Net Promoter Score: An S-Tier Behavioral Designer’s Guide

You have answered this survey. Everyone has. A screen slides up after a checkout, a support chat, or an app session and asks one question: “How likely are you to recommend us to a friend or colleague?” Below it sits a row of numbers from 0 to 10. You tap one, maybe without thinking, and move on. That tap, multiplied across thousands of customers, becomes a single figure that can decide an executive’s bonus, greenlight a product roadmap, or get a customer-success team reorganized by Friday.

The figure is called the Net Promoter Score, and for twenty years it has been the most worshipped metric in business. Boardrooms track it like a stock price. Consultants build practices around it. A whole industry of software exists to raise it by a point or two. The number feels objective, simple, and honest, which is exactly why it is so dangerous.

Here is the part the worship leaves out. The headline promise that launched NPS, that it was “the one number you need to grow,” mostly failed when independent researchers tried to reproduce it. The score throws away most of the information it collects. And the moment a company pays people to raise it, it quietly stops measuring anything real. Over the next several thousand words I will trace where NPS came from, how the machinery actually works, what its creator got genuinely right, where it breaks, what a recommendation is really measuring underneath, and how a behavioral designer uses this number without being fooled by it.

Speed Run Notes

• NPS asks one question, “how likely are you to recommend us, 0 to 10?”, then scores it as the percent of Promoters (9-10) minus the percent of Detractors (0-6). Fred Reichheld launched it in a 2003 HBR article.
• Its famous claim, that NPS is “the one number you need to grow,” largely failed to replicate. A 2007 Journal of Marketing study found it no better than plain old customer satisfaction at predicting growth.
• Collapsing an eleven-point scale into three buckets discards real information. A 6 and a 0 both count as Detractors, and two companies with an identical score can be nothing alike underneath.
• The number is trivially gamed. Tie a bonus to it and staff start begging customers for 10s. When a measure becomes a target, it stops being a good measure.
• In Octalysis terms, “would you recommend?” is a Core Drive 5 advocacy signal. You raise it by engineering the experience that produces advocacy, never by chasing the digit itself.
• Used honestly, as a doorway to the “why” and paired with Reichheld’s own accounting fix (Earned Growth), NPS earns its keep. Treated as objective truth, it flatters you while the business rots.

Table of Contents

Author Credibility: Yu-kai Chou

Yu-kai Chou created the Octalysis Framework after studying gamification since 2003 — years before the term entered mainstream vocabulary. As a Human-Systems Architect & Behavioral Designer, his framework has been applied by LEGO, Microsoft, Porsche, Coca-Cola, Salesforce, and MrBeast, impacting over 1.5 Billion Users.

Chou has taught the Octalysis methodology at Harvard, Stanford, Yale, Tesla, Google, BCG, and IDEO.

His work has been cited by Harvard, Stanford, MIT, Forbes, Wall Street Journal, Wired, US Department of Energy, NIST, NSF, NCBI, US Department of Education, ClinicalTrials.gov, and Google Scholar — with 3,700+ more academic publications. Explore his books here.

TEDxLausanne · Gamification to Improve Our World17 min

What Is the Net Promoter Score?

The Net Promoter Score is a customer loyalty metric built from a single question: “How likely is it that you would recommend this company, product, or service to a friend or colleague?” Respondents answer on a scale from 0 (not at all likely) to 10 (extremely likely). Those answers get sorted into three groups, and the score is the difference between the top group and the bottom group. That is the entire mechanic, and its brutal simplicity is the whole reason it conquered the corporate world.

The three groups are fixed. People who answer 9 or 10 are Promoters, assumed to be loyal enthusiasts who will keep buying and tell others. People who answer 7 or 8 are Passives, satisfied but unenthusiastic and easily poached by a competitor. People who answer anywhere from 0 to 6 are Detractors, unhappy customers who can damage your reputation through negative word of mouth. The score is calculated as the percentage of Promoters minus the percentage of Detractors. Passives count toward the total number of respondents but drop out of the arithmetic entirely.

