Customer effort score (CES): ask if it was easy, then ask what was hard

Customer effort score (CES) is a one-question survey that asks customers how easy it was to get something done with you, usually right after a support contact. The customer effort score in common use today asks people to rate their agreement with "The company made it easy for me to handle my issue" on a 1 to 7 scale, and you report it as an average or as the share of people who agreed. Use it after specific interactions where friction is the problem you want to find, and pair it with a short follow-up question, because the number tells you that something was hard and almost never what.

1. CES came from a 2010 argument that delight is overrated

The metric was introduced in "Stop Trying to Delight Your Customers," by Matthew Dixon, Karen Freeman and Nick Toman, in the July-August 2010 issue of Harvard Business Review. According to the HBR summary, their research covered more than 75,000 people who had dealt with contact center representatives or used self-service channels, and found that over-the-top service efforts made little difference. What customers wanted was a simple, quick solution to their problem.

The same summary says the authors introduced the Customer Effort Score and reported that it predicted loyalty better than customer satisfaction measures and the Net Promoter Score.

The article also listed five tactics, per the HBR summary: head off repeat calls by solving the next related issue, help reps handle the emotional side of a contact, cut down on channel switching, use feedback from struggling customers, and put problem solving ahead of speed. Those tactics are the real point. The score exists to tell you where to apply them.

2. Use the CES 2.0 wording, because the original question was confusing

The first version asked, "How much effort did you personally have to put forth to handle your request?" Jim Tincher, writing on CustomerThink in 2014, describes that phrasing as confusing and notes that CEB replaced it with an agreement statement: "The company made it easy for me to handle my issue."

The rewrite became known as CES 2.0. It got its own chapter, "The disloyalty detector: customer effort score v2.0," in the 2013 book The Effortless Experience by Dixon, Nick Toman and Rick DeLisi, according to its library catalog record.

The old question had two problems you can feel by reading it. "Put forth" is not how anyone talks, and a high score meant a bad experience (on effort-worded scales, higher numbers mean more effort, as Dovetail notes), which trips up both customers and the people reading the dashboard. The 2.0 version is plain English, and higher means easier.

Write it like this, with your company name filled in:

[Company] made it easy for me to handle my issue. 1 Strongly disagree, 2 Disagree, 3 Somewhat disagree, 4 Neither agree nor disagree, 5 Somewhat agree, 6 Agree, 7 Strongly agree

Change "handle my issue" to match the moment when the survey follows something other than support, for example "set up my account." Keep the agreement format and the direction.

3. Pick one scale and never compare scores across scales

Vendors disagree about the scale. Nicereply says CES 2.0 uses 7 points where the original used 5. Salesforce lists 1 to 5, 1 to 7 and 1 to 9 numeric scales, plus agreement phrases and emoji faces. Qualtrics shows an example running from "very easy" to "very difficult," and Dovetail shows both 1 to 7 and 1 to 10 examples.

Use the 7-point agreement scale. It matches the CES 2.0 wording, and seven points leave room to see movement without asking people to split hairs. Salesforce makes the important caveat plainly: scores are only comparable within the same scale type. A 4.2 on a five-point scale and a 4.2 on a seven-point scale are different stories.

Also write down which end is good. If someone on your team learned CES from an older source, they may still think a low number is the goal. (It is a short meeting, but an awkward one.)

4. Calculate it as an average, and also report the share who agreed

The most common formula is the plain average. Salesforce defines CES as the sum of all scores divided by the number of responses, and Nicereply calculates it the same way. Also track the percentage of people who answered 5, 6 or 7, meaning they at least somewhat agreed that you made it easy. Salesforce gives benchmarks for an agreement percentage as well as for averages.

Here is a made-up example. A payroll software company sends the CES 2.0 question after every closed support ticket for a week and gets 20 answers:

The average is 104 divided by 20, which is 5.2. The share who agreed is 14 of 20 (the 6, 5 and 3 people who answered 7, 6 and 5), which is 70%.

Report both. The average moves when a few people go from 6 to 7. The percentage is easier to explain to someone outside the team. Then look hardest at the four people who answered 1 to 3, because they are the reason you run the survey.

Short version: average for trends, percent agree for the slide, and the bottom of the scale for the work.

5. Ask right after one specific interaction, and only then

CES measures a moment, so send it while the moment is fresh. Salesforce recommends sending it right after a specific interaction ends, such as a closed ticket or a resolved chat. Qualtrics also recommends sending immediately after the touchpoint, using automated triggers, and keeping the survey to one or two questions.

Good moments to ask:

1. A support ticket or chat is marked resolved.
2. A customer finishes onboarding.
3. Someone completes a self-serve task, such as changing a plan.
4. Someone uses a new feature for the first time.

Dovetail adds purchases, how-to article reads and new feature rollouts as triggers. Do not send CES on a quarterly schedule to your whole customer list. Tincher calls CES a transactional measure that does not work as an annual relationship score. "Was it easy?" needs an "it," and a quarter is not one.

Add one optional open question after the scale, the two-question format Salesforce recommends: "What made it harder than it should have been?" A score of 2 with no explanation leaves you guessing. The same score with "I had to explain the problem three times" hands you the repeat contact problem from the 2010 article.

6. Use CES next to NPS and CSAT, since each one answers a different question

The three metrics get confused because each is one question with a number. They measure different things:

That table follows the comparison on the Salesforce CES page, which treats the three as complementary. Qualtrics makes the same split: NPS covers long-term loyalty across the whole relationship, while CES and CSAT cover a single touchpoint.

Choose CES over CSAT for support and self-serve flows. Effort points at a fixable step, such as a transfer, a form or a missing answer, while satisfaction also mixes in how the customer feels about the answer itself. Keep NPS as your relationship measure. The posts on Net Promoter Score and on what counts as a good NPS score cover how to read that one.

On the loyalty claim, Qualtrics quotes The Effortless Experience: 96% of customers with a high-effort service interaction became more disloyal, against 9% with a low-effort one. Dovetail attributes the same figures to Gartner. Both reach you secondhand and trace back to the people who created the metric, so treat them as a strong hypothesis to test in your own business.

7. Do not use CES to judge your product or your whole relationship

CES has limits that matter for the decision. Dovetail notes that it covers service rather than the overall business and does not account for cost, competitors or product quality. Qualtrics points out that a loyal customer can score low after one bad interaction, so a single score is not a verdict on the account.

It also fits poorly when you have few interactions to measure. A B2B company with 40 customers and a handful of support tickets a month will get scores too sparse to trend. In that case, skip the dashboard and call the people who had a hard time.

The question wording matters as much here as anywhere. Avoid adding "How easy was our amazing new portal?" style framing, which biased survey questions covers in detail, and keep the rest of the survey as short as the advice in customer survey questions.

8. Follow every low score with a conversation

A CES survey finds the hard moments, and the fix comes from asking the person what happened. Pick five customers who answered 1 to 3 this month and ask each to walk you through the contact from the first click, step by step. You will usually hear about a handoff, a repeated question or a dead end in self-service, the same failures the 2010 article told companies to remove.

If the people you need are outside your customer list, for example support leaders who have already cut channel switching at their own companies, Instant Expert can find them and send a paid invitation for a short call or a written answer, and you pay only if someone books or answers. The directory pages for customer experience consultants and customer support quality assurance experts are a place to start.

Your first step today: add the CES 2.0 question to the message that goes out when a ticket closes, with one open question under it, and read every answer below 4 by hand for a month. For customers who leave anyway, how to interview churned customers covers what to ask.