How to Choose Between CSAT, NPS and CES

Here is how I choose between CSAT, NPS and CES on the teams I work with. CSAT vs NPS vs CES comes down to the question you need answered. CSAT (Customer Satisfaction Score) measures satisfaction with one interaction. NPS (Net Promoter Score) measures loyalty to the whole relationship. CES (Customer Effort Score) measures how much effort a customer spent to get something done. None of them is "the best"; I pick the one that matches the decision I need to make.

Each has a free calculator: CSAT calculator, NPS calculator and CES calculator.

What is the difference between CSAT, NPS and CES?

	CSAT	NPS	CES
Question asked	"How satisfied were you with this experience?"	"How likely are you to recommend us to a friend or colleague?"	"The company made it easy for me to resolve my issue."
Scale	1–5	0–10	1–7 (agree/disagree)
Level	One touchpoint	Whole relationship	One task or journey
When to send	Right after an interaction	Two to four times a year	Right after a task
Weak spot	Mood of the moment	Vague without a "why"	Says little about attachment

How do you calculate CSAT, NPS and CES?

CSAT (%) = responses of 4 or 5 ÷ all responses × 100
NPS      = % promoters (9–10) − % detractors (0–6)        range −100 to +100
CES (%)  = responses of 5, 6 or 7 ÷ all responses × 100   ("agree" share on a 1–7 scale)
CES (avg)= sum of all scores ÷ number of responses        (alternative: the mean score)

Passives (7–8) count in the NPS total but in neither percentage.

Pick one CES method and keep it. The share and the average move differently, and switching breaks the trend.

A worked example with all three

Say one quarter of surveys gives you:

Survey	Responses	Result	Score
CSAT after tickets	500	400 answered 4 or 5	80%
NPS (relationship)	200	90 promoters (45%), 60 passives, 50 detractors (25%)	+20
CES after refunds	300	210 answered 5, 6 or 7	70%

All three can be true at once. Support handles single tickets well (80% CSAT). The refund process is harder than it should be (30% did not find it easy). And something in the wider relationship holds loyalty back (+20 NPS). One score on its own would have told only a third of the story. Check your own figures in the CSAT calculator, NPS calculator and CES calculator. If you are unsure whether 200 answers is enough, the survey sample size calculator will tell you.

Which metric should I use?

If your question is...	Use	Send it...
Are we handling tickets well?	CSAT	Within a day of the ticket closing
Where do customers struggle?	CES	Right after setup, billing changes, refunds or cancellations
Will customers stay and recommend us?	NPS	Two to four times a year, never after an open escalation

CSAT works best right after a specific interaction, while it is fresh. Link each score to the team or process it reflects. More detail in how to calculate CSAT.

NPS tracks the relationship over time and compares segments, for example new accounts against long-standing ones. The score is the least useful part. The follow-up question, "What is the main reason for your score?", is where the insight lives. More in how to calculate NPS.

CES belongs where effort is the risk. High effort wears loyalty down even when the outcome is fine. The Harvard Business Review article that introduced CES argued that reducing effort does more for loyalty than trying to delight customers, and that matches what I have seen in support queues.

What does a survey dashboard look like?

Metric	Target	This month	Last month	Change
CSAT (after tickets)	≥ 85%	80%	83%	−3 pts
CES (after refunds)	≥ 75%	70%	66%	+4 pts
NPS (quarterly)	≥ +25	+20	+18	+2
Survey response rate	≥ 20%	17%	19%	−2 pts

Watch the response rate too. A score from a shrinking, self-selected group can move for the wrong reasons. The survey response rate calculator tracks it.

What I've seen in practice

On one support team I worked with, we started calling back every customer who gave a CSAT of 1 or 2 within a day. Many of those partners had simply misunderstood a refund timeline, and a five-minute call fixed it. Those calls also gave us the most honest feedback we collected all year.

For a while I reported our NPS every quarter, the number moved a few points up or down, and nobody did anything with it. Once we grouped the comments by theme and sent each theme to the team that owned it, the same survey started producing decisions.

With CES, the low scores almost always traced back to process: too many handoffs, unclear instructions, being asked for the same booking reference twice. Each of those had an owner and a fix.

Three rules I now follow: cap surveys at one per person per month, skip transactional surveys while an escalation is open, and close the loop on every low score. Asking for feedback you then ignore does more harm than not asking. And I do not compare scores across industries: they depend on the sector and on how the question is worded. Your own trend is the benchmark.

Frequently asked questions

Which is better, NPS or CSAT?

Neither in general. CSAT is more precise for one touchpoint. NPS is better for tracking loyalty to the relationship over time. Most teams need both, each for its own job.

Can I send CSAT and NPS in the same survey?

You can, but the answers blur: customers rate the relationship based on the ticket they just had. I keep them separate and send NPS on its own schedule.

What is a good CES score?

It depends on the scale and the task. On a 1–7 "agree" scale, track the share of 5–7 answers per journey and compare it with the previous month rather than with other companies.

How many responses do I need?

Enough that the result would not change much with a few more answers. For small bases, report the count next to the score so readers can judge it.

Sources

• Bain & Company, Measuring Your Net Promoter Score: the NPS question, scale and calculation.
• Harvard Business Review, Stop Trying to Delight Your Customers (Dixon, Freeman and Toman, 2010): the research behind customer effort.
• American Customer Satisfaction Index: national customer satisfaction scores by industry.