What NPS measures, and what it never can

NPS asks a single question: how likely are you to recommend us? The answer is a statement of future intent, self-reported, in the moment. It is not satisfaction, not loyalty, not actual behavior, and not an explanation of anything. The danger was never the number itself. It is the weight organizations pile onto it: targets, bonuses, rankings, whole strategy reviews hung on a score that cannot say what to fix, where friction lives, or how customers actually behave when the survey window closes.

Why it matters to the business

The evidence counsels humility. Keiningham and colleagues, studying 21 firms and more than 15,500 interviews in the Journal of Marketing, found NPS predicted revenue growth no better than the standard satisfaction index it claimed to beat. CEB research found Customer Effort Score twice as predictive of loyalty as NPS. And stated sentiment misses defectors entirely: Bain's Reichheld found 60-80% of lost customers said they were satisfied or very satisfied just before leaving. Even Reichheld, who created NPS, now calls bonus-linking an abuse that breeds gaming, and pairs the score with an audited, accounting-based Earned Growth Rate so that hard revenue evidence keeps the survey honest.

How to use it

• Use relationship NPS sparingly, as a top-level trend line reviewed quarterly.
• Pair every score with an operational driver metric, such as first-contact resolution or repeat contacts.
• Cross-check scores against behavior: repurchase, retention, expansion, actual referrals.
• Never rank individuals or attach bonuses to the score, and audit sampling for pleading and gaming.
• Route every detractor comment into root-cause work with a named owner.

Where teams get it wrong

Teams turn the thermometer into the target. Bonuses get attached, the frontline pleads for tens, sampling quietly skews toward happy customers, and the score climbs while churn stands still. The number stops measuring loyalty and starts measuring how well the organization has learned to farm it. Used that way, NPS does not just fail to help; it actively hides the problems it was meant to surface.

Ask your team

• What decisions did our last NPS reading actually change?
• Does anyone's pay depend on the score, and what would they do differently if it didn't?
• When the score moved last quarter, did retention or revenue move with it?

A score without diagnostics is a verdict without evidence.