Reading your Net Promoter Score
Net Promoter Score distils customer loyalty into one number from −100 to +100. Here's the formula with a worked example, what a good score looks like, how you compare to your industry, the survey question itself, and how to move the number.
How to calculate NPS
Net Promoter Score comes from a single question — “How likely are you to recommend us?” — answered 0 to 10. Sort the answers into three groups, take the percentage of promoters, subtract the percentage of detractors, and that's your score.
NPS = % Promoters (9–10) − % Detractors (0–6)
Worked example: 200 people answer. 110 rate you 9 or 10 (promoters), 50 rate 7 or 8 (passives), and 40 rate 0 to 6 (detractors). Promoters are 110 ÷ 200 = 55%. Detractors are 40 ÷ 200 = 20%. Your NPS is 55 − 20 = +35.
Three things trip people up. Passives still count — they inflate the denominator and therefore drag both percentages down, which is exactly the intent: a 7 is not a good rating. The result is a number, not a percentage: +35, never 35%, because it is the difference between two percentages. And the 0–6 detractor band is deliberately wide, so a score that looks harsh is working as designed.
What is a good NPS score?
Any score above zero means more promoters than detractors. Beyond that, “good” is relative — here's the whole range, and your live result lands on this same scale the moment you calculate.
World-class — Exceptional loyalty. You're in the top tier of any industry.
Great — Strong loyalty — promoters clearly outweigh detractors.
Good — Positive, but with clear room to win over more customers.
Needs work — You have more detractors than promoters — worth digging into why.
How you compare by industry
Averages swing a lot by field — hospitality and ecommerce run high, telecom and utilities run low. Compare against your own field, and against your own past scores.
| Software / SaaS | 40 |
| Ecommerce / Retail | 45 |
| Financial services / Banking | 40 |
| Healthcare | 38 |
| Hospitality / Hotels | 50 |
| Travel / Airlines | 35 |
| Telecom / Utilities | 25 |
Turn the score into action
The number only moves when you follow up — and you should ask each group something different. Promoters to amplify, passives to close the gap, detractors to recover. Copy-paste these into your survey.
Amplify the love — turn enthusiasm into referrals and proof.
- What do you love most about working with us?
- Would you be open to leaving a quick review or sharing your story?
- Is there anyone you'd recommend us to that we could help?
- What nearly stopped you from giving us a 10?
Find the gap — one fixable thing stands between a 7 and a 9.
- What's the one thing we could do to earn a higher score?
- Where did our product or support fall short of what you expected?
- What would make you recommend us without hesitation?
- Is there a feature or change that would make your life easier?
Recover and diagnose — a fast, human reply can save the account.
- What went wrong, and how can we make it right?
- What was the most frustrating part of your experience?
- If you could change one thing about us, what would it be?
- Can we set up a quick call to understand what happened?
NPS survey questions
The standard wording is fixed, and it is worth keeping it that way: “How likely are you to recommend [company] to a friend or colleague?” on a 0–10 scale. The value of NPS is that it is comparable across companies and across years; rewriting the question breaks that, and a bespoke loyalty question is no longer an NPS.
What you can change is the follow-up. Always ask one open question after the rating — “What is the main reason for your score?” — because the number tells you the temperature and the comment tells you the cause. Keep it optional and keep it to one.
Timing matters more than wording. Relationship NPS goes out on a cycle (quarterly or twice a year) and measures how people feel about you overall; transactional NPS fires after a specific event. Whichever you run, do not send it immediately after a support ticket — that measures the ticket, which is CSAT's job. And never survey the same person twice in a quarter; over-surveying depresses response rates and biases who bothers to reply.
How to improve your NPS
The arithmetic points at where to spend effort. Because detractors subtract and passives do nothing, moving one detractor to a promoter shifts your score twice as much as creating a new promoter from scratch. Start at the bottom of the distribution, not the top.
That means closing the loop: contact every detractor who left a comment, personally, and fix what they raised. Teams that do this consistently see the biggest movement, partly because the fixes compound and partly because a detractor who feels heard often re-rates. Then tally the reasons rather than reacting to individual responses — the tally is your roadmap, and it usually points at a small number of recurring problems rather than a general mood.
Passives are the cheaper second move. They are usually satisfied but unenthusiastic, and the gap between a 7 and a 9 is often one unresolved friction point rather than a fundamental complaint. Speed of support is the most common lever: a help center that answers the routine questions and an AI agent that replies instantly remove the delays that keep people at 7.
Employee NPS (eNPS)
Employee Net Promoter Score applies the same arithmetic to staff instead of customers. The question becomes “How likely are you to recommend [company] as a place to work?”, still on 0–10, and the calculation is identical — percentage of promoters minus percentage of detractors. The calculator above works for eNPS unchanged; only the question differs.
Two differences are worth knowing. eNPS scores run lower than customer NPS almost everywhere, so comparing the two numbers is meaningless — anything above 0 is respectable and above +20 is strong. And anonymity is not optional: employees who suspect their answer is attributable will rate higher than they feel, which produces a comfortable number and no information. If your team is small enough that responses are identifiable anyway, an eNPS survey is probably the wrong instrument, and a conversation is the right one.