Because it is a subtraction of two percentages, the result runs from -100 (every respondent a Detractor) to +100 (every respondent a Promoter). A score above 0 is usually framed as “good,” above 50 as “excellent,” and above 70 as “world-class,” though as we will see those labels carry far less meaning than they pretend to. “Net Promoter,” “Net Promoter Score,” and “NPS” are registered trademarks held jointly by Bain & Company, Fred Reichheld, and Satmetrix Systems, which is part of why the specific 0-to-10, three-bucket recipe has stayed so uniform across two decades and thousands of companies.

For a behavioral designer, this number matters in two directions at once. It is a reading on the people you design for, an attempt to quantify whether your work turned customers into advocates. And it is a number that will sit on your own dashboard, tempting you to treat a tidy figure as the truth about a messy reality. Understanding what NPS does and does not capture is the difference between a useful instrument and a comfortable lie.

The Origin: Reichheld and “The One Number You Need to Grow”

Fred Reichheld spent his career at Bain & Company studying customer loyalty, and he had already made his name with a 1996 book, The Loyalty Effect, arguing that retaining customers was one of the most underrated engines of profit. His frustration was practical. Companies ran long, expensive satisfaction surveys with dozens of questions, produced thick reports nobody read, and still could not connect any of it to whether the business actually grew. He wanted one question that predicted growth and that an executive would actually act on.

The claim landed in December 2003 with a Harvard Business Review article titled “The One Number You Need to Grow.” Reichheld described roughly two years of research in which his team tested a battery of survey questions and linked the answers to customers’ real subsequent behavior, their repurchases and their referrals, and then to the growth rates of the companies involved. Across most of the industries studied, one question tracked growth better than the rest: how likely a customer was to recommend the company to a friend. The percentage of genuine enthusiasts, he argued, was the leading indicator every leader had been missing.

The idea spread with astonishing speed, helped along by two follow-up books, The Ultimate Question in 2006 and The Ultimate Question 2.0 in 2011. The appeal is easy to understand. A CEO cannot do much with a 40-page satisfaction study, but a single number that moves up or down quarter over quarter is something a leadership team can rally around, benchmark against rivals, and cascade through an organization. Reichheld had packaged a real insight about loyalty into a format executives could hold in their heads. Adoption exploded across banking, software, retail, telecom, and eventually almost everywhere.

What is worth holding onto here is the original framing. NPS was pitched as a predictor of growth, validated by correlation with company revenue. That specific claim, not the general idea that happy customers are good, is what later research put under the microscope. The distinction turns out to matter enormously.

How NPS Actually Works: Promoters, Passives, Detractors

Walk through the arithmetic once and the whole thing becomes concrete. Suppose you survey 200 customers. 120 of them answer 9 or 10, 50 answer 7 or 8, and 30 answer somewhere from 0 to 6. Your Promoters are 120 out of 200, or 60 percent. Your Detractors are 30 out of 200, or 15 percent. Your NPS is 60 minus 15, which gives you a score of +45. The 50 Passives shaped the denominator, diluting both percentages, but never appeared in the final subtraction. That is the standard calculation everywhere it is used.

One property of the subtraction trips up almost everyone the first time. Because Passives are excluded from the arithmetic, a company can lift its score either by converting Detractors into Passives or by converting Passives into Promoters, and those two moves differ wildly in effort and in payoff. A score can also sit perfectly still while the base underneath it churns hard, as fresh Promoters replace departed ones and the net figure barely flickers. You are always looking at the gap between two moving percentages, and a calm gap can rest on top of violent movement.

The cut points are not arbitrary in intention, even if they are questionable in practice. Reichheld’s argument was that only a 9 or a 10 reflects the kind of intense enthusiasm that actually drives someone to repurchase and refer, and that a 6 or below signals real dissatisfaction with a meaningful chance of negative word of mouth. The 7s and 8s represent a lukewarm middle that is present but not committed. The design deliberately treats loyalty as demanding, refusing to hand out credit for mild approval.

In practice, teams collect NPS in two distinct modes, and confusing them causes endless trouble. Relationship NPS surveys customers periodically, often quarterly, about their overall relationship with the brand, giving a broad temperature reading. Transactional NPS fires right after a specific interaction, a purchase, a support call, an onboarding flow, to measure that moment. The two numbers answer different questions and should never be compared as if they were the same metric. A related variant, employee NPS or eNPS, asks staff how likely they are to recommend the company as a place to work, importing all of the method’s strengths and every one of its flaws into the HR department.

