How to calculate CSAT, and when to ask for it
Updated
Customer Satisfaction Score asks one customer about one experience, right after it happened. The formula is a single division. What decides whether the number is useful is the timing of the question and whether you ask the unhappy customers what went wrong.
The question
“How satisfied were you with [the delivery / your visit / the help you received]?” answered on a 1–5 scale, from very dissatisfied to very satisfied. Name the specific experience: CSAT measures one interaction, and a vague question gets a vague answer.
The formula
CSAT = (number of 4 and 5 answers ÷ total answers) × 100. The result is a percentage: the share of customers who were satisfied or very satisfied.
This is called the top-two-box method, and it is the usual definition. Some teams report the average score instead; both are legitimate, but they are different numbers, so say which one you use and never compare one against the other.
Send your first CSAT survey today
Start from the CSAT template, share it by link, QR code, or embed, and read scores and reasons in one dashboard. Free for up to 5 surveys and 100 responses each — no credit card.
Get started for freeA worked example
| Score | Meaning | Responses |
|---|---|---|
| 5 | Very satisfied | 96 |
| 4 | Satisfied | 84 |
| 3 | Neutral | 30 |
| 2 | Dissatisfied | 18 |
| 1 | Very dissatisfied | 12 |
| Total | 240 |
CSAT = (96 + 84) ÷ 240 = 75%. The average score for the same answers is 954 ÷ 240 ≈ 3.98.
Why the percentage usually beats the average
An average hides the shape of the answers. A 3.98 could be nearly everyone giving a 4, or a split between delighted and furious customers — two very different businesses. The percentage asks a plainer question, “how many people left happy?”, and moves when that changes.
Whichever you use, also look at the 1s and 2s on their own. In the example, 30 of 240 customers (12.5%) were dissatisfied, and they are where the next improvement is.
When to send it
- Right after the experience: on the receipt, at delivery, at the end of a support conversation.
- One experience per survey, so the answer is about something you can act on.
- Keep it short — the rating plus one optional open question. Every extra question loses respondents.
CSAT vs NPS vs CES
| CSAT | NPS | CES | |
|---|---|---|---|
| Measures | Satisfaction with one interaction | Overall loyalty | How easy something was |
| Typical scale | 1–5 | 0–10 | 1–7 agreement |
| Best sent | Immediately after the interaction | Periodically | After a task: support, returns, signup |
| Answers | “Did that go well?” | “Would they recommend us?” | “Was that hard?” |
Running CSAT in Proterate Surveys
Start from the CSAT template and add conditional logic so anyone who answers 1 or 2 is asked what went wrong, while satisfied customers finish in one tap. Share it by link, by QR code on a receipt or table card, or embedded on your site. Scores and trends land in one dashboard you can filter by date, so you can see whether a fix actually moved the number.
Frequently asked
Is CSAT a percentage or an average?
Usually a percentage: the share of 4 and 5 answers. Some teams report the mean score instead. Both are valid, but they are not interchangeable, so label which one you report.
Should neutral answers count?
Yes, in the total. A 3 is not satisfied, so it lowers the percentage without counting as a complaint.
Can I use a 1–10 scale for CSAT?
You can, but then decide and document which scores count as satisfied, and keep it fixed. Changing the scale breaks comparison with your own past results.
Should I measure CSAT or NPS?
Often both. CSAT tells you whether a specific interaction worked; NPS tells you how customers feel about the business overall. They answer different questions and are sent at different moments.