The Gist
- NPS measures what customers say, not what they do. Behavioral data — repeat purchase, retention and share of wallet — provides a clearer, less distorted signal of value than stated recommendation intent.
- The predictive promise doesn’t consistently hold up. Independent research repeatedly fails to show a reliable link between NPS movement and revenue growth, challenging its status as a growth engine.
- When metrics become targets, governance breaks. Optimizing for score movement can divert focus from operational drivers of ROI, creating activity without measurable financial impact.
Author's Note:
I am going to approach this article a bit differently than my others. I wanted to write an article on the topic of Net Promoter Score, or NPS, a metric designed to measure how likely customers are to recommend a company to others. As part of my research, I sent a Q&A to Isabelle Zdatny from Qualtrics XM Institute and Jason Lynn, SVP of product at Rokt, and they graciously agreed to share their perspectives.
But instead ofimmediately infusing their responses into what follows, I am going to split this article into two parts.
In part 1 today, I will lay out my own impression of NPS, grounded only in market data, observable behavior and common sense, the same way I approach every topic.
In the second part, I will bring their responses into the conversation and see where my thinking holds, where it breaks and where the truth likely lives somewhere in between.
The goal is not to win an argument. It is to arrive at an objective view of NPS, informed by how it is actually used in the real world, and to assess where its value may reside, if at all.
Ready? Let's do this thing.
Table of Contents
- Uncle Giuseppe's: No NPS. Plenty of Eggplant Parm.
- What NPS Promises vs. What It Measures
- Why I Do Not Use NPS
- The Customer Loyalty Illusion, Revisited
- NPS: A Lagging Indicator Wearing Predictive Clothing
- When Measurement Replaces Management
- The Cost of Indirect Answers
- The Hypothesis Worth Testing
- Why ROI Settles the Argument
- Where This Leaves NPS
Uncle Giuseppe's: No NPS. Plenty of Eggplant Parm.
Uncle Giuseppe's Marketplace near my home is a useful reminder of how consumers actually behave. The store is loud, crowded and chaotic in the best way. The prepared food is excellent…no seriously…excellent. The eggplant parm! The homemade bread! The cherry pepper sauce: spicy! You have no idea!
The ingredients feel curated, not stocked. The experience rewards attention. I do not love Uncle Giuseppe's as a brand. I return because the experience consistently delivers value. That distinction matters, because it exposes the gap between how people act and how we measure them.
Uncle Giuseppe's is not asking me how I feel about it. It does not survey me on my likelihood to recommend it. It does not track my sentiment. It earns my return visits by delivering something worth coming back for, consistently. My behavior answers the question that most customer metrics spend pages trying to infer.
This is where the gap begins to show. Much of modern customer measurement relies on what people say they might do rather than what they actually do, even though the data tells us those answers are often compromised from the start. Survey responses are routinely inflated by social desirability and courtesy bias, shaped by how the question is asked, who appears to be asking it and how quickly the respondent wants to move on. Add low response rates, straight-lining and the overrepresentation of only the happiest or angriest customers, and the result is a metric built on partial, skewed and frequently performative input.
The problem is not that these questions have no value. The problem is that they are treated as authoritative signals when observable behavior already provides a clearer, more honest, complete and less distorted answer.
Related Article: Your Customers Aren't Quiet — They've Given Up on Your Surveys
What NPS Promises vs. What It Measures
Net Promoter Score was introduced in 2003 by Fred Reichheld with a bold claim. A single question, asking how likely a customer is to recommend a company, could predict growth. Subtract detractors from promoters, track the score, and revenue would follow.
That promise is why NPS spread so quickly. It is simple, standardized and executive-friendly. Bain's own research continues to argue that, in select industries, NPS explains a meaningful share of organic growth variance. That claim deserves to be stated clearly, because it is often cited as proof that NPS works.
The problem is what happens when those claims are tested outside narrow contexts.
Independent, peer-reviewed research has repeatedly failed to validate NPS as a reliable predictor of future revenue growth or customer spending. Large-scale studies comparing NPS to traditional customer satisfaction measures found no evidence that NPS performs better. In some cases, it performs worse.
A Marketing Science Institute study in 2023 examined more than 30 U.S. companies using advanced modeling and reached a blunt conclusion. NPS fails to predict revenue growth.
Even more telling is what happens when NPS is evaluated against the metric that actually matters for ROI: share of wallet. Research covering more than 250,000 consumer ratings across 650 brands found that NPS explains roughly 1% of the variance in share of wallet. Changes in NPS explain only 0.4% of changes in spend. That leaves nearly all revenue behavior unexplained.
At that point, the issue is no longer methodological nuance. It is practical relevance.
