Net Promoter Score (NPS): Definition & Calculation

Picture two managers in the same meeting with the same number on the slide. One customer base is split in half, the other mostly shrugs. Both report the same Net Promoter Score.

The Net Promoter Score is a loyalty metric: you ask customers how likely they are to recommend you on a scale from 0 to 10, sort the answers into promoters, passives and detractors, then subtract the detractor percentage from the promoter percentage. At empirio.ai, an online survey tool from Germany, matrix, ranking and NPS questions are free. By the end you will have done the math yourself, you will know how to read a score in context and you will see where the number flatters you.

📌 Key takeaways

• The NPS measures how likely customers are to recommend you, on a scale from 0 to 10.
• Promoters answer 9 or 10, detractors answer 0 through 6.
• NPS = % promoters − % detractors, passives (neutrals) do not count.
• The score falls between −100 (negative) and +100 (positive).
• Fred Reichheld introduced the NPS in the Harvard Business Review in 2003.

What is the Net Promoter Score?

The Net Promoter Score, or NPS for short, 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 (not at all likely) to 10 (extremely likely).

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.

Promoters, passives (neutrals) and detractors: the three groups

When the results are analyzed, the eleven answer options (0 to 10) collapse into three groups:

• Promoters at 9 and 10: recommend you actively and stick around.
• Passives (neutrals) at 7 and 8: content, but not attached.
• Detractors at 0 to 6: unhappy, and they tell people.

That split is the real core of the method, and it is lopsided, which throws almost everyone the first time. 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 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 +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 +31.

What is a good NPS score?

No universal threshold separates a good NPS from a bad one. Everything depends on industry, country and how the survey reached people, which makes borrowed numbers the most common source of error. The only thing you can say for certain is whether promoters outnumber detractors or the other way around. As a rough guide, our NPS calculator uses the ranges below. Read them as an aid to interpretation, not as a standard:

• Below 0: critical. Detractors outnumber promoters, so start with the open answers.
• 0 to +29: needs work. Promoters at least match detractors, but there is no cushion.
• +30 to +49: good. Fans clearly outnumber critics.
• +50 to +69: very good. At least half of all respondents are promoters.
• +70 to +100: excellent. Rare in practice, which is all the more reason to report the shares and the response count alongside it.

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.

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 survey 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.

Running the NPS yourself: question type, plan and privacy

Running the NPS yourself comes down to three decisions: a question type that stores the 0 to 10 scale as a scale instead of eleven loose checkboxes, a plan that covers the number of responses you expect, and a written agreement with whoever handles the answers on your behalf.

The question type matters more than it sounds. An NPS question built out of a text field or a dropdown leaves you eleven labels to count by hand, and every new wave adds the same manual work again. A dedicated NPS question type holds the scale at 0 to 10, labels both ends and sorts the answers into promoters, passives and detractors on its own, so the score comes out the same way every quarter.

Plan and privacy usually surface later, once the first wave has already gone out. Both are cheaper to settle before you send the link. A survey that collects answers from customers handles personal information even when you never ask for a name, and the paperwork for that sits between you and your provider rather than between you and your customers.

What a free plan should cover

A free plan carries an NPS program as long as the question type and the analysis are included rather than sold as an upgrade. empirio.ai keeps matrix, ranking and NPS questions in the free plan, along with the analysis tools, results as charts and real-time updating. The free plan covers 50 respondents per month, cross tabs and filters start with the Plus plan, and embedding a survey costs nothing (as of October 2026). Everything a first NPS wave needs sits in that list, so you can create a free online survey and read your first score the same week.

