“How much effort did you personally have to put forth to get this done?”
From Matthew Dixon, Karen Freeman & Nick Toman · CEB (later Gartner) researchers; originators of the Customer Effort Score
Why it works
Dixon, Freeman and Toman studied more than 75,000 customer interactions and found something counterintuitive: delighting customers barely moves loyalty, but making them work hard reliably destroys it. Effort turned out to be the strongest predictor of whether someone would stay, spend more, and speak well of you. The single-item framing works because effort is concrete and recent — a respondent can answer 'how hard was that?' honestly the instant a task ends, where 'are you satisfied?' invites a vague, polite shrug. The original CEB item was scored 1–5; the wording has since evolved (Gartner's later version is a 7-point agree/disagree statement that drops the word 'effort' entirely), so what makes CES comparable is asking the same effort question, the same way, after the same kind of moment for everyone.
When to ask
Immediately after a discrete task the customer just completed — resolving an issue, finishing onboarding, returning an item. CES is transactional by design: it asks about the effort of one specific interaction, not the relationship overall, so fire it while that interaction is still vivid.
Good follow-ups
- What was the hardest part of getting this done?
- Was there a moment where you nearly gave up?
- What would have made that effortless?
Watch out for
Asking about the relationship instead of a specific interaction. CES is not a how-are-we-doing question; aimed at the whole product it loses its predictive edge because there's no single effortful moment to anchor on. The fix is to tie it to a concrete just-completed task. The other trap is sampling only the wins — if agents fire it after smooth resolutions but skip the painful ones, the average flatters you and the metric stops warning you. Deploy it across all interactions of a type, and pair the score with an open 'what made it hard?' so you can actually remove the friction the number is flagging.
Where to ask
- In-product surveygreat
The native habitat — fire the single effort item the moment a task completes (a ticket resolved, a setup finished) so the experience is still fresh. CES was built to run exactly here, transactionally.
- Long-form surveygreat
Travels cleanly in an emailed post-interaction survey; the one-item scale stands on its own and the average only means something at sample size, which a survey delivers.
- User interviewworkable
Fine as a structured probe to hear why something felt hard, but one person's effort rating is anecdote — the CES average is a property of many responses, not a single conversation.
- Support chatworkable
Works as the post-resolution pulse support teams know best — but a single agent firing it ad hoc skews the sample; deploy it systematically across all resolved contacts, not just the ones that went well.
Reworded for this context: “How much effort did it take to get this sorted today?”
- Sales callpoor
CES measures the effort of getting something done with an existing product. A prospect who hasn't used it has no effort to rate, so the question lands as hypothetical and the answer is noise.
Pairs well with
Stage: Stick · A question popularized by Matthew Dixon, Karen Freeman & Nick Toman