What is customer satisfaction score (CSAT)?

Customer satisfaction score (CSAT) is a metric that captures how a customer felt about a specific interaction, product, or experience, expressed as a percentage from 0 to 100. It is collected by asking customers a single question — usually some version of "how satisfied were you with [the thing]?" — on a short rating scale, then calculating the share of responses that landed in the "satisfied" zone.

CSAT is the oldest and most widely deployed customer experience metric. It is built to be quick to answer, easy to compute, and granular enough to attach to a specific moment in the customer journey — the call that just ended, the order that just arrived, the help article that just resolved (or didn't resolve) a question.

This page covers what CSAT measures, where the methodology came from, how to calculate it, what scale to use, what a "good" score looks like, how CSAT compares to NPS and CES, where the metric is strong, where it falls short, and a short FAQ.

CSAT in one sentence

CSAT is the percentage of customers who, when asked, said they were satisfied.

What CSAT actually measures

CSAT is a transactional metric. It is meant to capture a customer's reaction to a specific, recent experience — not their overall opinion of the brand. The question is always pinned to something concrete: a support conversation, an order, an onboarding session, a feature release, a delivery.

That scoping is the whole point. CSAT is designed to give an operator a fast read on whether the thing that just happened went well. If support tickets that close on Tuesday score lower than tickets that close on Wednesday, that is a signal someone can investigate this week. If the post-delivery CSAT drops in a particular region, that is a signal a fulfillment team can act on.

Customer satisfaction as a concept is broad and fuzzy. CSAT as a metric is narrow and operational. The narrowness is a feature, not a bug.

Where CSAT came from

The modern CSAT methodology traces to the American Customer Satisfaction Index (ACSI), founded in 1994 at the University of Michigan by economist Claes Fornell. ACSI was modeled on the Swedish Customer Satisfaction Barometer, which Fornell had built in 1989, and was designed to produce a cross-industry, cross-economy measure of how American consumers felt about the goods and services they bought.

Fornell and his co-authors formalized the methodology in a 1996 paper that is still the academic reference for the index. The same paper established the practice of expressing satisfaction on a 0–100 scale, which is why CSAT scores today are almost always reported as a percentage even when the underlying survey uses a 1-to-5 or 1-to-10 rating.

The ACSI publishes a national satisfaction score every quarter. As of the Q4 2025 release, the U.S. national ACSI sat at 77 out of 100, where it has hovered for most of the last decade. That number is the benchmark most industry-level CSAT comparisons are built against.

How CSAT is measured

A CSAT survey has three moving parts: the question, the scale, and the formula.

The question

CSAT is a single-question survey. The standard wording is some variation of:

"How satisfied were you with [the experience/product/interaction]?"

The blank is where CSAT earns its operational value. The question can be scoped to a support ticket, a checkout, a return, a feature, an onboarding step, or anything else a team needs feedback on.

The scale

Three rating scales are common.

• 1-to-5 scale. The dominant format. Responses run from "very unsatisfied" to "very satisfied," and the top two boxes (4 and 5) count as "satisfied" responses.

• 1-to-10 scale. Often used by teams that already collect NPS on the same scale and want consistency. Responses of 9 and 10 typically count as "satisfied."

• Thumbs up / thumbs down or happy/neutral/unhappy. Binary or three-state. Used when a longer scale would be friction — embedded in an email signature, a chat widget, or a post-resolution prompt.

The choice of scale changes the texture of the data, but it does not change how CSAT is calculated.

The formula

The accepted CSAT formula is:

A "satisfied" response is one in the top of the scale — the 4s and 5s on a 1-to-5 survey, the 9s and 10s on a 1-to-10 survey, or the thumbs-up responses on a binary survey. Multiply by 100 and the output is a percentage.

Worked example: 200 customers respond. 150 of them give a 4 or a 5. CSAT = (150 ÷ 200) × 100 = 75%.

Qualtrics has noted that the top-two-box approach is the most reliable predictor of customer retention, which is why it has become the default even though "average all the responses" is mathematically simpler.

When to ask

CSAT works best when the experience is fresh. The most common trigger points:

• Immediately after a support interaction closes

• After an order is delivered

• After an onboarding step is completed

• After a product feature is released or first used

• A few weeks before a renewal decision

For depth on how to design the survey itself, see the Gladly guide on customer satisfaction surveys.

