NPS - Net Promoter Score: Definition & Calculation

Picture two managers walking into the same Monday meeting with the same number on the slide. One of them serves a customer base split down the middle, half of it delighted and half of it ready to walk. The other serves a customer base that mostly shrugs. Both report an identical Net Promoter Score.

The Net Promoter Score is a referral metric: you ask on a scale from 0 to 10 how likely a recommendation is, sort the answers into three groups and subtract the percentage of detractors from the percentage of promoters. The result always lands between minus 100 and plus 100. By the end of this article you will have worked the math yourself, you will know how to put a score in context, and you will spot the moments when the number promises more than it can deliver.

📌 Key takeaways

• The NPS measures willingness to recommend on a scale from 0 to 10.
• Promoters answer 9 or 10, detractors answer 0 through 6.
• NPS equals percent promoters minus percent detractors, passives drop out.
• The score always sits between minus 100 and plus 100.
• Fred Reichheld introduced the NPS in the Harvard Business Review in 2003.

What is the Net Promoter Score?

The Net Promoter Score is a customer loyalty metric built out of one question: how likely is it that you would recommend us to a friend or a colleague? Answers come in on a scale from 0 to 10.

American loyalty researcher Fred Reichheld developed the metric and presented it in December 2003 in the Harvard Business Review under the title "The One Number You Need to Grow". His argument was deliberately blunt: out of everything you could ask a customer, one question says the most about future growth, and that is the question about recommending you. Making a recommendation puts your own reputation on the line, which is exactly why it carries more weight than a checkmark next to "satisfied".

Worth knowing before the number reaches an investor deck: "Net Promoter" and "NPS" are registered trademarks of Bain & Company, Inc., NICE Systems, Inc. and Fred Reichheld. Applying the method is free and publishing your own score is fine, but commercial use of the marks requires a license and a source credit under Bain's own trademark and licensing terms (as of September 2026). For a survey inside a club, a class or your own store, none of that matters. For an annual report, it does.

Keep the NPS apart from ordinary satisfaction measurement. Satisfaction looks backward ("how happy were you?"), a recommendation looks forward to something a person might actually do. The two can sit far apart, because plenty of customers are content enough to stay and still not convinced enough to bring anyone with them.

Promoters, passives and detractors: the three groups

The eleven answer options of the NPS collapse into three groups: promoters at 9 and 10, passives at 7 and 8, detractors anywhere from 0 to 6. That split is the real core of the method, and it is lopsided, which throws almost everyone the first time.

So seven of eleven possible answers land in the critical bucket. Reichheld defends that cut with behavior rather than symmetry: in his data, people who gave a 9 or a 10 acted measurably differently from everyone else. They repurchased more often and brought new customers along. The line sits where behavior changes, not in the middle of the scale.

Running from 0 to 10 makes this an eleven point rating scale. Looks like a ruler with even spacing, but a ruler it is not: nobody can show that the step from 6 to 7 means the same thing as the step from 9 to 10. Which is precisely why the NPS never takes an average. Counting shares is the safer move.

How to calculate the NPS: formula and worked example

The NPS is calculated by subtracting the percentage of detractors from the percentage of promoters. Passives fall out of the subtraction entirely and only count toward the total number of responses. Report the result as a whole number, with no percent sign attached.

1. Sort the answers into three groups. Count how many responses land on 9 to 10, on 7 to 8 and on 0 to 6.
2. Turn the groups into percentages. Divide each group by the total number of responses and multiply by 100.
3. Subtract detractors from promoters. Whatever is left is your Net Promoter Score. From the three group figures you can also calculate the NPS directly.

Run it once with 200 completed questionnaires from a customer survey and the arithmetic stops feeling abstract.

48 percent promoters minus 25 percent detractors gives an NPS of plus 23. The 54 passive answers never enter the subtraction, yet they still push the result down indirectly, because they inflate the denominator and shrink both shares. Take the same 96 promoters and 50 detractors out of only 150 responses and the score climbs to plus 31.

💡 Tip

Always publish the three shares and the number of responses next to the score. A plus 23 from 200 answers and a plus 23 from 12 answers look identical on a slide and mean nothing like the same thing.

What is a good NPS score?

No universal threshold separates a good NPS from a bad one. Only one thing is certain: from plus 1 upward promoters outnumber detractors, below zero the criticism wins. Everything past that depends on industry, country and how the survey reached people, which makes borrowed numbers the most common source of error.

Your own time series is the reading that holds up. Measure the NPS across several quarters with the same wording, the same moment in the customer relationship and the same audience, and the movement starts telling you something. Bain & Company sells industry benchmarks under the name NPS Prism. The fact that comparison data can carry a business model at all tells you how little a bare score means on its own.

One effect gets underestimated in cross border programs over and over: people in different countries use the same scale differently. According to van Herk, Poortinga and Verhallen (2004) in the Journal of Cross-Cultural Psychology, a tendency to agree and a pull toward the extremes showed up more clearly in the Mediterranean countries studied than in northwestern Europe. The United States was not among the six countries in that study (Greece, Italy, Spain, France, Germany and the United Kingdom), so nothing in it places American respondents in either camp. For a US company the lesson runs the other way: if you survey across the States and Europe, do not pool the scores, and never treat a published European NPS as a yardstick for your own.

⚠️ Careful

Never benchmark your score against numbers other companies publish. Those figures come from different samples, different moments and often different languages. Compare your current score with your last one instead, and keep the survey identical so the comparison means something.

Asking the NPS question correctly

The NPS question only does its job when wording, scale and timing line up. Change the phrasing, shorten the scale or fire it off whenever you feel like it, and you are measuring something else, which breaks the comparison with your own previous round.

