What Does NPS Actually Measure?

COCenk Onur Ozan ·29 Sep 2026Published 29 Sep 2026 ·2 min read

We all know the NPS question by heart: “How likely are you to recommend this brand to a friend?” The question is simple and the answer looks simple too. Things get more complicated once we start talking about what the question really measures.

NPS measures an intention to recommend. It captures how satisfied a customer feels, how inclined they are to stand up for the brand, and the social risk they take on when recommending it to someone. It does not measure whether the customer actually buys again, actually recommends you to anyone, how much they spend or how much profit they leave behind. That is exactly where the distance between intention and behaviour opens up.

Behavioural science calls this the intention-behaviour gap. People act on the circumstances they find themselves in more than on what they have said. A customer who gave you a 9 can leave after a price increase, move to a competitor’s campaign or change their habits the day they find a cheaper alternative. That is why NPS on its own should not be read as a causal model.

Finding a correlation between NPS and revenue is not hard. What is hard is knowing which way the arrow points. Does a high NPS raise revenue, or do customers who are already loyal and spend a lot simply give high scores? This is the classic problem of reverse causality. Saying “NPS drives revenue” without building a regression model and without checking customer lifetime value by segment is methodologically weak.

The scale itself is sensitive to context. In Turkey, an 8 often means “pretty good”, yet NPS counts it as passive. In the US, a 7 is already read as a lukewarm verdict. B2B customers tend to give scores that sit very close together, with little variance. All of this changes how scores are distributed; the same experience can turn into a different number in a different setting.

There are situations where NPS is genuinely strong. It works very well when you measure it on the relationship as well as the transaction, analyse it by segment, model it together with customer lifetime value and break it down with driver analysis. At that point you stop treating NPS as a KPI to hit and start using it as an input to a model.

The approach I find sturdier is to build a chain instead of leaning on a single metric. The drivers of the experience connect to a behaviour, and that behaviour connects to a financial outcome. For example, you follow how the delivery experience shows up in repeat purchases, and how repeat purchases show up in customer lifetime value. In that chain, NPS sits in the middle as an intermediate indicator.

So I often ask the teams I work with the same question: are we setting NPS as a target to reach, or using it as a variable that explains our business model? The first creates a reporting culture, the second an analytical one.

Cenk Onur Ozan COO, Cloud4Feed

Writes about customer experience, behaviour and the decisions that follow from listening.

LinkedIn