The simple metric that’s taking over big business

Vanguard CEO Tim Buckley sees his firm’s Net Promoter Score every time he looks at the management dashboard on his screen. Bill Barton, CEO of California Closets, checks his company’s score first thing every morning and has built his highly successful business on it. Intuit product managers check their score every day and have been using it since it was invented 17 years ago. Some 40,000 employees use it at IBM, and the executive who oversees its use, Michelle Peluso, says, “It’s more than a metric. One could use the word ‘religion.’”

All this devotion for a particular measure of customer sentiment? It may seem bizarre, but the phenomenon is real and growing. At least two-thirds of the Fortune 1000 use the Net Promoter Score, including most or all of the financial service companies, airlines, telecom companies, retailers, and others. Quietly, steadily, without anyone much noticing, NPS has moved into the C-suites of most big companies and the owners’ offices of thousands of small ones—extending its reach deeply and broadly through the global economy. Skeptics and enemies have largely been vanquished. It is now used in every developed economy and many emerging ones. It’s pored over in all types of organizations, not just businesses; in Britain, the National Health Service uses it. As organizations everywhere obsess over the customer experience, NPS’s advance across industries and countries is, if anything, accelerating.

If you are somehow unfamiliar with the Net Promoter Score, you have nonetheless been touched by it—guaranteed. That’s because you have received an email or a phone call asking the one simple question at the foundation of the NPS empire: “On a scale of zero to 10, how likely is it that you would recommend [Company Name] to a friend or colleague?” You were probably asked a second question also: “Why did you give the answer you gave?”

And that, believe it or not, is pretty much it. Some companies add another question or two, and some alter the language slightly, but those two short questions are the essence of the system that inspires so much rampant enthusiasm. The results are so straightforward that they’re often expressed as emoticons—from green smiley faces to red angry faces.

Asking customers for a quick piece of feedback is just the beginning, of course. While the questions may be simple, the art of getting value from them is not. Talk to experienced users in a wide range of industries, and one message comes through loudly: NPS is a tool, very simple and very powerful. The simpler the tool, the more skill it requires; given a chisel, you can carve David or you can ruin a block of marble. Some organizations are using NPS skillfully and achieving significant competitive advantage. Others are misusing it and getting nothing or, worse, are deluding themselves and potentially their investors. The results a company gets from NPS depend on how well it’s used. And the wisdom on how to use it best is not always what one might expect.


To understand what makes NPS so compelling, it’s necessary to know where it came from. The metric was introduced to the world in a 2003 article in the Harvard Business Review by Fred Reichheld, a Bain consultant. Its roots are deep. For years Reichheld had been “focused on loyalty and treating people right,” he says. He found that “companies that focus on earning the loyalty of customers are taking over the world.” They make higher profits, pay employees more, keep employees longer, reward investors better—a virtuous circle. In 1996 he wrote The Loyalty Effect, a business bestseller that’s still in print, and kept working with clients who wanted their customers to be more loyal. “I realized it was a leadership thing and then realized that these things take a long time to play out,” he recalls. “So leaders needed a simple metric to find out, ‘Are we winning or losing with our customers?’ ”

A simple metric was unorthodox at the time, even blasphemous. The customer surveys of the day were massive, sometimes 100 questions long. Customers hated them, and managers needed an 80-slide deck to explain them to colleagues. Reichheld spent two years testing questions for his simple metric. “We picked likelihood to recommend because it best predicted future behavior of customers,” he says. “We picked it on a technical basis. Fortunately, it also makes people want to take action. If someone hears that someone won’t recommend them, they want to fix that.”

At least as important as the question is Reichheld’s way of stating the answer. A respondent who gives a 9 or 10 on the scale is a promoter; a 7 or 8 is a passive; a 6 or lower is a detractor. Subtract the percentage who are detractors from the percentage who are promoters, and that’s the Net Promoter Score, which can range from 100 to –100.

Crucially, it’s one number. You needn’t know how many people chose each number on the scale, or the weighted average; there are no charts or graphs. It’s one question with one answer. “You don’t have to explain in detail what the number means,” says Maurice FitzGerald, who used NPS when he ran the customer experience function at Hewlett Packard Enterprise’s software division and who has written a book on how to apply NPS. “It speaks to people so well. The simplicity of communication is absolutely why NPS is the standard.”

It’s not about chasing the number. It’s about understanding what our customers want and need from us.

