Choosing between NPS vs CSAT vs CES can feel like picking one dial on a crowded dashboard. Use NPS for loyalty, CSAT for satisfaction with a single interaction, and CES for how hard something was.
This comparison sits within the wider topic of customer experience metrics.
| Item | Value |
|---|---|
| NPS score range | -100 to +100 |
| NPS promoters | Scores of 9 and 10 |
| NPS detractors | Scores of 6 or under |
| CSAT “satisfied” ratings | 4 (satisfied) or 5 (very satisfied) |
| CES agreement scale (one common format) | 1-7 |
| Global average CSAT | 86% |
The scoring bands come from IBM and the CES scale and CSAT average from Help Scout. Figures can change and vary by survey design, so check the source before you rely on them.
What NPS, CSAT and CES each measure
Net Promoter Score asks how likely a customer is to recommend a company. IBM says it was introduced in a well-known 2003 Harvard Business Review article and has since become an important tool for understanding customer relationships.
NPS is a relationship metric. QuestionPro describes it as measuring overall brand perception rather than a single interaction.
Customer Satisfaction Score (CSAT) works at a smaller scale. It is often called a transactional metric because it focuses on a single touchpoint, such as a support call or a purchase, according to QuestionPro.
Customer Effort Score (CES) looks at friction. IBM explains that it determines how easy or difficult customers found accomplishing their goals, for example making a purchase or having an issue resolved.
In short, the three answer different questions:
- NPS: would this customer put their name behind us?
- CSAT: did this specific experience meet expectations?
- CES: was the task easy or a struggle?
How each score is calculated
NPS is the most formula-driven of the three. The steps are short:
- Ask the standard question, as QuestionPro words it, on a scale from 0 to 10.
- Sort replies into groups. IBM treats 9 and 10 as promoters, 7 and 8 as passives, and 6 or under as detractors.
- Subtract the share of detractors from the share of promoters. The result runs from -100 to +100, per IBM.
CSAT is a simple percentage. IBM defines it as the share of respondents who say they are satisfied (4) or very satisfied (5), so you divide those answers by all answers.
CES has no single fixed format. IBM says most surveys ask customers to rate difficulty from 1 (easy) to 5 or 7 (difficult), while Help Scout describes a scale from 1-7 where 1 is strongly disagree and 7 is strongly agree.
Because the direction of those scales differs, QuestionPro lists a need for clear wording to avoid confusion about scale direction as a drawback of CES. Label both ends of the scale before you launch.

Benchmarks need context
A score on its own says little. IBM advises benchmarking against industry norms, because every industry has slightly different baseline NPS scores.
IBM points to a 2024 Retently study that found significant differences between NPS averages across industries. In that example the average NPS for an insurance company is 80, whereas it is 39 for cloud and hosting (IBM).
CSAT benchmarks vary too. Help Scout reports a global average CSAT of 86%, but notes that this can vary wildly by industry.
Treat any published average as a rough reference point. The more useful comparison is your own score over time, using the same question, scale and timing.
Choosing the right metric for the moment
The common failure is a timing one. QuestionPro puts it plainly: the mistake is using the wrong metric at the wrong moment.
A practical way to match metric to moment:
- Use NPS when you want a periodic read on overall loyalty, not a verdict on one event.
- Use CSAT straight after a defined interaction, such as a resolved support case or a checkout.
- Use CES after a task where friction would drive customers away, such as getting an issue fixed.
Each has a blind spot. IBM notes that CSAT helps you understand satisfaction but does not determine whether a customer is likely to recommend the company, and that CES can be less representative of overall satisfaction because it focuses on a specific action.
If friction is your main concern, the guide to customer effort score covers that metric on its own.
For business-to-business firms, CustomerGauge recommends that B2B brands start with NPS. That advice comes from a vendor focused on account-level feedback, so weigh it against your own customer base.
Whichever you pick, the wording of the question shapes the answer. The guide to customer satisfaction survey questions looks at phrasing in more detail.
Using all three together
You do not have to choose only one. CustomerGauge research shows 49% of NPS users also measure an additional metric, most commonly CSAT (33%) or CES (1 in 6).
One suggested cadence comes from Zonka Feedback, which says the best CX programs use all three at different journey stages: NPS quarterly, CSAT post-interaction, and CES post-task completion. Treat that as a starting pattern rather than a rule.
The benefit of combining them is clarity. QuestionPro says that, used together with clear intent, they provide a structured view of loyalty, performance, and friction.
Keep each metric tied to one decision. If a score does not change what a team does next, it is probably the wrong question for that moment.
Frequently asked questions
Is NPS just another name for CSAT?
Show answer
No. NPS looks at the whole relationship and asks about willingness to recommend, while CSAT rates a single experience. A customer can be pleased with one call and still be unlikely to recommend the company.
Which metric suits a support team best?
Show answer
CSAT and CES both attach to specific interactions, so a team can send either one when a ticket closes. QuestionPro says that in industries like telecom and banking, where switching costs are low, CES often highlights process problems that CSAT may miss.
Why add an open-ended question to an NPS survey?
Show answer
A bare score shows how loyal customers are but not the reason. IBM says some companies include open-ended questions to dig deeper into which product features customers like.
Do you need all three metrics?
Show answer
Not necessarily. CustomerGauge observes that some companies will only need one tool to adapt for all their purposes, while others will need a few.