Put the question at the end of the questionnaire or let it stand entirely on its own. Ask it right after ten detail questions about shipping speed and packaging, and people answer with shipping speed and packaging in mind instead of the whole relationship. How you structure a questionnaire therefore helps decide what number comes out the other end.

Checklist for the NPS question

• Wording stays word for word identical across every wave.
• The scale runs the full 0 to 10, never shortened.
• Both ends carry labels, for example "not at all likely" and "extremely likely".
• An open follow up question asking why sits directly underneath.
• The moment in the customer relationship is the same every time.
• Answering stays voluntary, never a required field.

The open follow up question is the most valuable part of the whole survey. The number tells you something is off, the open question tells you what. Reichheld built it into the method from day one, and in practice it is still the first thing teams cut.

For staff the same logic runs under the name eNPS, asking how likely an employee is to recommend their own employer. Identical arithmetic, so you can calculate the eNPS exactly the same way, trickier reading, because groups are small and anonymity is hard to protect in a team of 30. Measuring employee satisfaction cleanly depends more on the process than on the metric.

What the NPS does not do

The NPS is a simplification on purpose, and every simplification bills you for it. Anyone putting the score in front of an executive team should know three limits, because they explain how two identical scores can describe completely different situations.

None of these limits makes the metric worthless. They only mark where its jurisdiction ends. As an early warning system the NPS earns its keep. As the sole basis for a decision, it does not.

The NPS throws information away

Eleven gradations become three groups, and three groups become one number. A company with 50 percent promoters and 50 percent detractors scores zero. A company with 20 percent promoters, 60 percent passives and 20 percent detractors also scores zero. The first describes a customer base pulled apart, loud voices on both sides. The second describes a customer base that is evenly lukewarm. Same number, two entirely different jobs to do.

Small samples make the NPS jump

Since the NPS is the difference between two shares, it swings harder than an average taken from the same sample. At 50 responses, one answer is worth 2 percentage points. Move a single person from detractor to promoter and the score travels 4 points without anything at the company having changed. Report a 5 point improvement off 40 responses and you are reporting noise. How big a sample needs to be is not a side question here.

The link between NPS and growth is contested

Reichheld's central claim was that the NPS predicts growth better than rival loyalty measures. Keiningham, Cooil, Andreassen and Aksoy tested it in 2007 in the Journal of Marketing, using data from 21 companies and more than 15,500 interviews from the Norwegian customer satisfaction barometer. One of the yardsticks they measured Net Promoter against was the American Customer Satisfaction Index (ACSI), the national cross industry index that has run since 1994, was developed at the University of Michigan through the National Quality Research Center at the Ross School of Business and rests on roughly 350,000 customer interviews a year. Their result: the claimed clear superiority of the NPS over other measures did not hold up in precisely the industries Reichheld had held up as showcases. The paper won the 2007 H. Paul Root Award from the Marketing Science Institute. So the academic fight over the NPS was never abstract for American readers. It was fought against the very index they already had.

The NPS is a thermometer, not a diagnosis. It tells you something is hot or cold, never why.

Common mistakes in NPS surveys

Most unusable NPS results come from the circumstances of the survey rather than from a slip in the arithmetic. Three patterns show up again and again, and all three cost nothing to avoid.

What they share is direction: each one pushes the score upward. Which makes them sneaky, because an inflated NPS never gets questioned. Good news rarely does.

The NPS question arrives at the wrong moment

Right after a complaint gets resolved, people answer relieved and generous. Three weeks later, considerably more sober. Both measurements are correct on their own, yet strung together they produce a time series that says nothing about the company and everything about when you chose to ask. Fix the moment once, then hold it, even when some other moment looks more flattering this quarter.

The NPS turns into a target instead of a measurement

Attach a bonus to the score and people start begging for points. "If you were happy today, please give us a 10, anything lower counts as a fail for us" is a sentence that destroys the metric outright. Reichheld named the problem himself in 2021 in the Harvard Business Review under the title "Net Promoter 3.0" and proposed a counterweight straight out of accounting, the Earned Growth Rate. That figure measures how much revenue growth genuinely comes from returning customers and the referrals they make.

Only part of the customer base gets asked

Send the survey to your newsletter list and you are asking people who already raised their hand. Email only active accounts and you are asking the ones who stayed. Both routes lift the score without anyone manipulating a thing. Whether your result speaks for your whole customer base comes down to who ends up in the sample, and the piece on representative online surveys covers what that takes.

Conclusion

The Net Promoter Score is one of the few metrics everyone on a team actually understands, and that is its biggest strength. Treat it as an early warning system and a conversation starter, not as proof. What matters is not the score itself but what sits in the open answers underneath it, and whether the number moves across several rounds measured the same way. A lone NPS with no shares, no response count and no point of comparison says almost nothing.

Where to go next

• Want to work out more metrics from your survey? Calculate metrics
• Planning the survey that goes with it? Customer surveys: guide and examples
• Answers already in hand? Analyzing a questionnaire in five steps
• Wondering why an average is the wrong tool here? Ordinal scale explained simply

Our reading tip: Anyone who wants the original argument in full will find it in Reichheld's 2003 essay in the Harvard Business Review. The counterposition sits in the 2007 study by Keiningham and colleagues in the Journal of Marketing. Read together, the two give you a more honest picture than any summary.

Want to measure the NPS in your own survey?

empirio.ai, an online survey tool from Germany, offers NPS as its own question type and the analysis calculates the score for you. Browse customer feedback templates