Someone has asked you to “report on customer experience,” and you have a pile of survey scores, a churn number from finance and a support team measuring how fast it closes tickets. Customer experience metrics turn that pile into a short set of numbers that leadership can follow and teams can act on. This article sits within our wider guide to customer experience analytics and covers the three types of CX metrics, how NPS, CSAT and customer effort score differ, how to calculate each one and how to choose and report the right customer experience KPIs.
Key takeaways
- CX metrics come in three types: perception (NPS, CSAT, customer effort score), behavior and outcome (retention, churn, repeat purchase, share of wallet) and operational (resolution time, first contact resolution).
- NPS tracks the overall relationship, CSAT tracks satisfaction with a moment, and customer effort score tracks how easy a task was.
- Always report a perception metric with its base (the number of responses) and test whether a change is significant before calling it a trend.
- Perception metrics earn their place only when you link them to outcomes such as retention and spend.
What are customer experience metrics?
Customer experience metrics are numbers that describe how customers experience your company. They fall into three types, and a useful program has at least one of each.
| Type | What it tells you | Examples | Typical source |
|---|---|---|---|
| Perception | How customers feel and think | NPS, CSAT, customer effort score | Relationship and transactional surveys, reviews |
| Behavior and outcome | What customers actually do | Retention, churn, repeat purchase rate, share of wallet | CRM, billing, product usage |
| Operational | How well you deliver the experience | Average resolution time, first contact resolution | Support tickets, web and app analytics |
Perception metrics are leading indicators: a drop often shows up before customers leave. Outcome metrics are the results the business cares about. Operational metrics are the levers teams can pull. The value comes from connecting all three, which is where most programs fall short.
How do NPS, CSAT and customer effort score compare?
The three perception metrics look similar because each is a single survey question. They measure different things.
| NPS | CSAT | Customer effort score (CES) | |
|---|---|---|---|
| Question | ”How likely are you to recommend us to a friend or colleague?" | "How satisfied were you with [interaction or product]?" | "[Company] made it easy for me to handle my issue.” |
| Scale | 0 to 10 | 1 to 5 | 1 to 7, strongly disagree to strongly agree |
| Calculation | % promoters (9–10) minus % detractors (0–6) | % answering 4 or 5 (top-2-box) | Average score, or % answering 5 to 7 |
| What it predicts | Loyalty and word of mouth | Satisfaction with a specific moment | Friction that can lead to repeat contact and churn |
| Best use | Relationship surveys, tracked over time | Transactional surveys after a purchase or contact | Service, onboarding and self-service tasks |
For the full NPS method, see what Net Promoter Score is. For CSAT question design and timing, see our guide to the customer satisfaction survey.
What is customer effort score and how do you calculate it?
Customer effort score measures how easy it was for a customer to get something done: fix a problem, return an item, set up an account. The common version is an agreement statement, “The company made it easy for me to handle my issue,” on a 1 to 7 scale where 7 is strongly agree.
You can report it two ways. The average score is the sum of all answers divided by the number of responses. The percent agree is the share answering 5, 6 or 7. Percent agree is easier for non-specialists to read, while the average picks up movement within the scale. Pick one as your headline and keep it.
Older versions ask how much effort the customer had to put in, where a lower score is better. If you inherit data from that wording, do not mix it with the agreement version in one trend line.
CSAT vs NPS: when should you use each?
The simplest rule is relationship versus transaction. NPS works best in a relationship survey sent on a schedule, such as quarterly, to a sample of all customers. It asks about the company as a whole, so it moves slowly and reflects many experiences at once.
CSAT works best in a transactional survey, sent straight after a purchase, delivery or support contact. It tells you which moments go well and which do not. Using NPS after every ticket, or CSAT as your annual loyalty measure, gives you a number that answers the wrong question.
What does a worked example look like?
The numbers below are illustrative, not from a real company. Suppose a quarter of surveys returns 200 responses to each question.
NPS (0 to 10 scale)
| Group | Responses | Share |
|---|---|---|
| Promoters (9–10) | 90 | 45% |
| Passives (7–8) | 60 | 30% |
| Detractors (0–6) | 50 | 25% |
NPS = 45% − 25% = +20. The step-by-step NPS calculation covers spreadsheet formulas and margin of error.
CSAT (1 to 5 scale)
| Score | 5 | 4 | 3 | 2 | 1 |
|---|---|---|---|---|---|
| Responses | 64 | 76 | 30 | 18 | 12 |
CSAT = (64 + 76) ÷ 200 = 140 ÷ 200 = 70%.
Customer effort score (1 to 7 agreement)
| Score | 7 | 6 | 5 | 4 | 3 | 2 | 1 |
|---|---|---|---|---|---|---|---|
| Responses | 40 | 50 | 40 | 30 | 20 | 12 | 8 |
Sum of scores = (7 × 40) + (6 × 50) + (5 × 40) + (4 × 30) + (3 × 20) + (2 × 12) + (1 × 8) = 280 + 300 + 200 + 120 + 60 + 24 + 8 = 992. Average CES = 992 ÷ 200 = 4.96. Percent agree = (40 + 50 + 40) ÷ 200 = 130 ÷ 200 = 65%.
Now suppose last quarter’s CSAT was 64%, also on 200 responses. Is the 6-point rise real? The standard error of the difference is the square root of (0.70 × 0.30 ÷ 200 + 0.64 × 0.36 ÷ 200), which is the square root of 0.0022, or about 4.7 points. Multiply by 1.96 and the 95% margin is about 9.2 points. A 6-point gain sits inside it, so report it as “up, but not yet significant.”
How do you choose CX KPIs for leadership and for teams?
Leadership and operational teams need different customer experience KPIs, because they make different decisions.
- Leadership: one relationship metric (usually NPS) plus one or two outcomes, such as retention and revenue from existing customers. Report quarterly, by major segment.
- Service and support: customer effort score and CSAT after contact, alongside resolution time and first contact resolution.
- Product and digital: effort and completion on key tasks, from surveys plus web and app analytics.
- Account and commercial teams: retention, repeat purchase and share of wallet by account, with NPS as context.
In our experience, the best test of a KPI is whether a named owner can change it within a quarter. If no one can, it belongs in a quarterly review, not a weekly scorecard. Tying these KPIs to financial results is the job of measuring customer experience ROI.
How should you report CX metrics?
Show the base next to every perception score, so a reader can see that “CSAT 82%” rests on 40 responses rather than 4,000. Test changes between periods and segments for significance, and flag the ones that pass.
Then link perception to outcomes. Compare churn among detractors with churn among promoters, or repeat purchase among customers who rated effort low versus high. Our guide to customer churn analysis shows how to set that comparison up. Once you can say which perception scores move with outcomes, a team can see what drives satisfaction and loyalty by segment and journey and focus on the drivers that matter.
What are the common mistakes with customer experience metrics?
- Tracking perception only. Without retention or spend alongside, you cannot show that a higher score matters.
- Mixing scales and wordings. A 1 to 5 CES and a 1 to 7 CES are not one trend line.
- Reporting scores without bases. Small segments swing widely from noise.
- Using the wrong metric for the moment. NPS after a password reset measures the wrong thing.
- Ignoring the comments. The verbatims explain the score; voice of customer analysis shows how to code them into priorities.
What should you do next?
List the CX metrics you report today and sort them into perception, outcome and operational. Fill the gaps, give each metric an owner, and add a base and a significance flag to every score. When the set is stable, a continuous view of satisfaction drivers, journey friction and emerging issues helps your team spot what changed between reporting cycles.