The mechanic’s genius and its curse are the same trait. Reducing a rich distribution of eleven possible answers to one signed number makes NPS effortless to report and impossible to argue with in a meeting. It also means that by the time the score reaches the boardroom, most of what customers actually told you has been thrown on the floor.

What Reichheld Got Right

It would be lazy to treat NPS as a scam, because Reichheld identified something real, and the metric’s staying power is not purely marketing. Give the man his due on several counts.

First, he tied a customer measure to the actual mechanics of organic growth. A company grows organically through two behaviors: existing customers buying again, and existing customers bringing in new ones. “Would you recommend?” gestures directly at the second engine, referral, in a way that a generic “how satisfied are you?” does not. Anchoring the question to advocacy rather than mere contentment was a genuine conceptual step, and it maps cleanly onto the customer decision journey’s loyalty loop, where advocates feed the top of everyone else’s funnel.

Second, he understood adoption as a design problem. A metric that is theoretically perfect and practically ignored is worth nothing. By compressing everything into one intuitive number on a common scale, Reichheld built something leaders would actually track and organizations could align around. There is a lesson here that applies far beyond surveys: the measure people will really use beats the measure that is merely more accurate on paper. Simplicity is a feature, and he engineered for it deliberately.

Third, and most underappreciated, the score was never supposed to travel alone. The serious version of the method pairs the rating with an open follow-up, “what is the primary reason for your score?”, and treats that verbatim comment as the real payload. The number tells you the temperature; the comments tell you why. Practitioners who use NPS well spend most of their energy reading and acting on the “why,” not admiring the digit. The closed-loop practice Reichheld pushed, calling Detractors back to understand and fix their problem, is where a lot of the method’s genuine operational value lives.

Strip away the overreach and a solid core remains: advocacy is a meaningful signal, simplicity drives real-world use, and the follow-up question is a doorway to insight. The trouble started when the industry fell in love with the digit and forgot the doorway.

Where NPS Falls Apart

The gap between what NPS claims and what it delivers is wide enough that careful researchers have been documenting it for nearly twenty years. Three problems matter most, because each one changes how a designer should treat the number.

The Growth-Prediction Claim Does Not Hold Up

The load-bearing promise of NPS was predictive superiority: that it forecasts growth better than the alternatives. In 2007, a team led by Timothy Keiningham published “A Longitudinal Examination of Net Promoter and Firm Revenue Growth” in the Journal of Marketing, using years of data from the Norwegian Customer Satisfaction Barometer across many firms and thousands of interviews. When they reran the analysis on the very industries Reichheld held up as showcases, they could not reproduce the “clear superiority” he had claimed. NPS correlated with growth about as well as ordinary customer satisfaction did, sometimes better and sometimes worse, but with no special magic. The paper won the Marketing Science Institute’s H. Paul Root Award for its contribution to marketing practice, so this was not a fringe potshot. The reasonable read is that the specific “one number you need” headline was the marketing, and the durable finding underneath was the far more modest “advocacy is one decent loyalty signal among several.”

It Throws Away the Scale

Statisticians have a straightforward complaint: NPS takes an eleven-point scale and crushes it into three categories, then discards one of them from the math. That is a lot of thrown-away information. A customer who scored you a 6 and a customer who scored you a 0 are both filed as Detractors and treated as identical, even though one is mildly disappointed and the other is furious. Two companies can post the exact same +30 while one has a wall of moderate 7s and 8s and the other has a violently split base of ecstatic 10s and hostile 2s. Those are completely different businesses with completely different risks, and the headline number hides the difference. Because the metric ignores Passives and lumps a wide range together, it is also statistically noisier than a simple average of the raw scores would be, which means quarter-to-quarter wobbles often reflect sampling noise rather than any real change in how customers feel.

Defenders answer that the crude bucketing is a feature, forcing a demanding definition of loyalty that a plain average would blur into mush. There is something to that. A designer should simply know the price of the simplification: you are trading statistical resolution for boardroom legibility, and the tidy result should never be mistaken for the richer reality it compressed.

