A definition of CSAT: the survey question, the formula, what counts as a good score, and where CSAT falls short. Full disclosure: unitQ publishes this guide and analyzes the verbatims behind survey scores.
CSAT, short for Customer Satisfaction Score, measures how satisfied customers are with a specific interaction or experience, such as a support conversation, a purchase, or an onboarding flow. It is captured with a single survey question, usually “How satisfied were you with X?” on a 1 to 5 scale, and reported as the percentage of respondents who picked one of the top two ratings. A CSAT of 85% means 85 out of every 100 respondents rated the experience a 4 or 5.
The formula, and the choices hiding inside it
The standard calculation is simple:
CSAT = (number of 4 and 5 responses ÷ total responses) × 100
Simple, but every part of it involves a choice. Some teams use a 1 to 7 or 1 to 10 scale. Some count only 5s as satisfied. Some report the mean rating instead of a top-box percentage. None of these variants is wrong, but they produce different numbers from identical customers, which is why comparing your CSAT to another company’s is rarely meaningful unless you know their method.
Timing is the other lever. CSAT is transactional by design; it works best asked immediately after the moment it measures. A satisfaction survey sent two weeks after a support ticket closes measures memory, not experience.
Why CSAT matters
CSAT earns its place for three reasons. It is cheap to run and easy for customers to answer, so response rates beat longer surveys. It attaches to a specific touchpoint, so a falling score points somewhere concrete: this flow, this team, this release. And it produces a trendline that frontline teams can own week over week.
Its limits are just as real. Only a small fraction of customers answer, and the ones who do skew toward strong feelings in either direction. The score also says nothing about cause; a 62% CSAT tells you something is wrong but not what. That is why the open-text comment attached to the rating is often worth more than the rating, and why teams increasingly analyze those comments with the same rigor as the score itself. The customers who never respond at all still leave signal elsewhere, in reviews and tickets, which is a reason not to let a survey metric stand in for the whole listening program.
A worked example
A subscription app’s support team runs CSAT on ticket close. The score sits at 82% for months, then drops to 71% in one week. The count of responses has not moved much, so the team reads the 1- and 2-rated verbatims and finds the same story repeated: refund requests are being answered with a policy macro that does not address the customer’s actual question. The fix is a rewritten macro plus an agent guideline, not a product change. CSAT recovers within two weeks. The score found the smoke; the verbatims found the fire.
See how unitQ compares on your data
A short demo, run on your own feedback.
Related terms
The brand-loyalty metric CSAT is often paired with.
The effort-based companion to a satisfaction score.
What over-asking does to your response rates.
Theming the open-text behind the number.
FAQ
The rating slipped. Find out why.
The rating tells you a touchpoint slipped; the comment underneath tells you why. See how teams read CSAT verbatims at scale in the best NPS, CSAT, and verbatim analysis tools, or look up any app’s free unitQ scorecard.