Net Promoter Score: Magic Metric or Fool’s Gold?

The Case Against NPS: Where the Criticism Has Force

The objections to NPS are not minor quibbles. Several of them cut to the heart of what the metric is actually measuring, and honest research practitioners have to take them seriously.

The question itself sets up an unrealistic scenario. Most people do not go around recommending brands to their friends and colleagues as a matter of routine. This is especially true in low-engagement categories. Pensions, insurance, utilities, healthcare providers, broadband suppliers: these are sectors where consumers hold strong opinions about their experience, but where organic word-of-mouth recommendation is not a natural behaviour. When someone is asked how likely they are to recommend their pension provider, they are not really being asked about a behaviour they recognise. The cognitive leap between the literal question and the underlying sentiment is larger in some sectors than others.

A personal example illustrates the point. After a recent GP appointment, the practice sent an NPS request. It felt so disconnected from any realistic behaviour that ignoring it seemed the only sensible response. And it is safe to assume that reaction is common, which creates a response bias problem: the people who bother to answer are likely to be those with stronger feelings, either positive or negative, which inflates the apparent polarisation of the score.

The Detractor threshold is arguably set too high. Classifying anyone who scores 6 out of 10 as a Detractor is a difficult position to defend. A 6 out of 10 is, in any reasonable reading, a broadly acceptable rating. Someone who gives a brand a 6 is not, in most cases, an active critic. They are somewhere in the middle ground, which is exactly where most customers sit. The effect of this classification is to systematically overstate the proportion of Detractors and produce NPS scores that are structurally lower than a more balanced threshold would generate.

NPS is also susceptible to gaming, and this is not a marginal problem. Because frontline staff are sometimes evaluated against NPS results, there are widespread accounts of customers being coached, pressured or subtly encouraged to give higher scores. A Fortune magazine investigation of NPS noted that car salespeople routinely tell customers that anything below a 10 hurts their pay, and that customers sometimes use the prospect of a high score as a bargaining chip. Reichheld himself has been unambiguous on this: linking NPS to employee compensation is, in his view, the single most damaging thing a business can do with the metric. It inflates scores, destroys the honest feedback loop the system was designed to create, and typically collapses NPS programmes within one to two years.

There is also a structural silence in NPS data. The customers least likely to respond to any customer feedback survey are those who had an unremarkable experience: neither frustrated nor delighted, just fine. This silent majority would tend to score in the Passive range, and their non-response means the score is shaped disproportionately by outliers. That is not a unique flaw in NPS; it affects most survey methodologies. But it is worth being explicit about.

Finally, as a single-question measure, NPS cannot tell you anything about what is driving the score or where to act. Knowing that your NPS has dropped five points is useful only if you already have other data that explains why. On its own, the metric is a signal without context.

The Academic Challenge: Does NPS Actually Predict Growth?

Reichheld’s foundational claim was not just that NPS was easy to use, but that it predicted company growth better than other measures. This claim has been tested, and the evidence is genuinely mixed.

The most frequently cited supporting evidence comes from a 2005 study by researchers at the London School of Economics, published in Brand Strategy under the title ‘Advocacy Drives Growth’ (Marsden, Samson and Upton, 2005). The study found that a seven-point increase in NPS correlated on average with approximately one per cent growth in revenue. It is worth noting, however, that critics have since observed that the NPS data and the revenue growth data in that study largely overlapped in time, meaning the correlation was contemporary rather than strictly predictive. Bain and Company have also argued that in most industries, NPS explains between 20 and 60 per cent of the variation in organic growth rates between competitors, and that NPS leaders typically grow at more than twice the rate of their peers.

However, subsequent academic research has reached more sceptical conclusions. A study by Dawes (2024), examining longitudinal data across airlines, supermarkets and insurance companies over periods of five to eleven years, concluded that NPS is not a reliable indicator of future revenue growth. Scholars have also challenged whether the likelihood-to-recommend question measures anything meaningfully different from standard satisfaction or repurchase-intent measures, and whether its superior predictive power, claimed in the original Harvard Business Review article, holds up under rigorous independent testing.

An independent replication study by MeasuringU, which revisited the original data Reichheld used in his 2006 book, found that NPS could explain approximately 38 per cent of the variability in company growth across seven industries when future rather than historical revenue growth was used as the dependent variable. That is considerably less than the figure Reichheld originally reported, but still a meaningful amount relative to other single-question behavioural measures.

The honest answer, then, is that the link between NPS and growth is real but not as clean as its most enthusiastic proponents claim. NPS is a reasonable proxy for sentiment and advocacy tendency. It is not the infallible predictor of commercial performance that some of its supporters have suggested.

The Case For NPS: Where It Earns Its Place

Despite the criticisms, NPS has survived and spread for reasons that have genuine force, and it would be intellectually dishonest to dismiss them.

