Customer Effort Score (CES) is a single-question post-interaction survey. After a resolved interaction — most often a return, a support contact, or a checkout — the customer is shown a statement like “[Brand] made it easy for me to handle my issue” and rates agreement on a 5- or 7-point scale anchored at “strongly disagree” / “very difficult” to “strongly agree” / “very easy.” It’s a friction reading on one touchpoint, not a brand-level number.
How the math works
Three reporting forms are common.
Three reporting forms
-
Top-2-box share
is the percentage of respondents who picked the top two “easy” boxes — the cleanest DTC reporting form, and the one that survives ordinal-scale concerns. -
Average score
on the scale is widely used in tooling but treats a Likert response as if the gaps between boxes were equal, which they aren’t (the same caveat the CSAT entry makes). -
Net easy
% easy minus % difficult — mirrors the promoters-minus-detractors arithmetic of NPS and produces a single signed number that’s easy to track over time.
Where it came from
CES was introduced by Matthew Dixon, Karen Freeman, and Nicholas Toman in the 2010 Harvard Business Review article “Stop Trying to Delight Your Customers,” which argued that reducing customer effort predicts loyalty better than investments in delight. The Corporate Executive Board (CEB, now part of Gartner) developed the metric; the longer treatment is in The Effortless Experience (Dixon, Toman, DeLisi, 2013). The original CES asked customers to rate effort on a 1–5 difficulty scale; the revised form — the one most surveys ship today — uses the “made it easy” agreement statement on a 7-point scale.
CES vs. CSAT vs. NPS
Three metrics, three different questions. CES asks “how easy was this” on a single resolved interaction — it’s operationally actionable on one touchpoint. CSAT asks “how satisfied were you” on the same touchpoint — a broader transactional read. NPS asks “would you recommend us” about the brand overall — a relational, brand-level signal. The framing isn’t a ranking; it’s fit-for-purpose. Use CES to diagnose a friction-prone interaction, CSAT for general per-touchpoint satisfaction, NPS as the relationship-level tracking signal. Brands that ship all three usually field CES on the touchpoints where friction is suspected, CSAT broadly, and NPS quarterly.
What a falling CES tells the operator
CES earns its keep as a per-touchpoint diagnostic, and the touchpoint where it drops tends to name the team that should investigate. A falling CES on the returns flow points at returns-portal UX, restocking-fee disputes, or refund timing. A falling CES on checkout points at form complexity, payment failures, or error states. A falling CES on support points at agent process, hold times, or a self-serve gap upstream. Aggregated to a single brand-level CES, the signal loses most of its diagnostic value — the question CES is built to answer is which interaction got harder, not how the brand is doing overall.