Related Article: Wasn't NPS Supposed to Be All But Gone in 2025?
Why I Do Not Use NPS
I do not use NPS in my work and likely never will.
That decision is not ideological. It is operational. When customer behavior is observable, spend is measurable, and market conditions are visible, stated intent adds little clarity. I can predict how someone will feel about a cheesecake recipe. The only insight that matters is whether they eat it and come back for another slice. If the recipe performs, the work shifts to consistency in execution, not pursuit of a score. And (almost) most importantly, if I survey folks about whether they like my cheesecake or not, they likely will tell me what I want to hear, not what I need to hear.
NPS asks people how they think they will behave in the future. Behavior answers that question with far greater precision. Repeat purchase, retention, expansion and share of wallet do not require interpretation. They either happen or they do not.
This is where NPS starts to resemble a distraction. It creates the appearance of discipline without improving decisions tied to ROI. Teams debate survey design, scoring bands and benchmarks while the real signals sit untouched in transaction logs and retention curves.
The Customer Loyalty Illusion, Revisited
This is where NPS begins to overlap with the loyalty problem I have written about before, though the two should not be treated as identical. In The Loyalty Program Illusion: Why Points Don't Equal Preference, I argued that most customer loyalty programs confuse repeat behavior with genuine preference and then mistake that frequency for future value. The data was clear then, and it remains clear now.
Decades of research show that emotional loyalty to brands is rare. Most repeat behavior is driven by habit, convenience and availability, not attachment. Points-based loyalty programs routinely subsidize behavior that would have occurred anyway. More than half of loyalty memberships sit inactive. In travel and hospitality, most earned points are never redeemed. McKinsey found that the majority of transactional loyalty programs fail within two years. The pattern is consistent. Frequency is easy to measure. Preference is not.
NPS inherits this same structural flaw. It treats repeated behavior as proof of preference and stated advocacy as proof of future value. Both beliefs collapse under scrutiny. Here's the bottom line: people return to what works. They recommend what feels safe to suggest. Neither reliably predicts how much they will spend tomorrow, how long they will stay, or how easily they will leave.
This does not make NPS useless in every context. It does make its elevation to a strategic centerpiece difficult to justify, especially when it reinforces the same illusion loyalty programs have promoted for years: that saying something nice about a brand is evidence of commitment, rather than a byproduct of convenience.
Related Article: The New King of Customer Experience? McDonald's. Seriously.
NPS: A Lagging Indicator Wearing Predictive Clothing
At best, NPS is a lagging indicator. It reflects past experiences and recent interactions. Presented honestly, that can be useful as a directional signal. Presented as a predictor of growth, it becomes misleading.
Governance is where this matters most. Metrics shape behavior. When incentives, dashboards and executive reviews revolve around NPS, teams optimize for movement in a number rather than movement in revenue. The organization feels busy. Progress appears measurable. The connection to ROI weakens.
That is the risk I want to test in the second part of this article.
When Measurement Replaces Management
The persistence of NPS is not an accident. It survives because it is easy to administer, easy to report and easy to socialize upward. A single number travels well in board decks. It creates a sense of order in an otherwise messy reality.
That convenience comes at a cost.
When organizations adopt NPS as a primary indicator of customer health, they often stop asking harder questions. This is a textbook case of Goodhart's Law, the principle that once a measure becomes a target, it stops reflecting reality. Instead of using NPS as a signal, teams begin managing to the score itself. The work shifts from understanding customers to improving a number.
As that shift happens, fundamental questions go unanswered. Are customers consolidating spend or fragmenting it? Are repeat buyers increasing order frequency or merely returning out of habit? Are retention gains masking declining share of wallet? These questions require operational analysis rather than surveys, and they demand access to behavioral data and a willingness to confront uncomfortable answers.
The Path Toward Metric Displacement
The result is metric displacement. Effort flows toward reporting and improving NPS at the expense of usability, retention, customer effort and behavioral insight. In high-stakes environments, this pressure predictably invites gaming. Employees plead for top scores. Survey timing gets manipulated. Negative responses quietly disappear. The organization appears to be improving while the underlying experience remains unchanged.
This is where governance enters the conversation.
Metrics shape priorities. When leadership teams rally around NPS, they often redirect resources toward score management rather than experience improvement. Survey cadence increases. Score targets appear. Teams debate promoter thresholds. None of that guarantees better outcomes for customers or the business.
Research consistently shows why this approach fails. NPS does not resolve these questions. It does not explain why customers allocate spend the way they do. It does not meaningfully predict changes in revenue. Even Bain's more favorable claims acknowledge variability by industry and context. That variability matters because it undermines the idea that NPS is a universal proxy for growth.