Privacy in NPS surveys: the duty first, then the fix

Privacy law in the United States starts with a contract rather than a consent banner. Under the California Consumer Privacy Act, as amended by the California Privacy Rights Act, a business that discloses personal information to an outside provider needs a written contract with that provider. The California Privacy Protection Agency sets out which terms belong in that contract in its CCPA regulations, in the section titled “Contract Requirements for Service Providers and Contractors” (§ 7051). Whether the CCPA covers your company at all depends on its size and its revenue, so settle that question first.

empirio.ai hosts survey data in the EU and provides a written data processing agreement, both already on the free plan (as of October 2026). EU hosting is the point that matters most once respondents sit in Europe, because European rules care about where the answers land. For a California business the written agreement is the piece to look at: read its terms against § 7051 before you sign, because the duty to have the right terms in place rests with the business and not with the provider.

Anonymity is a separate decision from the paperwork. An NPS survey works with no identifying field at all: no name, no email address, no account. Leave those out and the score and the open answers still read fine, you simply cannot go back to the person who wrote one. Follow-up is often worth the trade, and when it is, say so in the invitation instead of collecting contact details quietly.

Embedding the NPS question in your own website

Embedding the NPS question in your own website lifts the response rate, because nobody has to click through to a second page. An embedded question sits in the order confirmation, in the help center or in the footer of a logged-in area, and it carries your own layout instead of a provider's. Embedding a survey is free at empirio.ai (as of October 2026) and a first NPS question takes a few minutes once you create a free account.

Pros and cons of the NPS

The NPS shows up in so many companies because it is so simple. That same simplicity is also what makes it easy to poke holes in. Before you commit to the metric, take an honest look at both sides.

The case for the NPS

• One question that takes seconds to answer, which keeps your response rate high.
• A number between −100 and +100 that any department can read without a briefing.
• Keep the wording and the timing fixed and you get a trend line you can trust, quarter after quarter.
• It fits almost anywhere: after a purchase, after a support ticket or once a year across your entire customer base.
• Paired with the open follow up question, it tells you the reasons behind the score, not just the score.

The case against the NPS

• The distribution disappears: a polarized customer base and an indifferent one can land on the exact same score.
• Passives at 7 and 8 never show up in the result, even though they are often the largest group.
• With only a handful of responses, the score jumps around while nothing real has changed.
• It measures an intention, not a recommendation anyone actually made.
• Whether the NPS predicts growth better than other metrics is still contested in the research.

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.

How the NPS differs from other metrics

The NPS looks ahead and reflects loyalty to your company as a whole. Three other metrics often get mentioned in the same breath. Each of them also gets by with a single question, but measures something different:

• The CSAT (Customer Satisfaction Score) measures how satisfied someone was with one specific experience, such as a purchase or a support call. It looks back.
• The CES (Customer Effort Score) measures how much effort it took to get something done. It fits right after a process like placing an order or filing a complaint.
• The eNPS (Employee Net Promoter Score) puts the recommendation question to your own workforce. It shows whether employees would recommend their company as a place to work.

Which metric fits which situation, and where the three overlap, is laid out in our short comparison of NPS, CSAT and CES.

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.

Improving the NPS: from the number to the action

Improving the NPS means closing the loop between asking and acting: collect the score together with the open answer, read those answers for reasons that repeat, fix the one that comes up most often and tell the people who raised it what changed. A score nobody acts on drifts, and the same respondents quietly stop answering the second time around.

Three steps turn the next wave into something usable. Tag every open answer with a single reason, so the pile turns into a count instead of a mood. Pick one reason per quarter, the one with the most detractor mentions, and give it an owner and a date. Then write back to the people who raised it, briefly and without marketing, because a closed loop is the one part of the method customers actually notice.

Passives at 7 and 8 are the group most programs skip, and they are usually the largest. A passive is not indifferent, a passive is unconvinced, and the distance from 8 to 9 is almost always shorter than the distance from 5 to 9. Read the open answers from the 7s and 8s as their own pile, separate from the detractors. The reasons there tend to be concrete and small, a shipping window, a confusing invoice, one missing feature, and small is exactly what a single quarter can fix.

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

• 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

Want to measure the NPS in your own survey?

empirio.ai offers NPS as its own question type and the analysis calculates the score for you. Browse customer feedback templates