Customer satisfaction score (CSAT) explained. What it measures and how to use it.

The short answer. What CSAT actually tells you.

Customer satisfaction score, almost always abbreviated to CSAT, is the percentage of customers who rated an interaction positively on a short survey. The question is typically "How satisfied were you with [this interaction]?" answered on a scale of one to five or one to ten. Responses in the top two ratings (four or five on a five-point scale, or seven to ten on a ten-point scale) count as satisfied. Divide those by total responses and multiply by 100: that is your CSAT score.

It is one of three widely used customer experience metrics, alongside NPS (Net Promoter Score, developed by Fred Reichheld at Bain & Company and published in the Harvard Business Review in 2003) and CES (Customer Effort Score, which asks how easy an interaction was to complete). Each measures something different. CSAT is the most operationally specific of the three. It is tied to a single event, a support ticket closed, an onboarding session delivered, a product updated, which makes it useful for diagnosing exactly where an experience broke down and where it is working.

Why it matters.

The main reason to track CSAT is early warning. A consistently falling score at a specific touchpoint, say, post-support interactions or at the end of month one of a new client engagement, shows you where the experience is breaking down before churn confirms it. At that point, you still have time to act. By the time dissatisfaction shows up in renewal rates or in a negative review, the window for intervention has usually passed.

The American Customer Satisfaction Index (ACSI), which has tracked customer satisfaction across the US economy since 1994, consistently places economy-wide averages in the mid-70s out of 100. That figure varies by sector: software and technology companies tend to score higher, while utilities and airlines typically score lower. For a B2B professional services business, a score consistently above 80% generally reflects a well-functioning delivery experience. Sustained scores below 60% almost always point to a specific, addressable operational failure rather than an unmeasurable "customer attitude" problem.

In the context of a customer journey, CSAT data fills in the emotional picture at specific moments that broader analytics cannot capture. Your CRM might show that a client renewed for a second year, but it will not show whether that renewal came with enthusiasm or with mild dissatisfaction and no obvious alternative. CSAT, collected at the right moments, adds that layer. It is why customer journey mapping exercises almost always recommend instrumenting key touchpoints with brief satisfaction surveys rather than relying on renewal or retention data alone. On the B2B buying side specifically, the same instinct applies before a client even signs; see mapping the buyer journey for how to track satisfaction and friction through the stages that precede the sale, not just after it.

How it works.

The mechanics are straightforward. Send a one- or two-question survey as close to the interaction as possible, ideally within 24 hours. Response rates fall significantly after 48 hours, and feedback becomes less reliable as memory fades. For support interactions, the survey is usually triggered automatically when a ticket closes. For project milestones or delivery moments, it is sent manually or via your CRM within a day of sign-off.

Keep the survey short. One question asking for a rating, plus one optional open-text field asking "what could we have done better?" is the standard format. Adding more questions reduces response rates without adding proportionate insight. The open text field is where the useful detail lives: the scores tell you whether there is a problem; the comments tell you what the problem actually is.

Score calculation is consistent across implementations: count positive responses (top two on a five-point scale), divide by total responses received, multiply by 100. A survey with 45 positive responses from 60 total gives you a CSAT score of 75%. Most teams track this over time, by touchpoint, rather than looking at a single number in isolation. A score of 75% across all interactions is less useful than knowing that your onboarding scores at 88% while your post-implementation handoff scores at 58%.

Most CRM and helpdesk platforms can capture and report CSAT without additional tooling. If your CSAT data is currently sitting disconnected from your wider client reporting, that is usually a CRM setup issue worth addressing: the score becomes far more actionable when it sits alongside renewal dates, deal value and interaction history in a single view.

It is also worth deciding upfront which interactions you are going to survey and which you are not. Sending a CSAT survey after every email exchange will fatigue customers and devalue the data. The most useful approach is to survey at moments that genuinely matter to the relationship: the end of onboarding, after a support issue is resolved, at a quarterly review, or after a project milestone is delivered. Four or five meaningful data points per year per client is usually more valuable than weekly noise.

A practical example.

A B2B technology services business finds its overall CSAT is 79%. That looks reasonable. But when they break it down by interaction type, onboarding scores at 85% and technical support scores at 61%. The 79% average was hiding a meaningful problem in support.

Digging into the open-text comments on the support survey, the same themes appear repeatedly: long waits before a first response, and support agents answering a different question than the one asked. Two operational changes follow: a first-response SLA of four hours, with automatic escalation, and a revised intake form that forces clearer problem description before a ticket is opened.

Twelve weeks later the support CSAT is at 77%, and the overall score has moved from 79% to 83%. The improvement came not from changing the product or increasing headcount, but from diagnosing a specific breakdown and fixing it in a specific place. That is what CSAT is for.