DEBORAH CAMPBELL, vice president of consumer and marketing insights at Verizon

Having developed the world’s shortest customer survey, Reichheld then got even more radical. He told the world what it was. “We made it open-source, which was a revolutionary move,” he says. Consulting firms don’t give away their secrets. Gallup, J.D. Power, and others had built successful customer research practices based on proprietary methodologies. Not that Bain was averse to making money. But Rob Markey, who worked with Reichheld and now heads the firm’s NPS practice, recalls, “I tried to tell my partners that we should use NPS as widely as possible to learn faster. We wanted more companies to use it and share their results. Those that were really serious would benefit from our consolidating the experience of all these companies.” Besides, how could anybody keep the world’s shortest customer survey a secret?

That’s another reason NPS is everywhere. It’s not just simple, it’s also free. As a result, no one knows exactly how many companies use it; any organization can send out the survey entirely on its own. Which raises an obvious question: Is NPS a business or just a pro bono project at Bain?

Turns out NPS is a large and thriving industry. Any organization wanting to use NPS immediately faces a raft of questions: To whom do we send it? When? How many customers will respond, and how many constitute a valid sample? How often should they be polled? And the biggest question of all: Once we’ve learned our NPS, what should we do?

Several firms will be happy to help. Bain won’t disclose the dollar volume of its NPS practice but says the firm has handled nearly 750 NPS implementation projects over the past decade and has used NPS in thousands of other client projects. Last year SAP paid $8 billion for Qualtrics, a customer research and consulting firm, most of whose clients use NPS. A direct competitor is Medallia, which went public last year; it was worth $4.5 billion before the coronavirus clobbered the stock. Reichheld was a director of a smaller customer experience software firm, Satmetrix, which was bought by NICE Systems in 2017 for an undisclosed sum; Reichheld is no longer involved with it. Beneath these major players are dozens of smaller firms that advise on the use of NPS, plus dozens more that just handle the surveying.

The NPS industry even includes companies that once were bitter enemies of the metric. Gallup, seeing a direct threat to its consulting business, in 2006 called NPS “a popular idea that’s dead wrong.” That view has evolved. “We have to hand it to them,” says Gallup executive Jim Asplund. “People have been using it and finding value in it, so there must be something there.” Gallup even incorporates NPS into its work. Asplund says, “We look at our clients’ NPS data all the time and make use of it.”

At J.D. Power, “when NPS appeared, reaction was kind of mixed—it could be a competitor,” recalls Jay Meyers, the firm’s research chief. But five years ago Bain engaged Power to conduct NPS surveys, and even though that deal has expired, Power continues to measure NPS for clients. “Lots of them use it,” says Meyers. “We advise them on how to use it. We look at Net Promoter definitely as complementary now.”

As for the best way to use it, ask executives at major companies that use NPS and you’ll get a surprise. For all the work that went into devising this one simple metric, they say unanimously that the value doesn’t come from the number. Even at USAA, the financial services firm that continually posts some of the highest Net Promoter Scores in America, the score isn’t what counts. “The most important thing we’ve learned above all is what customers tell us, not the number,” says Julio Estevez-Breton, who oversees the research. Verizon, which gets some of the best scores in a low-ranking industry, has found the same. “It’s not about chasing the number,” says Deborah Campbell, who oversees Verizon’s use of NPS. “It’s about understanding what our customers want and need from us.”

This is a long-term metric. You can drive yourself crazy if you get hung up on the day-to-day or month-to-month. But if you use it for improving long-term, people will never go back.

DIEGO RODRIGUEZ, chief product and design officer at Intuit, which has used NPS since it was invented

That is, even though the score gets all the attention, it’s the second question—“Why did you give the score that you gave?”—that delivers the value. Even Bain agrees. “Anyone can hire a research firm and measure their score,” says Markey. “What’s more powerful, and required to get the real impact, is to make use of feedback from individual customers.” The score grabs everyone’s interest and is useful as a trend indicator, but the real importance of the first question is that it sets up the second question.

Parsing the answers to that second question is not easy for companies with thousands or millions of customers. But now technology can read responses and analyze them. “The real gem and actionable insights come from the verbatim transcripts,” says Marc Stein of Dell Technologies, “and we’ve learned the importance of bringing data science and machine learning to NPS.”

