Contact centers report two numbers that sound like they measure the same thing and do not. Customer Satisfaction Score, or CSAT, asks a customer to rate how the interaction felt. Customer Effort Score, or CES, asks how hard the interaction was. A customer can rate an experience highly on one and poorly on the other, and when that happens, CES is the better predictor of what the customer does next.
Where CSAT comes from and what it captures
CSAT is a single post-interaction question, typically "How satisfied were you with this experience," scored on a 1-to-5 or 1-to-10 scale. It has been the default contact center metric for decades because it is cheap to collect and easy to explain to a leadership team.
It also measures a narrow thing: how the customer felt in the moment, right after the call or chat ended. A rep who apologizes well, sounds friendly and moves quickly can produce a high CSAT score even when the underlying problem took three contacts to resolve. SQM Group, which benchmarks CSAT performance across more than 500 North American contact centers each year, puts world class CSAT at 85% or higher, a bar only a small share of centers clear.
CSAT also has a structural weakness: most customers never answer the survey. Retently's analysis of more than 25 million survey invitations sent across 600 e-commerce brands in 2025 found CSAT response rates in the single digits for email surveys, with in-app prompts performing meaningfully better.
Why Customer Effort Score exists
CES was introduced in a 2010 Harvard Business Review article by Matthew Dixon, Karen Freeman and Nicholas Toman, based on a study of more than 75,000 customer interactions across phone and self-service channels. Their finding: reducing the effort a customer has to expend to get an issue resolved predicted loyalty better than satisfaction ratings or Net Promoter Score. A customer who solved their problem in one contact with no channel-switching came back, even if they were not especially delighted. A customer who was satisfied with how a rep handled a problem that took four contacts to fix often left anyway.
CES asks a version of "how easy was it for you to handle your issue," typically scored on a 1-to-7 scale from very difficult to very easy. It measures friction, not mood.
How the two disagree in practice
The gap between the two shows up most clearly around automation. A chatbot or IVR flow can resolve a simple request instantly and score well on both metrics. It can also resolve a request badly, hand off to a human after the customer has already explained the problem twice, and still get a decent CSAT score if the human agent who finally solves it is warm and apologetic. CSAT rewards the recovery. CES penalizes the repeat contact and the channel switch that made the recovery necessary in the first place.
This is the specific reason CES matters for evaluating support automation rather than sitting as a second dashboard tile next to CSAT. If a company is measuring whether a new AI deflection flow, agent-assist tool or self-service option is actually working, satisfaction after the fact tells you less than effort during the process.
What to check before trusting either score
Three questions apply to any CSAT or CES program before treating the number as decision-grade.
What is the response rate, and who is missing. A CSAT score built on a single-digit response rate is a report on the customers who felt strongly enough to answer, not the customer base.
What scale was used, and has it changed. Moving from a 5-point to a 10-point scale, or changing what counts as "satisfied," shifts the reported number without changing anything a customer experienced. Compare scores only within the same methodology.
Where in the journey was the question asked. A CES question asked immediately after a resolved contact measures that contact. It does not capture effort spent earlier in the same journey, such as a failed self-service attempt before the customer called in.
What commonly goes wrong
The most common mistake is treating CSAT as proof that an automation deployment succeeded. A high CSAT score after a bot-to-human handoff can mask the fact that the bot failed to resolve anything and the customer had to repeat themselves. CES, tracked alongside first contact resolution, is more likely to surface that failure because it asks about the process rather than the mood at the end of it.
Neither metric replaces the other. CSAT is still useful for reading whether a specific interaction felt good. CES is the more honest signal for whether a support process, automated or not, is actually reducing what the customer has to do.
Sources: Harvard Business Review: Stop Trying to Delight Your Customers · SQM Group: Call Center Metrics and World-Class KPI Targets · Retently: State of Survey Response Rates