How to Calculate NPS
A Net Promoter Score comes from one question: How likely are you to recommend us to a friend or colleague? People answer from 0 to 10. The maths takes a minute. Most teams go wrong in the step after it: reading a score from 60 responses as if it were exact.
How to calculate NPS
NPS = % of promoters − % of detractors. Sort every response into three groups: promoters (9–10), passives (7–8) and detractors (0–6). Divide each group by the total number of responses, including passives, then subtract the detractor percentage from the promoter percentage. The result is a number from −100 to +100, not a percentage.
Worked example: 200 responses
| Group | Scores | Responses | Share of all 200 |
|---|---|---|---|
| Promoters | 9–10 | 110 | 55% |
| Passives | 7–8 | 50 | 25% |
| Detractors | 0–6 | 40 | 20% |
| NPS | 55 − 20 = +35 |
Passives never enter the subtraction, but they stay in the total. Drop them from the denominator and the same responses would give 110/150 − 40/150 = +47: a better-looking number that describes nobody.
What is NPS?
Net Promoter Score measures how willing customers are to recommend you. It was created by Fred Reichheld with Bain & Company and Satmetrix (now part of NICE). Bain uses a 0-to-10 scale on purpose: zero is never mistaken for a good score, and eleven points give people room to separate “fine” from “would tell a colleague”.
The three groups behave differently, which is why the score ignores the middle:
- Promoters (9–10) stay longer and recommend you unprompted.
- Passives (7–8) are satisfied but not attached; a competitor’s offer can pull them away.
- Detractors (0–6) are unhappy, most likely to churn, and the main source of negative word of mouth.
Common mistakes when calculating NPS
- Averaging the scores. 100 answers can average 7.9 out of 10 and still produce an NPS of +20. The average hides the split; NPS is built on it.
- Leaving passives out of the total. See the example above: +35 becomes +47.
- Writing it as a percentage. “35%” suggests a share of something. NPS is +35 on a scale from −100 to +100.
- Comparing scores from different channels as if they were one number. An in-app survey reaches active users; an email survey reaches whoever opens it. Track each channel separately.
How many responses do you need?
A score from a small sample moves a lot on its own. Using the example split above (50% promoters, 20% detractors, so NPS +30), the 95% margin of error is:
| Responses | NPS +30 could really be anywhere between |
|---|---|
| 50 | +8 and +52 (±22) |
| 100 | +15 and +45 (±15) |
| 200 | +19 and +41 (±11) |
| 400 | +22 and +38 (±8) |
| 1,000 | +25 and +35 (±5) |
So a drop from +34 to +28 between two months of 100 responses is not news. A change you can act on either persists over several periods or is large compared with the margin. (Calculation: the variance of one response is %promoters + %detractors − NPS², and the margin is 1.96 × √(variance ÷ responses).)
What is a good NPS score?
Any score above 0 means you have more promoters than detractors. Beyond that there is no universal “good”: typical scores differ a lot between industries, and a published benchmark is only comparable if it was collected the same way (same question, same channel, same kind of customer). The two comparisons that hold up:
- Your own trend, measured the same way each period, with the margin above in mind.
- The split behind the score. +30 from 50% promoters and 20% detractors is a different business from +30 from 35% and 5%: the first has a detractor problem, the second a passive one.
How to write the NPS survey question
Keep the standard wording so your score stays comparable: “How likely is it that you would recommend [product] to a friend or colleague?”, 0 to 10, with the ends labelled (“Not at all likely” / “Extremely likely”). Then ask one open question: “What’s the main reason for your score?” That answer is where the actionable part of NPS lives.
Two timing rules do most of the work:
- Ask after value, not mid-task. After someone finishes something in your product, not while they’re checking out or stuck.
- Don’t re-ask the same person too often. Once a quarter per user is a common rhythm; whatever you choose, keep it fixed so the periods compare.
How to collect NPS responses
- In-app: reaches active users at the moment they use the product; the answer can come with context about what they were doing.
- Email: reaches people who don’t log in often, including those drifting away. Useful for relationship NPS; the people who answer are self-selected, so read it separately.
- After support: measures the support experience more than the product; keep it separate from product NPS.
- Website: fine for signed-in customers; anonymous visitors aren’t customers yet.
How to run an NPS survey in-app with Userback
Userback’s in-app surveys have a native Net Promoter Score® question (0–10), so the calculation above is done for you:
- Start from the NPS survey template and add the follow-up question.
- Choose when it appears: after a set time, when someone clicks a specific element, on a rage click, on a custom event you send from your app, or from your own code with the JavaScript SDK (
Userback.openSurvey(...)). - Choose who sees it: a segment from your user data (for example, customers on a paid plan, or users who signed up more than 30 days ago), specific pages, and desktop, tablet or phone.
- Control repetition: show it once, or stop after a response and restart after a set number of days.
- Read the result: the survey overview shows the score with the breakdown of promoters, passives and detractors. Each respondent’s profile shows their NPS next to their other feedback, such as bug reports and feature requests, so a detractor’s score comes with what they’ve told you before.
- Share it: send responses to Slack, Microsoft Teams or a webhook, or export them as CSV.
Surveys are included on Userback’s Business and Business Plus plans, and you can try them free for 14 days.
FAQ
Can NPS be negative?
Yes, whenever there are more detractors than promoters. The lowest possible NPS is −100, when every respondent is a detractor.
Why are 7 and 8 not counted?
They are counted, in the total. Passives are left out of the subtraction because they are satisfied without being likely to recommend you, so they don’t push the score either way. They do dilute it.
Is NPS the same as CSAT?
No. NPS asks about recommendation and describes the overall relationship. CSAT asks how satisfied someone was, usually with one interaction, and is reported as the share of satisfied answers. Many teams run both: CSAT after specific moments, NPS on a steady rhythm. More in our guide to CSAT surveys.
Should I round NPS?
Report whole numbers. With the margins shown above, a decimal place suggests precision the score doesn’t have.
How often should I run an NPS survey?
Often enough to see a trend, rarely enough not to annoy people: many teams ask each user about once a quarter and read the score monthly or quarterly as responses come in.
Net Promoter®, NPS®, NPS Prism®, Net Promoter System® and the NPS-related emoticons are registered trademarks of Bain & Company, Inc., NICE Systems, Inc., and Fred Reichheld. Net Promoter Score℠ is a service mark of Bain & Company, Inc., NICE Systems, Inc., and Fred Reichheld.