NPS measures loyalty by asking how likely a customer is to recommend you, CSAT measures satisfaction with a specific experience, and CES measures how much effort it took a customer to get something done.
All three are customer-satisfaction metrics, but they answer different questions and use different scales. NPS runs from -100 to +100 and tracks the overall relationship. CSAT is reported as a percentage of satisfied users and tracks a single interaction. CES sits on a 1 to 7 effort scale and tracks friction. Picking the right one is mostly a question of what decision you are trying to make.
This guide defines each metric with its exact survey question, gives you real benchmarks, compares them side by side, and ends with a clear recommendation for early-stage teams. The short version: the score is the lagging indicator, the open-text comment behind it is the signal.
NPS: Net Promoter Score
NPS measures loyalty. It asks one question, verbatim: “How likely are you to recommend [product] to a friend or colleague?” Respondents answer on a 0 to 10 scale.
You then bucket the answers and do a little subtraction:
- Promoters (9 to 10). Loyal enthusiasts who will refer others.
- Passives (7 to 8). Satisfied but unenthusiastic, and ignored in the math.
- Detractors (0 to 6). Unhappy users who can damage your brand by word of mouth.
The score is the percentage of promoters minus the percentage of detractors, which lands somewhere between -100 and +100. If 60 percent are promoters and 20 percent are detractors, your NPS is 40. The trap is that ignoring passives makes the number jumpy on small samples, so NPS only stabilizes with a few hundred responses behind it.
What counts as good
Any positive score means more promoters than detractors. Above 30 is good, above 50 is excellent, above 70 is rare. But benchmarks are wildly industry-dependent: SaaS tends to average 30 to 40, while telecom and insurance often live in the teens. The only comparison that matters is against your own trend line and your direct competitors.
CSAT: Customer Satisfaction Score
CSAT measures satisfaction with a specific experience, not the whole relationship. The standard question is: “How satisfied were you with [experience]?” answered on a 1 to 5 scale, from very dissatisfied to very satisfied.
The score is the percentage of respondents who picked the top one or two boxes (4 and 5, or “satisfied” and “very satisfied”). If 80 of 100 people rate you a 4 or 5, your CSAT is 80 percent. It is intuitive, easy to explain to a board, and stabilizes on small samples because it averages across everyone instead of subtracting buckets.
Where it fits
CSAT shines right after a discrete moment: a support ticket closing, an onboarding step completing, a feature being used for the first time. Fire it transactionally, in context, while the experience is fresh. A typical good CSAT sits around 75 to 85 percent, though it skews high because dissatisfied users often just leave without answering.
CES: Customer Effort Score
CES measures friction. The modern phrasing is an agreement statement: “[Product] made it easy for me to handle my issue,” rated on a 1 to 7 scale from strongly disagree to strongly agree. Higher is better, meaning lower effort.
CES came out of research showing that reducing customer effort predicts loyalty more reliably than trying to delight people. Customers do not reward you for going above and beyond nearly as much as they punish you for being hard to deal with. So instead of asking “did we wow you,” CES asks “did we get out of your way.”
When CES is the sharpest tool
If your product lives or dies on friction (signup, checkout, self-service support, getting to first value), CES is the most actionable of the three. A high-effort score points directly at a flow you can fix, where an NPS number just tells you people are unhappy without saying where. Use it right after the user finishes a task you care about.
NPS tells you whether they like the relationship. CSAT tells you whether they liked a moment. CES tells you whether you got out of their way. None of them tell you what to build next. The comment does.
NPS vs CSAT vs CES At A Glance
Same three metrics, lined up across the dimensions that actually matter when you are choosing one.
| NPS | CSAT | CES | |
|---|---|---|---|
| What it measures | Loyalty and the overall relationship | Satisfaction with a specific experience | Effort to get something done |
| The question asked | How likely are you to recommend us? | How satisfied were you with this? | We made it easy to handle my issue. |
| Scale | 0 to 10, scored -100 to +100 | 1 to 5, reported as % satisfied | 1 to 7 agree/disagree |
| When to use | Quarterly, with real volume | Right after a discrete moment | Right after a friction-heavy task |
| Biggest weakness | Noisy on small samples, says nothing about why | Skews high, only covers one touchpoint | Narrow scope, weak on emotional loyalty |
Which One Should You Actually Track?
The honest answer for most early-stage teams is: not NPS, at least not yet. Here is the reasoning, in order.