CSAT vs. NPS vs. CES

CSAT is one of three single-question metrics that show up in almost every customer experience program. They measure different things, ask different questions, and are useful in different moments.

The shortest way to keep them straight: CSAT is about the moment, NPS is about the relationship, CES is about the friction. Most mature CX programs run all three because each one answers a question the others cannot.

A note on overlap: CSAT and CES are sometimes confused because both fire post-interaction. The difference is that CSAT asks how the customer felt about the experience, while CES asks how hard the experience was. A customer can find something easy and still be unsatisfied, or find something hard and still be satisfied — which is why running both produces a more complete picture than running either one alone.

What is a good CSAT score?

There is no universal "good" CSAT score, because expectations vary by industry, channel, and what you are measuring. The closest thing to a global anchor is the U.S. national ACSI of around 77.

Industry benchmarks from Retently's 2026 CSAT study give a more granular read:

• Consulting: 83

• Digital marketing agency: 83

• Financial services: 81

• Property management: 78

• Ecommerce and retail: 77

• B2B software and SaaS: high 70s

• Education: 68

• Internet and software services: 64

• Construction: 30

• Communication and media: 26

Three rough bands hold up across most data sets:

• Below 50% — actively dissatisfied. Almost half the customers responding are unhappy. Needs immediate diagnosis.

• 50–70% — neutral zone. Most customers are neither delighted nor angry. Stable but not durable, and not a moat.

• 70–90% — healthy range. This is where most well-run CX programs operate.

• Above 90% — exceptional. Often achievable only on a narrow interaction (a specific support team, a specific channel) and very rarely sustained across the entire customer base.

The more useful question than "is my CSAT good?" is "is my CSAT moving in the right direction?" A score trending up by two points a quarter usually means more than a one-time score that is high.

What are the strengths of CSAT?

CSAT has stuck around for thirty years because the strengths are practical.

It is cheap to deploy. One question, any channel. CSAT can be sent over email, SMS, chat, IVR, in-app, on a receipt, or after a return. The implementation cost is low and the data starts arriving the day the survey goes live.

It is easy on the customer. A single question with a five-point scale takes seconds. Response rates are usually higher than for longer surveys.

It is operationally useful. Because the score is tied to a specific interaction, an operator can act on it. A CSAT drop in one queue or one fulfillment center is investigable in a way that a brand-level NPS dip is not.

It is widely benchmarked. The 0–100 percentage format is the same across vendors, so internal teams, industry peers, and external researchers can compare scores. ACSI's quarterly release makes the cross-industry comparison possible at no cost.

It pairs well. CSAT, NPS, and CES are complementary, not competitive. A program that runs all three sees both the moment-level signal (CSAT, CES) and the relationship-level signal (NPS) without either one fighting the other.

What are the limitations of CSAT?

CSAT is not a complete picture of the customer relationship and it should not be treated as one.

It only captures the moment. CSAT measures how the customer felt about one interaction. A customer who scores a support call a 5 can still cancel their subscription next week, and a customer who scores a return a 2 can still be a lifelong fan. Satisfaction is not loyalty.

It is biased toward extremes. People who had a great experience or a terrible one are more likely to respond. The merely-fine customers — the middle of the satisfaction distribution — often skip the survey entirely. That makes raw CSAT data noisier than it looks.

It is culturally inflected. Research has shown that respondents from collectivist cultures tend to avoid extreme ratings, while respondents from individualist cultures lean into them. A global CSAT score with a uniform scale can flatten real differences between regions.

"Satisfied" is a soft word. A 4 out of 5 means "satisfied" — not "delighted," not "loyal," not "will tell a friend." A program optimized purely against CSAT can end up rewarding "fine" and missing the gap between "fine" and "great." That gap is where churn and loyalty actually live.

It is a lagging indicator on its own. CSAT tells you how the last interaction went. It does not tell you why. To get to the why, the score has to be combined with conversation data, behavioral signals, and operational metrics — which is where modern AI-driven CSAT analysis has the most leverage. Our take on that is in how to use AI to improve CSAT scores and a CSAT improvement plan that actually works.

The practical takeaway: CSAT is a strong default signal, but it is a starting point, not a finish line. It tells you that something is up. It does not tell you what to do about it.