Customer Success

Customer Satisfaction Score: Calculate and Use CSAT

Calculate customer satisfaction score with a worked example, response coverage and a practical follow-through register. Keep CSAT separate from NPS and retention.

By Finta Editorial TeamPublished November 15, 2026Last reviewed November 15, 2026

A clear aqua glass bowl holds a subset of emerald and amber stones while others remain outside.

What you will learn

  • Define the experience, question and scale behind a CSAT score.
  • Calculate satisfaction and response coverage with different denominators.
  • Translate feedback into owned follow-through without overstating the result.

Customer satisfaction score, or CSAT, summarizes how satisfied respondents say they are with a particular experience, product or service. Calculate it using a stated positive-response rule, then show the question, scale, response count and collection period alongside the percentage. That context helps a team choose what to investigate; the score alone does not establish satisfaction across every customer or predict renewal.

What does CSAT stand for in the context of customer satisfaction?

Twilio’s CSAT definition describes satisfaction in relation to customer expectations. The experience matters. A person can be happy with a support interaction and still dislike the product, or value the product while finding a particular process difficult.

Ask about one recognizable experience at a sensible point after it occurs. Avoid combining speed, accuracy and friendliness into one question if you need to know which dimension to improve. State who was eligible to answer and how repeat responses from the same person will be handled.

A five-point scale can run from very dissatisfied to very satisfied. A common convention counts the top two responses as satisfied. Other instruments use different scales, so preserve the actual method instead of comparing percentages with different definitions.

How do you calculate CSAT?

IBM’s CSAT calculation explanation uses positive responses divided by total collected responses. For the five-point convention used here, ratings of 4 or 5 are positive. Write the rule before looking at the answers.

CSAT = valid satisfied responses ÷ all valid responses × 100. Missing answers are not positive responses. Decide and disclose how invalid or duplicate submissions are excluded. Do not divide by invitations when labeling the result as CSAT; invitations belong to response coverage.

Fictional five-point satisfaction survey

RatingMeaningValid responses
5Very satisfied8
4Satisfied6
3Neither satisfied nor dissatisfied3
2Dissatisfied2
1Very dissatisfied1
TotalAll valid answers20

The satisfied count is 8 + 6 = 14. CSAT is 14 ÷ 20 × 100 = 70%. If 50 eligible people were invited, response coverage is 20 ÷ 50 × 100 = 40%. The remaining 30 invitees have unknown satisfaction. They should not be recorded as satisfied, dissatisfied or neutral.

What does the score tell you?

The fictional score tells you that 14 of 20 valid respondents selected the defined positive ratings. It does not establish the satisfaction of all 50 invitees. Report the underlying distribution, because two surveys can share a headline score while having different numbers of very dissatisfied respondents.

Use the survey-bias lesson to inspect who was invited and who answered. An improved percentage can reflect a changed mix of respondents, a different question or a different collection moment. Compare like with like before describing an improvement.

Keep CSAT separate from recommendation, effort, retention and observed task success. A satisfaction answer does not prove a task was completed accurately, a customer will renew or a business outcome occurred. For wider account evidence, use the customer health score guide.

Build a feedback-to-action register

Suppose the fictional survey concerns onboarding. Several respondents mention unclear preparation instructions. Do not jump from that pattern to “customers hate onboarding.” Preserve the actual issue and test what would help.

Original satisfaction follow-through register

ObservationQuestion to resolveOwner and next evidence
Three responses describe unclear preparationWhich instruction or handoff caused confusion?Onboarding owner reviews the actual instructions and offers a relevant clarification
One response reports a missing appointment linkWas the link absent, expired or not found?Operations checks the delivery receipt and controlled test
The busiest customer segment rarely answeredDoes the collection method exclude people with less time?Research owner reviews invitation and response coverage before generalizing

Try the register with the missing appointment link. First check what the person received, then test the intended route. The useful next action could be repairing a broken destination, clarifying the instructions or correcting a delivery problem. Record the evidence that distinguishes those explanations before choosing the fix.

A proposed change and a completed change are different states. Record the responsible person, the action they accept and what will count as completion. If you contact a respondent, respect the permission and purpose under which the feedback was collected. Avoid exposing personal comments in a broad shared summary.

How can a team improve CSAT responsibly?

  1. Read the distribution and comments before choosing a fix.
  2. Check whether the problem is service delivery, expectation-setting or the measurement itself.
  3. Choose one bounded improvement with an owner and a way to verify completion.
  4. Keep the question and eligible population stable for a comparable follow-up, or disclose any changes.
  5. Review both feedback and actual operational evidence. Do not pressure people to provide a favorable rating.

A customer business review can connect feedback to current goals and decisions. It should not turn the score into a substitute for the customer’s explanation.

Copyable AI prompt

Try this with Aurora AI in Finta, Codex, or Claude. Share only material you are authorized to use, remove unnecessary personal information and review the draft.

From [question and rating scale], [eligibility and invitation count], [authorized de-identified response data] and [collection dates], calculate CSAT using the stated positive-response rule. Show valid, duplicate, invalid and missing counts separately. Calculate response coverage with its own denominator. Summarize themes with supporting observations and propose a follow-through register. Do not infer nonrespondent satisfaction, invent benchmarks, contact respondents or change records.

Keep the evidence attached to the next action

Bring the question, response distribution and follow-through register to your next customer review. Explore Finta’s Customer Voice Researcher and Customer Health Analyst for preparing that discussion from authorized material. Keep survey collection and scoring in their responsible systems, and give each accepted improvement an owner and completion check.

Sources and method. By Finta Editorial Team. Sources checked November 15, 2026. Worked examples and worksheets are original fictional teaching materials, not customer results.