Culture and Gaming Corrupt the Number

Two more forces quietly poison cross-comparisons. The first is culture. Willingness to hand out a 9 or a 10 varies dramatically by country. Customers in the United States give top marks freely, while customers in many parts of Europe and Asia reserve the highest ratings for something close to perfection, so a “7 out of 10” from a reserved respondent may signal the same real feeling as a “9” from an effusive one. Comparing the NPS of a company’s American and Japanese divisions, or benchmarking a global brand against a regional one, can therefore be close to meaningless. The second force is the human one: the number is easy to game, and the next section is devoted to how that plays out. Once you know a metric can be inflated by coaching customers rather than serving them, every published benchmark deserves a raised eyebrow.

What’s Really Happening When Someone Recommends You

To use NPS wisely, you have to understand why a stated willingness to recommend carries any signal in the first place, and where that signal leaks.

Start with what a recommendation actually is. Telling a friend to try something puts your own reputation on the line. If the product disappoints them, some of that disappointment sticks to you for having vouched for it. That makes advocacy a costly signal: because it risks something, a genuine intention to recommend reflects real confidence in a way that a bland “I’m satisfied” does not. This is the psychological reason Reichheld’s question had more predictive juice than the standard satisfaction item. You are asking people whether they would spend social capital on you, which is a higher bar than asking whether they were pleased.

This is also why advocacy sits at the very top of everyone else’s acquisition funnel. A recommendation from a friend is the highest-trust input a prospective customer can receive, because it arrives pre-vouched by someone they already believe. Classic models of how strangers turn into buyers, like the AIDA model, all hinge on that first spark of attention, and a personal recommendation is the most efficient way to strike it. NPS is trying, however imperfectly, to gauge the supply of people willing to strike that spark on your behalf.

Now the leaks. The most important one is the gap between what people say and what they do. NPS measures intention to recommend, and stated intentions are a famously imperfect guide to behavior. Plenty of people who tap a sincere 9 never actually refer anyone, and some quiet 7s become your most valuable evangelists. A survey captures a feeling in a moment; it does not capture the referral that may or may not happen three months later. Reichheld’s later work, discussed below, is an admission that this gap is real and needs a behavioral backstop.

There is also a memory problem sitting underneath the score. When a customer answers, they do not compute the average quality of every interaction they have ever had with you. They answer from what they can recall, and human memory of an experience is dominated by its most intense moment and its ending rather than its overall arc. A single painful support call near survey time can sink a score built over years of quiet satisfaction, and a delightful final touch can lift it past what the full relationship deserves. The number you collect is a reading on remembered experience, which is a specific, editable, and sometimes misleading thing. For a designer, that is not only a caveat but also a lever: the moments you choose to make memorable move the score more than the average quality ever will.

NPS vs CSAT, CES, and the ACSI

NPS is one option in a small family of customer-feedback metrics, and choosing well means knowing what each one is actually for.

CSAT (Customer Satisfaction Score) asks, straightforwardly, how satisfied you were, usually with a specific interaction, on a scale of something like 1 to 5. It is the oldest and most intuitive of the group and it excels at transactional, in-the-moment feedback: was this support call good, was this delivery on time. Its weakness is a ceiling effect, since most people click “satisfied” out of politeness, which limits how much it discriminates.

CES (Customer Effort Score) asks how much effort you personally had to expend to get your issue resolved. It came out of a 2010 Harvard Business Review article by Matthew Dixon and colleagues, “Stop Trying to Delight Your Customers,” which argued that reducing customer effort predicts loyalty better than trying to exceed expectations. For support and service journeys, effort is often a sharper predictor of churn than either satisfaction or recommendation intent, which connects directly to the friction-reduction logic behind the EAST behavioral framework: make the good path easy and people stay on it.

The ACSI (American Customer Satisfaction Index) is a different animal, an academic, multi-question index maintained at the University of Michigan and grounded in a published model of satisfaction, expectations, and perceived value. It is more rigorous and more predictive in the research literature than a single-item measure, but it is also heavier and slower, which is exactly why a lightweight metric like NPS won the popularity contest despite being less careful. Reichheld’s real edge over the ACSI was never accuracy. It was that a busy executive would actually use his number, a point that echoes the “will people actually adopt this?” test at the heart of good design thinking.

The practical takeaway is unglamorous. Use CES for service and support friction, CSAT for transactional satisfaction, a rigorous index like the ACSI when you need research-grade accuracy, and NPS when you want a simple, widely benchmarked relationship signal that leadership will actually watch. No single number is the “right” one, and the marketing that pretended otherwise is precisely what got the industry into trouble.

