NPS: the number you present to the board that says nothing.

NPS appears in almost every board deck. It often arrives alone, large, with a green or red arrow depending on the quarter. It is presented as the health indicator of the customer relationship. It reassures or worries. It rarely triggers a decision.

Not because NPS is wrong. Because we ask it to say things it cannot say.

What NPS actually measures — and what it does not

NPS measures the declared intention to recommend at a given moment, among a population of respondents. That is its exact definition. No more, no less.

It does not measure actual loyalty. It does not measure retention. It does not measure the value of respondents versus non-respondents. It does not distinguish an active promoter — who actively refers you to their network — from a passive promoter who clicked 9 out of habit. It aggregates signals of very different nature into a single number.

A score of 42 can mean many things. It can mean your most engaged customers recommend you strongly and your most dissatisfied customers no longer respond. It can mean the reverse. It can mean your response rate has fallen 30% and only satisfied customers bothered to complete the survey. The score does not tell you which of these scenarios is yours.

Response rate: the number behind the number

The first thing to look at behind an NPS is the response rate. If your survey is sent to 10,000 customers and 600 respond, your NPS is calculated on 6% of your base. The remaining 94% have an opinion you have not measured.

The problem is not statistical. It is behavioural. Highly satisfied customers respond. Highly dissatisfied customers respond. Customers who have stopped caring about you — those who leave silently, who do not renew, who gradually reduce their purchases — often do not respond. That silence is not neutral. It is valuable. And it is invisible in your score.

An NPS rising on a falling response rate may well indicate your engaged customer base is shrinking, and only historical loyalists continue to interact with your surveys. The number improves while the relationship erodes.

An NPS without a response rate is a fraction where you only see the numerator.

When strategic customers leave in silence

Aggregated NPS particularly masks the loss of high-value customers. These customers typically represent 20% of the base and 60% of revenue. When they are dissatisfied, they rarely leave noisily. They shift their budget to a competitor, gradually reduce engagement, or let the contract expire without seeking renewal.

Meanwhile, your overall NPS can stay stable. Volume customers, more numerous, continue to respond to surveys. The score holds. The portfolio, silently, shrinks.

This is why an unsegmented NPS fails to protect what matters. It treats equally a customer worth €50,000 in annual revenue and a customer worth €500. Their weight in the score is the same. Their weight in your income statement is not.

What to cross-reference with NPS to make it useful again

NPS becomes an instrument once you stop reading it alone. Four cross-references change the picture.

First: segmentation by customer value. Calculate your NPS separately for strategic customers, growth customers, and volume customers. The trajectories often diverge. If your strategic customer NPS is falling while the aggregate rises, you have a problem the overall figure masks.

Second: verbatims. A score without verbatims is a measure without a diagnosis. Verbatims say why. They identify recurring irritants, moments of breakdown, unfulfilled promises. Without them, you know something is wrong. You do not know what to correct.

Third: behavioural data. Compare declared NPS with observed behaviours: purchase frequency, renewal rate, inbound contact rate, escalation rate. A promoter who contacts customer service three times per month has a different profile from a promoter with no contacts. Declaration and behaviour sometimes tell two different stories.

Fourth: trend, not point-in-time score. A stable NPS of 42 over six months says something different from a 42 rising over twelve months or just fallen from 55. Direction and speed of movement often have more value than the score itself.

What we must stop asking it to say

NPS cannot be your sole customer relationship objective. When it becomes so, three drifts appear systematically.

First drift: short-term management. Teams optimise for the next score. Surveys are sent after positive experiences. Customers are not solicited after an incident. Sales teams are pushed to "prepare the ground" before survey time. The score rises. The relationship does not improve.

Second drift: conflating satisfaction with recommendation. A customer may be highly satisfied with your service and never recommend you — because your offer is too technical, too niche, too personal to be passed on in their network. Conversely, a customer may recommend you out of habit without being particularly engaged. The correlation between NPS and actual word-of-mouth is weaker than assumed.

Third drift: no link to business value. An NPS that does not connect to revenue, retention rate, or wallet share remains a comfort indicator. It says customers are happy. It does not say whether that satisfaction generates measurable economic value.

NPS has a place in CX management. Not the top place, and certainly not the only place. It is one signal among others, useful when segmented, contextualised, cross-referenced with real behaviours, and linked to concrete decisions. Presented alone at a board meeting, it creates the impression of steering. It is steering nothing.