Meanwhile, the most reliable signals of performance remain underused. Transaction data, retention cohorts and share-of-wallet analysis provide direct insight into ROI. They are less elegant, harder to summarize and more demanding to interpret. They also tell the truth.
Related Article: What Is Net Promoter Score (NPS)? A 2025 Roadmap
NPS vs. Behavioral Metrics: What Actually Predicts ROI?
A side-by-side look at how Net Promoter Score compares to direct behavioral indicators when the goal is measurable growth.
| Dimension | Net Promoter Score (NPS) | Behavioral & ROI Metrics |
|---|---|---|
| Primary Input | Stated likelihood to recommend | Observed customer actions |
| Data Source | Survey responses | Transaction logs, retention data, share of wallet |
| Bias Exposure | High (social desirability, courtesy bias, low response rates) | Low (behavior either occurs or it does not) |
| Time Orientation | Lagging, framed as predictive | Directly tied to current and longitudinal performance |
| Executive Appeal | Simple, standardized, board-friendly | Complex, requires analysis and interpretation |
| Governance Risk | Score optimization, gaming, metric displacement | Operational improvement tied to financial outcomes |
| Link to Revenue | Weak and inconsistent in independent studies | Directly measurable impact on retention and growth |
| ROI Utility | Directional at best | Decision-grade input for investment and allocation |
The Cost of Indirect Answers
There is a pattern here that extends beyond NPS. Marketing and CX functions have a long history of favoring indirect measures when direct ones are available. Loyalty programs reward frequency instead of preference. Engagement metrics reward interaction instead of impact. NPS rewards stated intent instead of observed behavior.
Each of these tools promises insight without confrontation. Each creates the illusion of progress through activity. Each risks diverting attention from what actually determines performance.
The research does not argue that customer sentiment is irrelevant. It argues that sentiment, when isolated from behavior, is insufficient. Studies comparing NPS to satisfaction scores, recommendation intent and other attitudinal measures repeatedly arrive at the same conclusion. These metrics explain little about future spend. They explain even less about changes in spend.
From an ROI perspective, that limitation is decisive.
If a metric cannot reliably inform allocation decisions, prioritize investment or predict revenue movement, its role should be constrained. That does not mean it disappears entirely. It means it stops masquerading as a growth engine.
The Hypothesis Worth Testing
This is the position I am putting at risk.
NPS may function as a conversation starter or a directional signal in limited contexts. From my perspective, its necessity collapses once real behavior is visible and measurable. In organizations serious about ROI, customer interaction, consistency of execution and market response outperform survey-based intent as decision inputs.
In other words, the work does not begin with a score. It begins with what customers do.
Why ROI Settles the Argument
When debates about NPS become circular, it is usually because the discussion drifts away from outcomes. Scores rise or fall. Benchmarks shift. Internal narratives form. None of that resolves the central question executives ultimately care about. Did the business grow because customers changed their behavior?
The evidence suggests NPS rarely answers that question. Across industries, the relationship between recommendation intent and spend remains weak. Share-of-wallet variance remains largely unexplained. Longitudinal studies fail to show consistent links between NPS movement and future revenue growth. These are not edge cases. They are repeated findings across large samples and extended timeframes.
From an ROI perspective, this is decisive. If a metric cannot meaningfully inform investment decisions or predict financial outcomes, its role should be limited. That does not make it fraudulent. It makes it secondary.
What reliably explains performance are observable actions. Customers return or they do not. They consolidate spend or they fragment it. They tolerate price changes or they defect. These behaviors capture preference without asking for it. They also remove interpretation from the equation.
This is why I prioritize direct interaction, behavioral data and market signals over survey instruments. When the cheesecake sells out and customers come back asking for it again, the feedback loop is complete. The work then becomes consistency, not measurement.
Where This Leaves NPS
Stripped of mythology, NPS appears less like a growth lever and more like a comfort mechanism. It gives organizations a sense of control in environments where customer behavior feels unpredictable. It offers simplicity where complexity is uncomfortable. It signals diligence even when impact remains uncertain.
That explains its persistence. It does not justify its elevation.
The industry data does not argue for eliminating NPS in all cases. It argues against treating it as decisive. It argues against confusing movement in a score with movement in revenue. It argues for restraint.
So, that's it. That's my thoughts on NPS. Am I right? Am I wrong? Is it somewhere in between where the truth will shine? There's only one way to find out: wait for part 2, which will be released soon on CMSWire.
For now, I'm off to Uncle Giuseppe's for some eggplant parm. I'm going to eat it. Enjoy it. And go back tomorrow, despite the fact they've never surveyed me.
Learn how you can join our contributor community.