CSAT score guide: how to calculate, read, and improve customer satisfaction
CSAT (Customer Satisfaction Score) measures how satisfied customers are with a specific interaction, calculated as: CSAT = satisfied responses ÷ total responses × 100. On the standard 1–5 scale, 'satisfied' means the top two boxes (4s and 5s). For example, 150 satisfied responses out of 200 total gives a CSAT of 75%. It is a transactional metric — send it immediately after the interaction you want to measure.
What is a CSAT score?
CSAT is the percentage of customers who say they were satisfied with a specific experience — a support ticket, a delivery, an onboarding call, a store visit. The question names the interaction directly: 'How satisfied were you with your support experience today?' Respondents answer on a scale, most commonly 1–5 from 'very dissatisfied' to 'very satisfied'.
Because it targets one moment rather than the whole relationship, CSAT is precise: a low score points at a specific, fixable touchpoint. That is also its limit — a customer can rate one interaction highly and still be unhappy with you overall, which is why teams pair CSAT with a relationship metric like NPS.
How do you calculate a CSAT score?
CSAT = (number of satisfied responses ÷ total responses) × 100
On a 1–5 scale, 'satisfied responses' are the top two boxes: ratings of 4 and 5. Everything else — 1, 2, and 3 — counts only in the denominator.
Worked example
You receive 200 responses to a post-support survey: 90 customers rate it 5, 60 rate it 4, 30 rate it 3, 12 rate it 2, and 8 rate it 1.
- Satisfied responses = 90 + 60 = 150
- Total responses = 200
- CSAT = 150 ÷ 200 × 100 = 75%
Note that the 30 neutral ratings (3s) drag the score down even though those customers were not dissatisfied. That is intentional: top-two-box scoring sets the bar at clearly satisfied, not merely not-unhappy.
Scale mapping at a glance
| Rating (1–5 scale) | Label | Counts as satisfied? |
|---|---|---|
| 5 | Very satisfied | Yes |
| 4 | Satisfied | Yes |
| 3 | Neutral | No |
| 2 | Dissatisfied | No |
| 1 | Very dissatisfied | No |
What is a good CSAT score?
There is no single universal benchmark — CSAT varies by industry, touchpoint, and how the question is worded. As broad practitioner guidance: scores in the 75–85% range are generally considered healthy for most customer-facing teams, scores above that are strong, and anything trending downward matters more than the absolute level. Support interactions usually score differently from deliveries or onboarding, so benchmark each touchpoint against its own history rather than against each other or published figures.
When should you send a CSAT survey?
Immediately after the interaction, while the memory is fresh. Waiting even a day or two blurs the feedback and lowers response rates. Common triggers:
- Support ticket marked resolved → survey within the closing email or right after
- Order delivered → survey the same day the package arrives
- Onboarding or training completed → survey at the final session
- In-product flow finished → one-question prompt right in the app
Keep transactional CSAT surveys to exactly two items: the 1–5 rating and one optional open comment ('What's the main reason for your rating?'). Five-second surveys get answered; anything longer becomes homework and your response rate — and representativeness — drops.
How do you improve a CSAT score?
- 1Read every 1 and 2 with the interaction record attached — low scores almost always cluster around a few root causes
- 2Tag comment themes (wait time, unclear answer, handoffs, product defect) and fix the biggest theme first
- 3Close the loop: reply personally to dissatisfied customers; recovery itself lifts future scores
- 4Watch the 3s — neutrals are the cheapest group to convert into 4s and 5s
- 5Re-measure after each fix so you can tie score movement to specific changes
What are the limits of CSAT?
Three to keep in mind. First, response bias: very happy and very unhappy customers answer more often than the quiet middle, so your score reflects the vocal ends. Second, it is a snapshot of one moment, not loyalty — pair it with NPS for the relationship view. Third, scores are sensitive to question wording and scale choice, so never change the question text mid-stream if you want comparable trends. For running the surveys themselves, Formkii's free survey maker includes rating-scale questions, real-time charts, and CSV export, which is all CSAT requires.
Frequently asked questions
What is the CSAT formula?
CSAT = satisfied responses ÷ total responses × 100. On the standard 1–5 scale, satisfied responses are the ratings of 4 and 5 (the top two boxes). The result is a percentage from 0 to 100.
Do 3s count as satisfied in CSAT?
No. On a 1–5 scale, a 3 is neutral and only counts in the total, not as a satisfied response. Top-two-box scoring deliberately sets the bar at clearly satisfied (4 or 5), not merely neutral.
What is a good CSAT score?
It varies by industry and touchpoint, but as general guidance scores around 75–85% are considered healthy for most customer-facing teams. Your own trend per touchpoint is a more reliable benchmark than published averages, which use inconsistent methods.
What is the difference between CSAT and NPS?
CSAT measures satisfaction with one specific interaction and is reported as a percentage. NPS measures overall loyalty by asking how likely someone is to recommend you, and is scored from −100 to +100. CSAT is transactional; NPS is relational — most teams run both.
Can you use a 1–7 or 1–10 scale for CSAT?
Yes, some teams do, counting the top two boxes (6–7 or 9–10) as satisfied. The 1–5 scale is the most common because it is fast to answer and easy to label. Whatever scale you pick, keep it constant so your trend stays comparable.