CSAT measures satisfaction with a specific product, feature, or interaction. It answers "did that just work for you?" better than any other metric.
The question
"How satisfied were you with [specific thing]?"
Answered on a 1-5 or 1-7 scale. Common scales:
- 1-5: Very Dissatisfied, Dissatisfied, Neutral, Satisfied, Very Satisfied
- 1-7: More granular version
- Some teams use 1-10 (aligns with NPS)
The formula
CSAT = %Satisfied Responses out of Total Responses × 100
On a 1-5 scale, "satisfied" typically means 4 or 5. Some methodologies use only the top response (5 out of 5).
Example: 100 responses. 80 answered 4 or 5. CSAT = 80%.
What good looks like
Benchmarks depend on what you are measuring:
- Support ticket CSAT: 80%+ is best-in-class, 70-80% is good
- Product feature CSAT: 60-75% is normal, 75%+ is strong
- Onboarding CSAT: 70-85% is typical
Product-level CSAT tends to be lower than support CSAT because product satisfaction requires the product to work perfectly, while support satisfaction requires the interaction to feel helpful.
When to use CSAT over NPS
Use CSAT for specific moments:
- After a support ticket resolves
- After a feature is used (in-app survey)
- After onboarding step completes
- After billing interaction
- After a new feature release
Use NPS for overall sentiment. CSAT tells you what worked or did not on a specific touchpoint. NPS tells you how customers feel overall.
Common mistakes with CSAT
1. Averaging scores instead of taking percentage satisfied
An average of 4.2 out of 5 sounds good but hides polarization. 50% of respondents scoring 5 and 50% scoring 3.4 has the same average as everyone scoring 4.2, but the customer experience is very different.
2. Only surveying satisfied users
If you only send CSAT surveys to customers who completed something successfully, your CSAT is artificially high. Survey after all interactions, not just successful ones.
3. Not correlating CSAT to downstream behavior
CSAT is only useful if you can act on it. Correlate CSAT scores to churn, expansion, and support ticket volume. Low CSAT should predict problems. If it does not, your survey is not measuring what you think.
CSAT vs NPS vs CES
Three different metrics for different questions:
- NPS: Would you recommend us?
- CSAT: Are you satisfied with [specific thing]?
- CES: How easy was it to accomplish [specific task]?
See NPS vs CSAT vs CES for how to pick which one for each situation.
To score your setup for detecting satisfaction issues before they turn into churn, take the 60-second Churn Health Check.