NPS in the Real World

The abstract debate settles down once you watch NPS operate in specific settings, each with its own trap and its own payoff.

SaaS and Subscription Businesses

Software companies live and die on renewals, so relationship NPS run quarterly and cross-referenced against churn cohorts is a common early-warning system. The value is real when a falling score in a segment prompts an investigation before those accounts cancel. The trap is survivorship: your unhappiest customers have often already left and stopped answering surveys, so a subscription NPS can drift cheerfully upward precisely because the Detractors keep churning out of the sample. When your base is shrinking, a rising score can be a symptom of churn rather than a sign of health. This shows up constantly in modern subscription tools, from streaming services to project-management apps, where leadership celebrates a climbing quarterly score while the product’s heaviest early adopters quietly cancel and drop out of the sampling frame. The number is measuring the contentment of the survivors.

Retail and Transactional Journeys

In retail, e-commerce, and support, transactional NPS captures a specific moment: the checkout, the delivery, the returned item, the resolved ticket. Here the score is most useful as a fast operational signal tied to a named touchpoint, where a dip points at a broken process you can go fix. It becomes least useful when a company averages a thousand of these transactional pulses into one corporate headline number and then presents it as the health of the entire brand relationship, blending two things that measure different realities.

Employees: eNPS

Asking staff how likely they are to recommend the company as a workplace has become a standard pulse check. It is quick and trend-able, and a sharp drop is worth taking seriously. It also inherits every weakness of the parent metric, plus a new one: employees know their answers may not be as anonymous as promised, which can inflate the score through simple self-protection. Treat eNPS as a smoke alarm, not a diagnosis.

The Closed Loop

The single most valuable operational use of NPS has nothing to do with the headline figure. It is the closed loop: routing every Detractor to a human who calls them back quickly, understands what went wrong, and fixes it. Done well, this turns a measurement exercise into a recovery engine, and the recovered customer often becomes more loyal than one who never had a problem. The score is just the trigger. The follow-up conversation is the product.

The Elephant in the Room

Here is the failure mode that quietly destroys the metric from the inside, and almost every company that uses NPS at scale eventually walks into it. The moment you attach a bonus, a performance review, or a job to the score, you stop measuring customer loyalty and start measuring how hard your staff will push to inflate a number.

You have felt this as a customer. A car dealership hands you the keys and says, “you’ll get a survey, and anything less than a 10 counts as a fail for me, so if there’s any reason you can’t give a perfect score, please tell me now.” A support agent signs off with “if you were happy today, a 10 really helps my team.” Those are not requests for honest feedback. They are coaching sessions designed to move the metric without moving the underlying reality by one inch. The score climbs, the dashboard glows, and the actual experience is exactly what it always was, or worse, because energy that could have gone into serving customers went into managing their ratings.

This is a textbook case of what is often called Goodhart’s Law, sharpened by the anthropologist Marilyn Strathern into a single sentence: when a measure becomes a target, it ceases to be a good measure. A metric works as a thermometer, quietly reading the temperature. The instant you turn it into a thermostat that people are graded on, they will grab the dial. NPS is unusually easy to grab, because a single motivated employee can lean on a single customer at the exact moment of the survey. The number that was supposed to keep companies honest becomes an elaborate machine for lying to themselves, and the lie is comfortable because everyone’s incentives point the same way. This is confirmation bias wearing a lanyard: leadership wants the number to be up, the frontline is paid to make it up, and no one in the chain is rewarded for noticing that it means less every quarter. It is also the dark twin of good reward design, where reinforcing the proxy instead of the real behavior reliably produces the proxy and nothing else.

The escape is not a better survey. It is a structural rule: keep the measure and the incentive apart. A company that genuinely wants to know its NPS must make sure that no individual’s pay or standing depends on it, so that no one on the frontline has a reason to coach the answer. That is hard, unglamorous discipline, and it is the only thing that keeps the number worth collecting.

How to Apply NPS with the Octalysis Framework

The Octalysis Framework organizes human motivation into eight Core Drives. NPS is not one of them; it is a measurement that sits downstream of them. Its real value through an Octalysis lens is diagnostic, because the framework tells you which drives you have to engage to turn a customer into the kind of advocate the survey is trying to detect. Once you see that, the futility of chasing the number directly becomes obvious.

