Customer Satisfaction Score (CSAT)
CSAT measures how satisfied a customer was with a specific interaction — a support contact, a delivery, a branch visit, an onboarding. It is asked immediately afterwards, while the experience is fresh, usually as a single rating question.
CSAT = (Number of satisfied responses ÷ Total responses) × 100
On a 1–5 scale, "satisfied" conventionally means the top two boxes (4 and 5). On a 1–7 or 1–10 scale the threshold shifts, so the scale has to be stated whenever the number is reported — a CSAT of 80% means nothing without it.
Where it fits among the three
| Metric | Asks about | Timing |
|---|---|---|
| CSAT | Satisfaction with this interaction | Immediately after |
| NPS | Likelihood to recommend, overall relationship | Periodic, or after a key moment |
| CES (Customer Effort Score) | How hard the customer had to work | Immediately after |
They are complementary, not substitutes. CSAT is the most useful of the three for operational improvement, because it attaches to a specific interaction you can go and look at.
Getting it right
- Ask immediately. CSAT asked a week later measures memory, not experience.
- Ask about one thing. "How satisfied were you with this interaction" is answerable. "How satisfied are you with us" is not CSAT, it is a relationship question.
- Keep it to one question plus free text. Response rates collapse with length, and the free text is where the actionable information lives.
- Segment before you average. CSAT by issue type, channel, team and agent tells you where to act. A single company-wide figure does not.
- Read it with resolution. A satisfied customer whose issue was not resolved is usually being polite about the agent. CSAT and FCR together are far more informative than either alone.
The non-response problem
Typical CSAT response rates are low, and the customers who respond are not a random sample — strongly satisfied and strongly dissatisfied customers both over-respond. This matters most when comparing periods: a change in CSAT can reflect a change in who answered rather than a change in service.
Watching the response rate alongside the score, and treating large month-on-month swings with suspicion, costs nothing and prevents a lot of wrong conclusions.
Avant One℠ CX attaches satisfaction responses to the case they came from, so a score can be traced to the interaction, the agent and the issue type rather than sitting in a separate survey tool.
Frequently asked questions
What does CSAT stand for?
Customer Satisfaction Score.
How is CSAT calculated?
Satisfied responses divided by total responses, times 100. On a 1–5 scale, "satisfied" normally means ratings of 4 or 5.
What is the difference between CSAT and NPS?
CSAT measures satisfaction with a specific interaction. NPS measures likelihood to recommend, which reflects the overall relationship.
When should CSAT be asked?
Immediately after the interaction. Delay turns it into a memory test.
What is a good CSAT score?
Benchmark against your own segmented trend. Scores move with sector, channel, issue type and the rating scale used, so a cross-industry number tells you nothing about your own operation.