- Under a few hundred users, lean on CSAT or CES. Both give you a usable read from a handful of responses, where NPS will bounce around so hard you cannot tell a real change from sampling noise.
- Pick CES if your problem is friction. If users churn during onboarding or rage-quit a checkout, a 1 to 7 effort score fired right after that flow points straight at the thing to fix.
- Pick CSAT if you want a simple, board-friendly satisfaction read. It is the easiest to explain and the easiest to wire to a specific moment like a support reply or a feature launch.
- Add NPS once you have volume. When you can reliably get a few hundred responses a quarter, NPS becomes a clean relationship trend line and a number investors recognize.
But here is the point underneath all of it. Every one of these numbers is a lagging indicator. By the time your NPS drops, the thing that caused it shipped weeks ago. The score does not change your roadmap. What changes your roadmap is the open-text comment a user leaves next to the score: “the export keeps timing out,” “I could not find where to invite my team,” “your mobile view is broken.” That is the signal. Treat the score as a prompt to go read the comments.
Related Reading
- Why Companies Ignore Customer FeedbackRead
- Best User Feedback Tools (2026 Comparison)Read
- How To Organize Feature Requests Without DrowningRead
If You Want To Try Usero
A score tells you something is wrong, but never what to fix. Usero captures the open-text “why” behind every rating, clusters the recurring themes so duplicates stop piling up, and can open a draft pull request with a first-pass fix for the request that keeps coming back. The widget is open source on npm and drops into a React app in three lines. Spin up a widget and start collecting the comments behind your numbers this week.
Frequently Asked Questions
What is a good NPS score?
Any positive NPS (above 0) means you have more promoters than detractors. Above 30 is good, above 50 is excellent, and above 70 is world-class. Benchmarks vary by industry though: SaaS averages 30 to 40, while financial services and telecom often sit in the teens. Compare against your own past scores and your direct competitors, not a global number.
Should a startup track NPS or CSAT?
Early on, track CSAT. NPS asks how likely someone is to recommend you, which requires enough usage and enough peers to recommend you to, so it is noisy under a few hundred respondents. CSAT measures satisfaction with a specific interaction and gives you a usable signal from your first dozen responses. Add NPS once you have the volume for it to stabilize.
Can you use NPS and CSAT together?
Yes, and most mature teams do. They measure different things: NPS captures overall relationship and loyalty, CSAT captures satisfaction with a specific moment like onboarding or a support ticket. A common setup is a quarterly NPS survey for the relationship plus transactional CSAT pings right after key interactions. Just do not average them together, they are different scales.
What is CES and when does it matter?
CES (Customer Effort Score) asks how much effort a user had to expend to get something done, usually on a 1 to 7 agree/disagree scale. It matters most for support and self-service flows, where research shows reducing effort predicts loyalty better than delighting customers does. If your product lives or dies on friction (onboarding, checkout, ticket resolution), CES is the sharpest of the three.
How many responses do you need for NPS to be meaningful?
As a rough floor, aim for at least 100 responses before you read much into an NPS number, and ideally several hundred. Because NPS subtracts detractors from promoters and ignores the middle, a handful of responses can swing the score by 20 or 30 points. CSAT and CES stabilize on smaller samples because they average across all responses.
What is the difference between NPS, CSAT, and CES?
NPS measures loyalty (would you recommend us) on a -100 to +100 scale. CSAT measures satisfaction with a specific experience, reported as the percentage of users who rated it positively. CES measures how easy it was to get something done, on a 1 to 7 effort scale. NPS is about the relationship, CSAT and CES are about specific moments.
Are these scores worth tracking at all for an early-stage product?
The number itself is a lagging indicator and rarely changes a roadmap on its own. What is worth tracking is the open-text comment attached to every score, because that tells you why someone rated you the way they did. Treat the score as a trigger to read the comment, and the comment as the thing that actually drives what you build next.
Continue reading
How to Collect User Feedback In-App: 7 Methods Compared
How to collect user feedback in-app: 7 methods compared (widget, microsurveys, session replay, voting, chat) with response rates and what to ship first.
9 min read
AI-Generated PRs From User Feedback: How It Works
How AI-generated PRs from user feedback work: collect, cluster, draft a GitHub pull request, then a human reviews and merges. Honest about the limits.
8 min read
Feature Request Management: A Complete Framework
A repeatable 5-stage framework for feature request management: capture, cluster, prioritize with RICE, decide, and close the loop.
10 min read
Build a feedback loop your team actually uses
Usero collects, clusters, and turns user feedback into shipped fixes.
Get started free