How this instrument works
Customer Effort Score comes from a single survey question asked right after a support interaction — usually some version of 'how much effort did you personally have to put forth to handle your request?' — answered on a numbered scale where 1 is very low effort and 7 is very high effort. This instrument's formula is the plainest arithmetic possible: add up every respondent's rating and divide by how many people answered. Nothing is weighted, nothing is trimmed; the average is the metric.
The research behind it, published by the Corporate Executive Board and popularized in a 2010 Harvard Business Review article, argued that reducing customer effort predicted repeat purchase and referral more reliably than trying to delight customers did. That claim reframed how contact-center and product teams read their surveys — instead of chasing a warmer interaction, a team chases a shorter, plainer one, and this reading is the number they watch to see whether that is happening.
A support-operations lead or CX manager reaches for this figure after a specific interaction type — a billing dispute, a return, a password reset — rather than for the business as a whole, because effort is local to a process. The average also hides its own shape: it cannot distinguish a batch where everyone rated the process a middling 4 from a batch split evenly between people who rated it 1 and people who rated it 7, so a team that only watches the single number can miss a process that is smooth for most customers and badly broken for a minority.
- Add every respondent's rating from the batch of surveys you are scoring and enter that total as Sum of all effort scores (1-7 scale).
- Enter how many people actually answered that batch as Number of responses — count completed ratings only, not surveys sent or opened.
- Read Customer effort score — the instrument divides the total by the response count for you.
- Compare the reading against a previous period measured with the same question wording and the same 1-to-7 scale, since a five-point CES survey and a seven-point one are not interchangeable.
Worked example — 350 points across 100 responses
Take a support team closing out a week of surveys tied to one interaction type, each respondent rating the effort of resolving their issue from 1 (very easy) to 7 (very hard). The ratings summed to 350 points across 100 completed responses. Enter 350 as Sum of all effort scores (1-7 scale) and 100 as Number of responses, and the instrument divides one by the other to return a CES of 3.5 — the exact midpoint of the seven-point scale.
A reading of 3.5 sits squarely between the low-effort end and the high-effort end, so this team's process is neither obviously smooth nor obviously broken on its own. That midpoint result is what usually sends a CX manager back to the free-text comments attached to the highest-scoring responses, looking for the one step, a forced phone transfer, a repeated identity check, an unclear self-service article, that dragged the average up toward the middle instead of down toward 1.
Questions
Why does a lower CES mean a better result?
The scale measures effort, not satisfaction, so a 1 means a respondent found the interaction very easy and a 7 means they found it very hard. That inverted direction is the most common misreading of this metric — a team used to scores like Net Promoter Score or Customer Satisfaction, where climbing numbers are good news, can glance at a rising CES and read the trend backwards, mistaking a harder process for an improving one.
Where does the Customer Effort Score formula come from?
It comes from research published by the Corporate Executive Board, now part of Gartner, and popularized in a 2010 Harvard Business Review article that found effort reduction predicted repurchase and referral better than delight did. That original study used a five-point scale; many support platforms, including the one assumed here, now run a seven-point version, so check which scale a benchmark was built on before comparing it against your own reading.
Does averaging the ratings hide anything important?
Yes — a simple mean can produce the same 3.5 from a batch where almost everyone rated effort a 3 or 4, or from a batch split evenly between people who rated it 1 and people who rated it 7. Those are very different support experiences wearing an identical number, so a distribution of the raw ratings, not just the average this instrument returns, is what tells a consistently middling process apart from one that is broken for part of its customers.
How many responses does a CES reading need before it means much?
More than a handful — a dozen responses can swing several points on a single extreme rating, while a few hundred settle into a stable average. A CES built from under thirty responses is worth watching as an early signal, but the response count sitting next to the score matters just as much as the score itself, since a thin sample and a settled trend can produce the identical number.
Is Customer Effort Score the same thing as Customer Satisfaction Score?
No. CSAT asks how satisfied a respondent felt with an interaction, purchase, or product, and a higher rating is the good outcome on that scale. CES asks narrowly how much effort resolving one specific issue took, and a lower rating is the good outcome here. A customer can call an interaction highly satisfying while still logging real effort to reach that outcome, so the two figures answer different questions.
What counts as a strong Customer Effort Score?
It depends on scale and wording, since not every survey uses the same one. On the seven-point, lower-is-better scale this instrument assumes, contact-center benchmarking studies commonly describe averages near 2 as low-effort and averages above 4 as high-friction. Benchmarks quoted from the original five-point CEB research need converting before they say anything useful about a seven-point reading like this one.
References
- Harvard Business Review — original Customer Effort Score research (Dixon, Freeman, Toman, 2010)
- U.S. Small Business Administration — Marketing and sales guide
Read this first: This instrument shows arithmetic, not advice. Real offers add fees, taxes and terms that vary by lender and place — verify the figures against your actual paperwork before deciding anything.