, 80% is good to about ±7.8 points at 95% confidence.
What is the margin of error of a CSAT score?
The margin of error of a CSAT score is the range around the score that sampling alone can explain. For the 240 responses in the example it is ±5.2 points at 95% confidence, so a CSAT of 78.3% could be anywhere from 73.1% to 83.5%. It shrinks with the square root of the number of responses.
How many responses do I need for a CSAT survey?
Up to 97 responses for a margin of error of ±10 points at 95% confidence, and up to 385 for ±5 points. Those are the widest case, a score near 50%: a score further from 50% needs fewer. Halving the margin of error takes four times the responses.
How do I know if a change in CSAT is statistically significant?
Run a two-proportion test on the satisfied and total counts of both surveys. If the p-value is below 0.05, the change is larger than sampling noise at 95% confidence. In the example on this page, a rise from 70.8% to 78.3% on 240 responses each is not (p = 0.0592).
What is the difference between CSAT, NPS and CES?
CSAT asks how satisfied someone was and reports the share who chose a satisfied answer. NPS asks how likely they are to recommend you on a 0 to 10 scale and subtracts the percentage of detractors from the percentage of promoters. CES asks how easy something was to do, and is usually reported as an average or a share who agreed.