The recommendation itself is an act of Core Drive 5 (CD5): Social Influence & Relatedness. When someone tells a friend to try your product, they are spending social capital, signaling identity, and drawing on the web of relationships that CD5 governs. A Promoter is, in Octalysis terms, a customer for whom your product became something worth associating themselves with in front of other people. If you want more Promoters, you have to build experiences worth being seen endorsing, which is a CD5 design problem long before it is a survey problem.

Underneath the strongest advocacy sits Core Drive 1 (CD1): Epic Meaning & Calling. The customers who recommend you most fiercely usually believe your product stands for something they want to be part of. This is the drive at the heart of Reichheld’s own later framing, that durable growth comes from genuinely serving customers rather than extracting from them. Meaning turns a satisfied buyer into a missionary, and no amount of survey engineering manufactures that.

Core Drive 2 (CD2): Development & Accomplishment supplies another route to advocacy. People enthusiastically recommend tools that made them visibly better at something, more productive, more skilled, more capable. When your product is the reason a customer accomplished something they are proud of, the recommendation follows almost automatically, because sharing it also shares their own success story.

Core Drive 4 (CD4): Ownership & Possession deepens the bond. As customers invest time, data, and customization into a product, they develop a sense of psychological ownership over it, and people defend and promote what feels like theirs. And on the opposite pole, Core Drive 8 (CD8): Loss & Avoidance is what animates your Detractors: a bad experience creates a felt loss that people are driven to warn others about, which is why negative word of mouth travels faster and lands harder than praise. The Detractor is CD8 firing outward.

Naming the drives reframes the entire exercise. If you treat NPS as something to be raised through Core Drive 3 (CD3): Empowerment of Creativity & Feedback, by cleverly tuning survey wording, timing, and coaxing, you are optimizing the thermometer instead of the weather. The advocacy the score is trying to catch is produced by CD5, CD1, CD2, and CD4, and undermined by CD8. Engineer those drives and the number rises as a genuine side effect. Chase the number itself, by begging customers for 10s, and you activate the most manipulative, trust-eroding kind of Black Hat design, which corrodes the very advocacy you were trying to measure. The score is an output. Design the inputs.

How to Use NPS Without Fooling Yourself

Because the number is so easy to worship and so easy to game, every practice below is about keeping it honest and putting it to work rather than putting it on a pedestal.

1. Never tie individual pay or reviews to the score. This is the first and most important rule, and it is the one companies break most often. The instant someone’s bonus depends on NPS, the number starts measuring pressure instead of loyalty. Keep the measure and the incentive in separate rooms.
2. Treat the score as a doorway, not a destination. The rating tells you almost nothing on its own. The open-ended “why did you give that score?” is the actual asset. Spend your energy reading, coding, and acting on the verbatim comments, and treat the digit as merely the flag that tells you where to look.
3. Pair it with an accounting-based twin. Reichheld himself conceded the survey’s weakness. In his 2021 book Winning on Purpose and the companion “Net Promoter 3.0” article, he introduced the Earned Growth Rate, a measure built from your actual books rather than a survey: the revenue growth that comes from returning customers plus the new customers they refer. Because it is drawn from behavior rather than stated intention, it closes the say-do gap that always haunted the original score. If you take NPS seriously, track Earned Growth alongside it.
4. Read the distribution, not just the headline. A +40 built on placid 7s and 8s is a different business from a +40 built on a war between 10s and 2s. Watch your Detractor percentage and your Promoter percentage as separate lines, and look at the full spread of scores, so a single blended number never hides a polarized base.
5. Compare yourself to yourself. Cross-cultural and cross-industry benchmarks are mostly noise, thanks to the response-style differences described earlier. Your own trend line over time, using a consistent method and audience, is the only comparison you can trust. Ignore the vanity of stacking your score against a rival’s.
6. Close the loop within days, not weeks. The operational return on NPS comes from contacting Detractors fast, understanding the failure, and fixing it. A recovered customer is often more loyal than one who never complained. If you are not acting on individual responses quickly, you are collecting a number for decoration.
7. Design for the drives, not the digit. The durable way to raise advocacy is to engineer the Core Drives that produce it: give customers something worth recommending (CD5), a mission worth joining (CD1), and accomplishments worth sharing (CD2). The score is the shadow. Move the object and the shadow moves with it.