Simplicity is not a superficial virtue. One of the consistent findings across research is that the metrics that get acted on tend to be the ones that non-research audiences can understand, remember and track. NPS produces a single number that can be placed on a management dashboard, compared against competitors, and monitored over time. It does not require statistical literacy to interpret. That is a real advantage in organisations where insight needs to land with commercial teams rather than solely with research departments.

NPS also provides a consistent, comparable basis for tracking brand loyalty over time and benchmarking against the competitive set. In a customer satisfaction study or a brand tracking programme, the ability to see whether the score is improving or declining, and how it compares to competitors in the same category, has genuine strategic value. Even if the absolute score is subject to the methodological criticisms outlined above, the relative movement and competitive positioning remain meaningful, because the same structural anomalies apply equally to all brands being measured. This is the relative comparability argument, and it is robust.

The Promoter behaviour data is also more compelling than sceptics sometimes acknowledge. Research published by Temkin Group in 2017, based on 10,000 US consumers rating 331 companies across 20 industries, found that compared to Detractors, Promoters are over four times more likely to repurchase from a company, over five times more likely to forgive a company if it makes a mistake, and over seven times more likely to try new offerings. Those are meaningful behavioural differences, and they speak directly to brand loyalty strategies and customer loyalty marketing priorities.

NPS also functions as a useful forcing mechanism within organisations. When a leadership team is tracking a single customer sentiment metric, it creates clarity about what direction is desirable and what it means to improve. It is imperfect, but imperfect clarity is often more useful than perfect ambiguity.

NPS and Brand Loyalty: Understanding the Connection

The question of whether NPS is a useful tool for building brand loyalty depends partly on what we mean by brand loyalty. If we define loyalty narrowly as retention, NPS is an imperfect proxy; plenty of customers who score 6 or 7 continue buying from the same brand out of inertia, cost of switching or limited alternatives. A retention rate does not tell you about the quality of the relationship. NPS, for all its limitations, attempts to reach beyond passive retention to something closer to advocacy.

Genuine brand loyalty, in the strategic sense, rests on emotional engagement, trust and the belief that a brand consistently delivers something meaningful. NPS captures sentiment about whether a customer’s experience has reached the threshold for active recommendation. That is a demanding standard, and brands that consistently produce high proportions of Promoters tend to be those that have done something right in terms of customer experience, product quality, or emotional connection.

Brand loyalty programmes and customer loyalty marketing campaigns benefit from NPS as a direction indicator: it tells you whether the overall relationship is trending towards advocacy or away from it. What it cannot do is tell you what is driving that trend, which is why NPS is most valuable when it sits within a broader measurement framework that includes qualitative research, customer journey analysis and attitudinal tracking. Increasing brand loyalty requires understanding not just the score, but the experience and emotional dynamics behind it.

The Brandspeak Position on NPS

As researchers, we are by instinct purists. We care about question design, data quality and the integrity of the analytical conclusions we draw. The criticisms of NPS outlined in this article are ones we take seriously, and we would never recommend NPS as a standalone measure of brand health.

That said, we are pragmatic. We accept that the likelihood-to-recommend question acts as a reasonable proxy for brand sentiment, even in categories where people would not literally recommend the brand to their friends. Respondents generally understand what is being asked of them, and their score reflects the overall quality of their relationship with the brand rather than a literal assessment of their recommendation intentions.

We are also persuaded by the relative comparison argument. In brand tracking studies and customer satisfaction programmes, NPS is most valuable not for its absolute value, but for its directional movement and competitive positioning. Those comparisons hold even if you reject the absolute score as a perfect measure of advocacy.

Our position on gaming is unambiguous: if frontline staff are measured against NPS results, the score rapidly ceases to measure anything useful. Reichheld himself recommends de-linking NPS from employee compensation entirely, and we strongly endorse that view. A coerced NPS is not a measurement; it is a performance.

Where we add value for clients is in reporting NPS results in context, drawing on the full dataset to understand what is driving the score and where the brand can take concrete action to improve it. A score on its own is the beginning of a conversation, not the end of one.

Conclusion

Net Promoter Score is neither the magic metric its most fervent advocates claim, nor the fool’s gold its harshest critics suggest. It is a tool: genuinely useful when deployed intelligently, misleading when gamed or treated as sufficient on its own.

Its value lies in its simplicity, its comparability and its ability to surface the proportion of customers whose experience has reached the threshold of active advocacy. Its limitations lie in its single-question design, the structural over-counting of Detractors, its vulnerability to manipulation, and its inability to explain what is driving the score.

For organisations serious about building brand loyalty, NPS is best understood as one component of a richer measurement framework, not a substitute for it. It answers one important question about how consumers feel. For the harder questions about why they feel that way and what to do about it, you need more. To find out how Brandspeak can help you get that deeper understanding, visit our brand tracking and customer satisfaction research pages, or get in touch directly.