What Net Promoter Score Isn't Telling You About Your Brand
The Number Every Boardroom Trusts
The Net Promoter Score has become the closest thing corporate brand management has to a universal language. Invented by Bain & Company consultant Fred Reichheld and popularized through a 2003 Harvard Business Review article, NPS is now used by roughly two-thirds of the Fortune 1000, according to Bain's own published research. It reduces brand health to one question, 'How likely are you to recommend us to a friend or colleague?', and one number: the share of promoters (scoring 9-10) minus the share of detractors (scoring 0-6). Bain's analysis reports that sustained value creators carry Net Promoter Scores roughly twice the average company's, a statistic that has made NPS a fixture in investor presentations and quarterly board decks. But a growing body of independent research argues the score conceals as much about brand health as it reveals.
What the Independent Evidence Actually Shows
The Revenue-Growth Claim Doesn't Replicate
Reichheld's original claim was that NPS was 'the one number you need to grow.' That claim has not held up under independent testing. A 2023 Marketing Science Institute working paper, authored by researchers including Jon Krosnick and Yphtach Lelkes and analyzing survey data tied to more than 30 U.S. companies across sectors using Bayesian modeling, concluded that NPS fails to predict revenue growth, and that none of the commonly proposed methodological fixes to the scoring formula changed that result. A separate peer-reviewed study in the Journal of Marketing, authored by Tim Keiningham and colleagues and honored with the Marketing Science Institute's H. Paul Root Award, replicated Reichheld's own methodology and found that traditional satisfaction measures correlated with firm growth at least as strongly as NPS did.
It Barely Explains Customer Spending
The gap between NPS and actual purchasing behavior is stark. Research covering more than 250,000 consumer ratings across 650 brands, reported by customer-experience trade outlet CMSWire, found that NPS explains only about 1% of the variance in share of wallet, and that changes in NPS explain a mere 0.4% of changes in customer spend. A separate peer-reviewed study in the Journal of the Academy of Marketing Science, which tracked seven U.S. sportswear brands over five years, reached a more nuanced conclusion: the standard transaction-based NPS had limited predictive value, and only a broader 'brand health' version of the metric, sampling non-customers as well as customers, showed a meaningful link to future sales growth.
It Misses the Customers Who Already Left
A methodological critique published in the International Journal of Market Research, by researchers Robert East, Jenni Romaniuk, and Wendy Lomax, identifies a structural blind spot: because NPS and comparable indices like the American Customer Satisfaction Index (ACSI) only survey current customers, they systematically exclude ex-customers and never-customers, the groups the authors find generate the most negative word-of-mouth about a brand. The ACSI itself, a federally referenced methodology maintained by a consortium including the University of Michigan's Ross School of Business and used to track satisfaction with U.S. federal agencies, the U.S. Postal Service, and roughly 200 corporations according to its own methodology report filed with the U.S. government's Office of Management and Budget, was cited in that same critique as sharing NPS's insensitivity to dissatisfaction.
Why the Score Is Easy to Distort
Beyond its weak link to growth, NPS is structurally vulnerable to manipulation. Customer-experience research aggregator SupportExp documents how frontline staff can nudge scores upward simply by asking for a '9 or 10' before a survey goes out, by implying employee evaluations depend on the result, or by administering the survey face-to-face, where customers are demonstrably less willing to criticize. Trade publication Research-Live, covering the long-running academic dispute over NPS, reported that Tim Keiningham's team, attempting to replicate Reichheld's founding analysis, called the underlying research 'nonsense, based on bad research,' while ACSI's own founder, Claes Fornell, dismissed Reichheld's rebuttals as lacking a basis in statistics and measurement. Bain itself has acknowledged the problem from the inside: in a Harvard Business Review piece co-authored by Reichheld and reproduced on Bain's site, the firm concedes that 'inexperienced practitioners' have abused the score by linking it directly to employee bonuses, and that many published NPS figures never disclose whether an independent third party ran the survey, turning the number into what Bain itself calls a 'vanity statistic.'
What a Single Score Cannot Tell You
- It cannot distinguish a passive customer who is quietly shopping competitors from one who is simply reserved, since both may score 7 or 8.
- It cannot capture the 80%-plus share of negative word-of-mouth that Bain's own research attributes to detractors, because the score aggregates that group into a single subtracted percentage.
- It cannot reach the people most likely to warn others away from a brand, since ex-customers and never-customers fall outside standard NPS and ACSI sampling frames.
- It cannot certify itself, since there is no standard requirement that NPS results be independently verified before appearing in investor or marketing materials.
Where This Leaves Brand Leaders
None of this means NPS is worthless as a directional signal; even its harshest academic critics concede it offers a fast, low-cost snapshot of customer sentiment. What the evidence supports is narrower: treat NPS as one input, tracked over time and segmented by customer type, rather than as a standalone predictor of growth or a proxy for full brand health. Pairing it with data that captures dissatisfied customers who have already left, and insisting on independently verified rather than self-reported figures, addresses the specific gaps that peer-reviewed research and Bain's own published guidance have both identified.