None of these steps ask you to run a cleverer survey. They ask you to hold the number loosely, read what is underneath it, back it with behavior, and never let it become the thing your people are secretly optimizing. That is the whole discipline of using a metric without being used by it.

NPS Was the Beginning, Not the End

Fred Reichheld set out to find one honest question that predicted growth, and he found something genuinely useful: advocacy is a real signal, simplicity drives adoption, and the “why” behind a score is where the value hides. Then the idea got oversold into “the one number you need,” the independent research declined to confirm the superlative, and an entire industry spent two decades chasing a digit that its own frontline was quietly inflating. Both stories are true at once, which is exactly why NPS is such a useful thing for a designer to think hard about.

The lesson is not to throw the metric out. It is to demote it from oracle to instrument. Read it as a thermometer that reports the temperature, and keep it off the list of targets your people are graded on. Chase the verbatim comments, back the score with an accounting measure of real behavior, keep it far away from anyone’s bonus, and remember that the loyalty it is trying to detect is manufactured upstream in the Core Drives rather than in the survey design. Do that, and NPS becomes one honest reading among several. Worship it, and it becomes a mirror that shows you whatever you were hoping to see. A number is only as good as the humility of the people reading it, and that is a truth that reaches well past this one survey. To keep building the judgment that separates a real signal from a flattering one, keep going with the guides below.

Frequently Asked Questions

What is a Net Promoter Score in simple terms?

The Net Promoter Score is a customer loyalty metric based on one question: how likely you are, on a scale from 0 to 10, to recommend a company to a friend or colleague. Answers of 9 or 10 are Promoters, 7 or 8 are Passives, and 0 through 6 are Detractors. The score is the percentage of Promoters minus the percentage of Detractors, producing a figure between -100 and +100. It is meant to be a quick, single-number reading of how many of your customers are enthusiastic enough to advocate for you.

Who invented the Net Promoter Score?

NPS was created by Fred Reichheld, a loyalty expert at the consulting firm Bain & Company, working with Satmetrix and Bain. He introduced it in a December 2003 Harvard Business Review article titled “The One Number You Need to Grow,” then expanded it in his 2006 book The Ultimate Question and its 2011 sequel. “Net Promoter,” “Net Promoter Score,” and “NPS” are registered trademarks held by Bain & Company, Reichheld, and Satmetrix.

How is NPS calculated?

You sort every survey response into three groups: Promoters (scored 9-10), Passives (7-8), and Detractors (0-6). You then calculate what percentage of all respondents are Promoters and what percentage are Detractors, and subtract the second from the first. Passives are counted in the total but excluded from the subtraction. For example, 60 percent Promoters and 15 percent Detractors gives an NPS of +45. The possible range runs from -100 to +100.

What is a good NPS score?

By common convention, any score above 0 is considered acceptable, above 50 is considered excellent, and above 70 is called world-class. These labels should be treated with heavy skepticism, because “good” varies enormously by industry and, crucially, by culture. Customers in some countries hand out 9s and 10s freely while others reserve top marks for near-perfection, so the same underlying feeling produces very different scores. The only benchmark you can really trust is your own score measured consistently over time.

Is NPS actually a good predictor of business growth?

Not as good as its original marketing claimed. Reichheld pitched NPS as the single best predictor of growth, but a 2007 study in the Journal of Marketing by Timothy Keiningham and colleagues failed to replicate that superiority, finding that NPS predicted growth about as well as conventional customer satisfaction, no better and no worse. Advocacy is a legitimate loyalty signal, but the idea that this one number reliably outperforms all others did not survive independent testing.

What is the difference between NPS, CSAT, and CES?

They measure different things. CSAT (Customer Satisfaction Score) asks how satisfied you were, usually with a specific interaction, and is best for transactional feedback. CES (Customer Effort Score) asks how much effort it took to get your issue resolved and often predicts churn in service journeys better than the others. NPS asks how likely you are to recommend the company and is aimed at overall loyalty and advocacy. Each fits a different job, and no single one is universally superior.

What are relationship NPS, transactional NPS, and eNPS?

