Customer effort score (CES) and how to measure it

To work it out in Gartner’s 2019 version, ask customers how far they agree that your company made it easy to resolve their issue, from 1, the strongest disagreement, to 7. The score is the share who answer 5 or above, so if 28 of 40 customers choose a 5, 6 or 7, your customer effort score is 70%.

The measure comes from a study by the Corporate Executive Board, later known as CEB, published in Harvard Business Review in 2010. Gartner bought CEB in 2017 and still publishes research on the score. The 2010 version asked about effort on a 1 to 5 scale, where a low score was good.

For a small software company, effort usually means waiting overnight for a reply, explaining the problem twice or being sent from the chat to email. You can cut most of it without sending a survey at all.

Deacon is an AI customer support agent that answers your customers from your own docs in seconds, at any hour, and shows you where they get stuck. It has a free plan.

How to calculate customer effort score

Gartner’s 2019 article counts the share who at least somewhat agree, which is a 5 or above. Other tools average the ratings instead. Here are 40 responses on the seven-point scale, read both ways.

Gartner’s way, 28 of the 40 are a 5 or above, so the score is 70%. Averaged, the ratings add up to 209, or 5.2 out of 7. HubSpot reports the average, and its surveys call a 6 or 7 Easy, which here is 22 responses, or 55%. That’s three different numbers for the same 40 customers.

The 2010 scale and Gartner’s run in opposite directions

The 2010 question asked customers how much effort they had to put in, so a low score was good. Gartner’s 2019 version turned it into a statement to agree with, where a high score is good.

So an average of 2.2 is a good result on the 2010 scale and a poor one on a seven-point agree scale. Check which version a figure uses before you compare it with your own.

When it’s asked and who answers

Customer effort score asks about one support conversation, once it’s over. HubSpot’s help says the survey is often sent soon after a ticket is resolved, and you pick the ticket status that sends it.

It’s one question on purpose. HubSpot won’t let you add a second, and says each extra question cuts a survey’s response rate by 30 to 50% on average, without saying where that figure comes from.

Even so, many customers never answer. Gartner’s research on the score from June 2026 opens by saying that low response rates limit how reliable and useful survey-based effort scores are.

The customers who found it hardest are the easiest to miss. Someone who gave up on your docs and left never started a conversation, so they never got the survey.

What is a good customer effort score?

There’s no published benchmark you can rely on. Gartner’s own research on measuring the score is for its clients, and its 2019 article gives no average. The 2010 article gives one reference point. Osram Sylvania cut its score from 2.8 to 2.2 on the five-point scale by coaching its reps to rephrase words like “can’t” and “won’t”. That put it 18.5% below the average CEB saw for B2B companies. It’s one company in 2010, on the old scale, so compare your score with your own earlier months, measured the same way.

What the research does show is how much effort matters. These figures come from CEB’s study, read on 27 September 2026.

Gartner puts it more starkly. Its 2019 article says 96% of customers with a high-effort service interaction become more disloyal, against 9% with a low-effort one, though it doesn’t give the sample.

Three ways customer effort score misleads

The scales run in opposite directions. A falling score is good news on the 2010 scale and bad news on Gartner’s, and a share and an average of the same responses look nothing alike.

Small numbers swing it. With 15 responses, each one is worth almost 7 points of Gartner’s score.

It tells you effort was high, not why. The 2010 article and Gartner’s both pair the score with counts of repeat contacts, transfers and switches from one channel to another, because those show where the effort came from.

Customer effort score vs CSAT and NPS

CSAT asks how satisfied a customer was with a conversation, and NPS asks how likely they are to recommend you. The CSAT entry compares all three in a table. In CEB’s study, effort was a better guide than either to whether customers would stay and spend more after a service conversation.

Where effort comes from when you answer support alone

The 2010 article names the usual causes. Customers have to get in touch again, get transferred, repeat themselves or switch from one channel to another, and having to get in touch again is by far the biggest.

A founder rarely transfers anyone, so for you effort comes down to three things. How long customers wait for your first reply is your first response time. Whether they had to write back is your first contact resolution, and whether the problem ended solved is your resolution rate. You can count all three without a survey.

The questions show you where the effort is. When the same question keeps coming from one page, that page is making customers work. Customer insights goes through what each kind of question points to. Effort is one part of the wider customer experience, and customer service experience covers what makes support feel easy or hard.

How Deacon takes the effort out of asking

Deacon answers your customers from your own docs in seconds, at any hour and in the language they write in, in a chat on the page they’re already on. When your docs don’t cover a question, it says it doesn’t know instead of guessing and asks for the visitor’s email. You reply from the dashboard, and if they still have the chat open, your reply appears there under your name, so they don’t have to start again by email.

Deacon doesn’t run surveys. What it shows you is where customers had to work. Analytics › Widget gives you Answered, the share of messages Deacon attempted that it answered, and Waiting for a reply, the count of everyone who left their email and is still waiting to hear from you. Once one page’s questions go unanswered more often than anywhere else, it names that page.

Topics group the questions people keep asking, count who asked and label most groups as a how-to, a bug report, a feature ask or another type. A big how-to group is effort you can remove with one written answer. Write it from the Answer this button on a question Deacon couldn’t answer, and Save & check shows the reply Deacon now gives. From then on the next customer can get it in seconds, and Deacon needs you a little less.

Chatbot analytics covers everything the dashboard counts. Every feature is on the free plan, which needs no card.