CES vs CSAT: what each metric measures and when to use which

Customer Effort Score measures how hard something was. Customer Satisfaction Score measures how good it felt.

Both ask about a single interaction rather than the whole relationship, which is why they get confused, and both can move in opposite directions on the same touchpoint without either being wrong.

The short version:

Those two sentences resolve most of the question. The rest of this page covers the parts that decide whether the metric you pick actually tells you anything: how the calculations differ, what each one predicts, and what to do when they conflict.

Picking the wrong metric doesn't produce a wrong number. It produces a correct number about something you weren't asking, which is harder to notice.

A support team measuring CSAT after a difficult refund may see a strong score because the agent was gracious, while the refund process itself remains a three-step ordeal.

The satisfaction reading is accurate. It simply isn't measuring the process, so the process never gets fixed. Run CES on the same interaction and the friction surfaces immediately.

The reverse happens too. A frictionless self-service flow can score well on effort while customers remain unhappy with the outcome, because ease of use and getting what you wanted are different things. Effort is a property of the process. Satisfaction is a judgment about the result.

CES is usually phrased as an agreement statement rather than a direct question, which is what distinguishes it from a satisfaction item. Common formulations ask customers to agree or disagree that the company made it easy to handle their request, or ask directly how easy it was to resolve their issue.

CSAT asks about satisfaction with a defined thing: an interaction, a purchase, a product, a support experience. The wording stays scoped to one item, because whole-company satisfaction is a different measurement.

This is the difference teams most often miss, and it makes the two scores non-comparable even when both are expressed as numbers on similar scales.

Because one is an average and the other is a proportion, a CES of 5.8 and a CSAT of 82% carry no arithmetic relationship. Neither can be converted into the other, and putting them side by side in a dashboard without labeling the method invites exactly that mistake. If you need the CSAT side in detail, how to calculate a CSAT score covers the denominator and scale decisions that determine the result.

Some teams report CES as a top-two-box percentage instead, to make it visually consistent with CSAT. That is defensible, but it has to be documented, because it changes the number substantially and breaks comparability with any external CES reference.

CES scales vary more than CSAT scales, which is a practical annoyance when comparing across sources.

Always confirm scale direction before interpreting a CES figure from an outside source. An unlabeled CES of 3 could be excellent or dire.

Both are transactional and both should fire close to the interaction, but the useful windows differ slightly. CES is best sent immediately, while the effort of the task is still fresh, because effort is remembered less accurately than feeling. CSAT tolerates a slightly longer delay when the thing being judged is a product rather than an interaction, since satisfaction with a purchase may need a few days of use to form.

CES is often positioned as the stronger churn predictor in service-heavy contexts, on the reasoning that effort is a more reliable driver of defection than momentary satisfaction. Gartner research cited on SurveyMonkey's customer effort score guide reports that customers with low-effort interactions are substantially more likely to repurchase and to increase spending.

CSAT is the better instrument for tracking the quality of specific interactions over time and for comparing performance across teams, channels, or products. It is more sensitive to changes in how something was delivered, which makes it useful operationally even where it predicts loyalty less well.

Neither is a loyalty metric in the way Net Promoter Score is. NPS is relational and asks about the overall relationship, while CES and CSAT are both transactional. How CSAT and NPS differ covers that boundary, and what CES is and how to measure it covers the effort side in depth.

Conflicting readings on the same touchpoint are informative rather than a data problem. Each pattern points somewhere specific.

Usually not simultaneously. Adding a second rating question to the same survey lengthens it and tends to reduce completion, and the two items are correlated enough that the second adds less than its cost.

More workable approaches:

CES and CSAT answer different questions about the same interaction: one about difficulty, one about satisfaction. Choose based on whether the outcome you're trying to change is a process or a result, keep the calculation methods clearly separate in reporting, and read disagreement between them as a signal about where the problem lives.

To go deeper on either metric or on building a program around them:

Net Promoter, Net Promoter Score, and NPS are trademarks of Satmetrix Systems, Inc., Bain & Company, Inc., and Fred Reichheld.