Relationship NPS surveys customers periodically about their overall relationship with a brand, giving a broad loyalty temperature. Transactional NPS is collected right after a specific event, such as a purchase or support call, to measure that moment. Employee NPS, or eNPS, asks staff how likely they are to recommend the company as a place to work. They share the same 0-to-10 mechanic but answer different questions and should never be directly compared to one another.

Can NPS be gamed, and why do people say it is flawed?

Yes, easily. When bonuses or performance reviews are tied to the score, staff start coaching customers toward 10s, and the number rises without the experience improving at all. This is Goodhart’s Law in action: when a measure becomes a target, it stops being a good measure. Critics also point out that collapsing an eleven-point scale into three buckets throws away information, that Passives are ignored in the math, and that two companies with identical scores can have completely different customer bases. The fix is to keep the measure separate from any incentive.

What is Earned Growth, or NPS 3.0?

Earned Growth is a measure Reichheld introduced in his 2021 book Winning on Purpose and the “Net Promoter 3.0” article to fix the survey’s biggest weakness. Instead of asking customers about their intentions, it is calculated from a company’s actual financial records: the Earned Growth Rate captures revenue growth from returning customers plus the new customers they refer. Because it is built on real behavior rather than stated intent, it is meant to serve as an accounting-based twin to the traditional survey score.

How does NPS relate to gamification and the Octalysis Framework?

In the Octalysis Framework, recommending a product is an act of Core Drive 5 (Social Influence & Relatedness), and the strongest advocacy is fueled by Core Drive 1 (Epic Meaning & Calling), Core Drive 2 (Development & Accomplishment), and Core Drive 4 (Ownership & Possession), while Detractors are driven by Core Drive 8 (Loss & Avoidance). The practical implication is that NPS is an output metric: you raise it by designing the Core Drives that produce advocacy, not by chasing the number directly, which only tempts you into manipulative tactics that erode trust.

References

• Reichheld, F. F. (2003). The one number you need to grow. Harvard Business Review, 81(12), 46–54.
• Reichheld, F. F. (1996). The Loyalty Effect. Harvard Business School Press.
• Reichheld, F. F. (2006). The Ultimate Question. Harvard Business School Press.
• Reichheld, F. F., & Markey, R. (2011). The Ultimate Question 2.0. Harvard Business Review Press.
• Reichheld, F., Darnell, D., & Burns, M. (2021). Winning on Purpose: The Unbeatable Strategy of Loving Customers. Harvard Business Review Press.
• Reichheld, F., Darnell, D., & Burns, M. (2021). Net Promoter 3.0. Harvard Business Review, 99(6).
• Keiningham, T. L., Cooil, B., Andreassen, T. W., & Aksoy, L. (2007). A longitudinal examination of net promoter and firm revenue growth. Journal of Marketing, 71(3), 39–51.
• Grisaffe, D. B. (2007). Questions about the ultimate question: Conceptual considerations in evaluating Reichheld’s Net Promoter Score. Journal of Consumer Satisfaction, Dissatisfaction and Complaining Behavior, 20, 36–53.
• Dixon, M., Freeman, K., & Toman, N. (2010). Stop trying to delight your customers. Harvard Business Review, 88(7/8), 116–122.
• Fornell, C., Johnson, M. D., Anderson, E. W., Cha, J., & Bryant, B. E. (1996). The American Customer Satisfaction Index: Nature, purpose, and findings. Journal of Marketing, 60(4), 7–18.
• de Haan, E., Verhoef, P. C., & Wiesel, T. (2015). The predictive ability of different customer feedback metrics for retention. International Journal of Research in Marketing, 32(2), 195–206.
• Strathern, M. (1997). “Improving ratings”: Audit in the British university system. European Review, 5(3), 305–321.

Related Reading

• The Octalysis Framework: The Complete Guide to Gamification’s 8 Core Drives
• The Consumer Decision Journey: Inside the Loyalty Loop
• The AIDA Model: Attention, Interest, Desire, Action
• Design Thinking: The Five-Stage Human-Centered Process
• Confirmation Bias: Why We See the Numbers We Want to See
• Psychological Ownership: Why “Mine” Changes Everything

WOULD YOU LIKE YU-KAI CHOU TO WORK WITH YOUR ORGANIZATION?