CES (Customer Effort Score) has emerged alongside NPS and CSAT as the third pillar of customer measurement. Since Harvard Business Review published its research showing "loyalty comes from reducing effort, not from delighting," CES has become especially central in customer support and SaaS.
But CES has one property that sets it decisively apart from the other metrics. CES measures "process," not "emotion." Use it with the same instincts as a satisfaction score, without grasping that property, and CES never shows what it's good for. This article covers the basics — then gets into the field method of reading CES against behavioral logs.
What CES is
CES measures how much effort the customer had to exert to accomplish their goal. Two common question formats:
Original (CES 1.0)
"How much effort did you have to put in to resolve your issue?" (1–5: Very high effort — Very low effort)
Updated (CES 2.0)
"The company made it easy for me to handle my issue." — To what extent do you agree? (1–7: Strongly disagree — Strongly agree)
The updated format is now the standard. Asking "agree with this positive statement" produces more stable responses with less cultural bias.
Why CES matters
Harvard Business Review's research found:
- 96% of customers who have a high-effort experience trend toward disloyalty (churn, switching to a competitor)
- Reducing effort drives loyalty more than "exceeding expectations" does
The implication:
Not making them think "wow, amazing!" — but not making them think "what a hassle" — is what actually prevents churn.
"Reduce effort over delight" is not universal — where CES fits and where it doesn't
Here is an important caveat most CES explainers skip.
That famous HBR conclusion was obtained in a customer-support context. When "a customer with a problem comes to get it solved," reducing effort is righteous. Nobody is looking for "delight" in a support interaction.
But not every customer touchpoint is a problem-solving touchpoint. At touchpoints where the experience itself is the value, CES becomes the wrong lens.
| Touchpoints where CES fits | Touchpoints where CES doesn't |
|---|---|
| Support inquiries | Hospitality at a premium service |
| Cancellation / returns | Entertainment, hobby-driven products |
| Onboarding setup | Brand experience, in-store experience |
| Sign-up / checkout flow | Exploratory experiences where "the joy of browsing" is the value |
Ask "was that easy?" at a touchpoint on the right-hand side and the customer thinks, "well... I came here to enjoy that part." Treat CES as a metric dedicated to "touchpoints where effort should be reduced." For touchpoints where the effort itself is enjoyable, use CSAT or NPS.
How to compute CES
For a 7-point scale, score is the mean of responses (no special calculation like NPS). Example:
- Response 1: 6 / Response 2: 5 / Response 3: 7 / Response 4: 4 / Response 5: 6
- Mean = 5.6 → CES = 5.6 / 7
Rough benchmarks (7-point):
- Below 5.0: Needs work
- 5.0–6.0: Standard
- Above 6.0: Strong
When to use CES
CES is best right after a touchpoint. That is non-negotiable.
| Touchpoint | When to send |
|---|---|
| Support case | Immediately after resolution |
| Onboarding | Right after setup complete |
| Cancellation | On the cancellation-complete screen |
| Return / refund | Right after the process completes |
The longer the gap from the experience, the more the "memory of effort" fades and the more answers regress to the middle.
CES vs. NPS vs. CSAT
| Dimension | CES | NPS | CSAT |
|---|---|---|---|
| Measures | Effort at a touchpoint | Overall recommendation | Satisfaction with experience |
| Nature | Process metric | Relationship metric | Emotion metric |
| Churn prediction | Strong | Medium | Weak |
| As an exec metric | △ (touchpoint-bound) | ◎ | ○ |
| Directness to action | High | Medium | High |
CES is the strongest for churn prediction and touchpoint improvement — it complements NPS and CSAT rather than replacing them.
A CES score alone lies — read it against behavioral logs
This is the core of the article. We said CES is a "process metric." Precisely because of that, staring at a CES score on its own is almost meaningless.
The reason is simple: CES is subjective. For the exact same "support case with 3 exchanges over 30 minutes," a low-expectation customer answers "that was easy" (high score) and a high-expectation customer answers "what a hassle" (low score). The score alone cannot tell you whether "the process was good" or "the customer's expectations were just low."
So the real way to use CES is to cross the CES score (subjective) against objective behavioral logs. Example logs to cross against:
- Time to resolution
- Number of exchanges
- Number of agent hand-offs
- Re-contact rate for the same issue
Build a matrix on these two axes and the CES score suddenly becomes readable.
| Good log (low effort) | Bad log (high effort) | |
|---|---|---|
| High CES score | ✅ Healthy. Keep this touchpoint | ⚠️ Warning sign. The customer is conditioned to low quality. Expectations are merely low |
| Low CES score | ⚠️ Expectations too high / under-communicated. Fix the communication | 🔴 Improvement target is clear. Fix the process itself |
Watch the top-right cell especially — "high score × bad log." If you only looked at the CES score, you'd call this touchpoint "fine." In reality the customer may have given up and come to accept low quality as normal. That is a future churn candidate. Chase the subjective score alone and you miss this warning sign entirely.
Pitfalls
Touchpoint must be specific
CES is about a specific experience. "How much effort for the service overall?" produces noisy data, because different respondents have different scenes in mind.
Always pair with open text
For low scores (1–3), follow up with:
"What should we change to make this easier?"
Open-text answers there are typically a goldmine of improvement hints.
High score, still churns
A customer with a high CES can still churn over pricing or fit. CES is not sufficient on its own — combine with other indicators.
Concrete actions when CES is low
When you identify a low-CES touchpoint (one whose log is also bad), the standard playbook:
- Expand pre-contact FAQ — raise self-service resolution rate
- Add a chatbot — automate triage
- Cut steps in the process — fewer form fields
- Show progress — "2 steps left"
- Prevent hand-off ping-pong — first agent owns through to resolution
All of these reduce customer effort.
Summary
How to use CES:
- CES is dedicated to "touchpoints where effort should be reduced." Don't use it where the experience itself is the value
- Ask right after the touchpoint, in the agree/disagree (2.0) format
- Always pair with open text
- Don't judge on the score alone — read it in a four-quadrant grid against behavioral logs
The single biggest payoff of using CES correctly is catching the "high score × bad log" warning sign.
Tell Repoan's AI chat "Build a post-support survey in CES format" and you get an agree-disagree scale survey immediately. Or start from the post-support inquiry template and modify.
When to use CES with CSAT and NPS is covered in NPS vs CSAT. The AI analysis feature (see AI response analysis) can extract "what is driving the effort" automatically from open text.
Related
- The "why" behind effort: How to discover customer insights
- The silent majority AI can't see: Why run surveys in the AI era
- Improvement cycle with CES: Running a survey PDCA cycle