What is customer effort score: CES explained for service and support teams
Customer effort score, usually shortened to CES, is a post-interaction survey metric that measures how much work a customer had to do to get their issue resolved, scored on a single question and used to predict whether that customer comes back. It asks about one interaction, not about the relationship.
The premise behind it is narrow and worth saying plainly: in service, loyalty is lost far more often than it is won, and the thing that loses it is friction.
What is the CES question, exactly?
Two versions are in circulation and teams regularly mix them up.
The original (2010) asks: "How much effort did you personally have to put forth to handle your request?" It is scored from 1 (very low effort) to 5 (very high effort). Watch the direction. On this scale a low number is the good result.
The revised version, generally called CES 2.0 and dated to 2013, drops the word effort from the question entirely. It presents a statement, "[Company] made it easy for me to handle my issue," and asks for agreement on a 1 to 7 scale where 1 is strongly disagree and 7 is strongly agree. A high number is now the good result, and the sentence puts the responsibility on the company rather than on the customer.
Most survey tooling ships CES 2.0 by default. If your dashboard reads 5.6 and someone in the room remembers CES as "lower is better", you have a reporting problem before you have a service problem.
How is customer effort score calculated?
Three methods are in common use. Pick one and never quietly change it.
- Mean score: add every response, divide by the number of responses. The most common approach.
- Percentage agreement: the share of responses scoring 5, 6 or 7, reported the way a top-box satisfaction score is reported.
- Net score: percentage agreement minus percentage disagreement (1 to 3). Some teams prefer it because it moves visibly.
Timing matters as much as arithmetic. Send it within minutes to hours of the contact closing, attached to a specific ticket. A CES asked quarterly, detached from an event, produces a number nobody can act on.
Where the research came from
CES was introduced by Matthew Dixon, Karen Freeman and Nicholas Toman in Stop Trying to Delight Your Customers, Harvard Business Review, July/August 2010. The study behind it was run by the Customer Contact Council, a division of the Corporate Executive Board, and surveyed more than 75,000 B2C and B2B customers over three years about service interactions in non face-to-face channels.
The argument, in their own numbers:
- Of customers reporting low effort, 94 percent expressed an intention to repurchase and 88 percent said they would increase their spending. Only 1 percent said they would speak negatively about the company.
- 81 percent of customers who had a hard time solving their problem reported an intention to spread negative word of mouth.
- CES outperformed both customer satisfaction and net promoter score at predicting behaviour in customer service interactions.
- 84 percent of customers said their expectations had not been exceeded in their most recent interaction, while 89 of the 100 customer service heads surveyed said exceeding expectations was their main strategy.
That last pair is the whole case in miniature. The industry was investing in delight and delivering effort.
CES explained: a practical example
A support team runs CES 2.0 on every closed ticket and averages 5.8, which looks healthy. Split by resolution path, tickets closed at first contact score 6.4 and tickets transferred once score 4.1. Nothing was wrong with the agents on those transferred tickets. The routing was wrong, and the score found it.
When customer effort score is the wrong metric
CES is a good metric with a narrow remit, and most of the damage done with it comes from stretching it past that remit.
- It measures the journey, not the answer. A customer told "no" quickly, clearly and politely can rate the ease highly and still leave. CES will not see that departure coming.
- It only hears from people who had a problem. It says nothing about the customers who never contacted you, which is most of them.
- It cannot replace a relationship measure. It was built to evaluate service interactions and it was validated on service interactions.
- Low effort is not always the goal. Where the experience is the product (a test drive, a fitting, a tasting menu), time and involvement are the value, and optimising them away damages the thing you sell.
- It is easy to game. Agents who can see their own CES will close tickets fast to protect it. Always read it next to a reopen or repeat-contact rate.
- Small samples move it a lot. A team handling 200 tickets a month at a 12 percent response rate is reading 24 answers. Treat that as a direction, not a number.
One more piece of honesty about the source. The claim that delight does not pay has been argued over ever since, and the defensible reading is not that delight never works. It is that removing effort is the cheaper and more reliable lever, and that most organisations have far more of it to remove than they think.
Where CES sits next to your other numbers
CES is an outcome measure. It tells you friction exists and never where it is. The operational metrics around it do the locating: first contact resolution and transfer rate at the service desk, routing accuracy in support triage, and first-time fix rate once a job goes to the field. A CES that drops for a month is a prompt to go and read those.
It also has a specific job after a failure. Effort during the fix, not the size of the goodwill gesture, is what decides how a service recovery lands. Measure the recovery interaction the same way you measure any other.