Sometimes what companies discover is undramatic but important. It could be a broad issue, for example, “clients who think in general that some of our tech documentation has to be overhauled and improved,” says IBM’s Peluso, who implemented NPS in previous jobs at Citigroup and Travelocity. Or it could be a narrow issue, such as “an account team that didn’t realize a client was having trouble onboarding their offering.” Constantly addressing those roadblocks, which the company might never have known about otherwise, builds a cumulative effect. “I’ve been blown away by how this data can be used for continuous improvement of customer experience,” says Boris Groysberg, a Harvard Business School professor who has studied NPS. “The ROI over five years is so much higher than you can get in three to six months.”

At other times the value of the “verbatims” is their power to center a manager’s thinking. At Intuit, chief product and design officer Diego Rodriguez says, “I can log in at any point and see the verbatims coming in. It’s very grounding and humbling.” That’s why California Closets CEO Barton starts his day reading the previous day’s verbatims: “I want the first voice in my ear every day to be the customer’s voice. It just grounds me.”

The customer isn’t always happy, and while listening is nice, it doesn’t help. Only action does. Responding to non-promoters is called “closing the loop” in the NPS world, and it’s vital. “I just had a bad experience with an airline,” says Peluso. “When I got an NPS survey, I wrote back about my experience, but then I thought, ‘Why am I writing this? I know you won’t respond.’ They didn’t, and I’m a really high-level ­flier.” Her view: “Don’t ask the question if you’re not going to do something about it. It’s like the airline—they’re making me angrier.”

Even in a big retail business where closing the loop seems daunting, it isn’t impossible. “From frontline employees to our executives, we call customers back,” says Carolyn Saunders, who oversees customer experience at Scotiabank, a Canadian bank that has expanded aggressively through Latin America using NPS. “We’ve conducted over a million callbacks since November 2016.”

Because NPS is so simple, many companies are applying it beyond its original use, often by measuring the NPS of competitors. It isn’t necessary to get the other guys’ customer lists; companies hire survey firms to call or email people at random, asking which companies in a given industry they do business with. In fact, many companies use third-party firms to learn their own score because the result is more objective. “The biggest bias is to send a survey request under your own name,” says Reichheld. “You can kick up your score 30 to 40 points by saying, ‘I’m with such-and-such company, and what did you think of us?’”

Net promoter all-stars

CURB APPEAL Workers at a Chick-Fil-A in Pembroke Pines, Fla., deliver orders to customers in their vehicles in March as part of new pandemic protocols. Chick-Fil-A consistently receives much higher Net Promoter Scores than most of its fast-food rivals.
Joe Raedle—Getty Images

To calculate their Net Promoter Scores, companies subtract the percentage of customers who rank them between 0 and 6 on a 10-point scale (detractors) from the percentage who give them a 9 or 10 (promoters). Here are the leaders in 10 industries. NPS true believer USAA tops two categories.

Industry Lowest Average Highest
Retail banking 16 34 78 (USAA)
Credit cards 11 38 57 (Discover)
Property and casualty insurance -18 30 70 (USAA)
Wealth management 18 34 53 (Vanguard)
Wired network operations -31 -7 23 (Verizon Fios)
Wireless service (postpaid) 3 33 58 (Consumer Cellular)
Department stores 29 36 42 (Nordstrom)
Grocery stores -1 29 69 (H-E-B)
Drugstores -20 -12 -4 (Walgreens)
Quick service and fast-casual restaurants 2 21 60 (Chick-fil-A)

Companies are even finding they can learn a lot by surveying people who aren’t customers at all. Asking the NPS questions of prospects who didn’t buy is highly revealing. Surveying the company’s own employees is also a strong trend; finding out if they would recommend their workplace to a friend or colleague, and why, can be eye-opening.

Like all simple tools, NPS can be used destructively. Longtime users agree on the No. 1 thing not to do: “If you want a bad NPS implementation, start by incentivizing the scores,” says Nacho Deschamps, a top executive at Scotiabank. When pay is at stake, employees can be marvelously creative. The resulting data will be worse than useless because managers won’t realize it’s fake. Many companies nonetheless link compensation for some employees to Net Promoter Scores.

Other damaging errors “can be done with the best of intentions and produce terrible results,” says ­Markey. For example, at an auto insurance company that was collecting feedback on its claims process, “some genius decided to ask only those customers whose claims had been paid, because it meant the process was over.” Of course it also meant the company didn’t hear the opinions of customers whose claims had not been paid. A common blunder occurs when a score is attributed only to the customer’s last point of contact. For example, a customer starts a transaction online, fails, contacts the call center, gets put on hold, struggles to describe the problem, eventually gets it fixed—and then receives an NPS survey about the phone rep. The web team, the source of the customer’s problem, hears nothing.

The chief remaining critic of NPS as a metric is Timothy Keiningham, a professor of marketing at St. John’s University in New York City. He argues that his statistical analyses show NPS is not “the single most reliable indicator of a company’s ability to grow” and is no better than several other measures of customer sentiment at predicting profitability. But to NPS believers, he’s focused on the wrong issue, since they all agree that the number alone isn’t what gives them value. (Professor ­Keiningham did not respond to interview requests.)

As NPS becomes ubiquitous, many companies, especially those below an industry’s top tier, have taken to ignoring the first law of NPS—it isn’t about the number—and issuing press releases bragging that they’ve achieved the highest NPS in their industry. Be skeptical of such claims. The numbers are rarely comparable because of differences in how, when, and from whom the results were obtained. They may delude the company’s own managers, “but worse,” says Markey, “is convincing investors that your deluded score is meaningful.”

Not even NPS’s most ardent fans think it’s the only customer experience tool they need. They all use many others as well. “It’s not the only way you get insights on what your customers need and believe to be true,” says Amy Cribbs, a customer experience executive at Vanguard, which has used NPS since it was created. For example, “it doesn’t always give you the insight on where that passion for your brand comes from. We are big believers in it, but there are other things you have to do.”

In addition, technology could someday make NPS obsolete. “You could envision a world where a rep is talking to a customer live on the phone, with the conversation being transcribed in real time and scored for customer sentiment and bucketed for issues,” says Stein of Dell Technologies, another charter member of the NPS users’ club. “Today the accuracy just isn’t there. But it will be.” Only a little further into the future, all customer experience data could be biometric and continuous, detected from a customer’s pulse, voice, and breathing. It’s plausible. And who would need NPS?

Until then, it’s hard to see the metric waning. Someone would have to invent a tool with a higher ratio of value to simplicity. In 17 years, no one has done it. Most important—fittingly for a system that originated as a way of studying loyalty—its users are extraordinarily steadfast. “This is a long-term metric. You can drive yourself crazy if you get hung up on the day-to-day or month-to-month,” says Rodriguez of Intuit, which has used NPS since it was invented. “But if you use it for improving long-term, people will never go back. We’re going to ask the same questions in the same way from now until the cows come home.” For now, the cows are nowhere to be seen.


How to cheat at NPS—and why you really shouldn’t

“Any metric can be gamed,” says Vanguard’s Amy Cribbs. For that reason in part, she says, “we do not bake NPS into individual reward systems.” But some companies do, incentivizing workers to scam their employers in ways that can lead to costly mistakes in the quest to earn customer loyalty. Among the techniques:


Seemingly everyone in the field of customer experience has heard about car salesmen who, at the end of the sale, tell customers, “You’re going to receive a survey, and if you don’t give me a 10, I won’t be able to feed my kids.” Maurice FitzGerald, who used NPS at HP Enterprise and has written a book on the measure, says this is “the rule rather than the exception” in U.S. auto retailing. Savvy customers sometimes beat the salesman to the punch by saying, near the end of the negotiation, “Take off another $500, and I’ll give you a 10.”


Responses can be influenced in many subtle ways. Managers whose bonuses are tied to Net Promoter Scores may send a survey in which 9 and 10 are green, 7 and 8 are yellow, and everything else is red. Or they may send a separate email saying only 9 and 10 count. Or, before the Net Promoter question, they may ask a question designed to get a favorable response; respondents want to be consistent and are thus more likely to answer the NPS question with a high number.

FitzGerald reports a related stratagem: Offer an incentive like entry in a prize drawing to everyone who responds; people won’t believe they can win if they give a low score.

Skewing the sample

Make sure the surveys go only to people who are happy or are friends. This is “easiest where a transaction is taking place by phone, and the script requires the employee to ask whether the customer is willing to take a survey,” says FitzGerald. “Simply omit to ask the unhappy ones.” Send online surveys to customers who have just ordered a new product, not to those whom you’re hounding to pay a late bill.

Ballot stuffing

“A telecom provider had retail stores where they were getting feedback via SMS,” says Bain’s Rob Markey. “The customer’s phone number was captured by the point-of-sale system, and the text message was sent at the end of the customer visit. Some employees figured out they could temporarily change the customer’s number to their own number, give themselves a 10, then change the number back.”

A version of this article appears in the June/July 2020 issue of Fortune.

Leave a Reply

Your email address will not be published. Required fields are marked *