Somewhere in your product, right now, a modal is waiting for a customer to finish loading a page so it can ask them, on a scale of 0 to 10, how likely they are to recommend you to a friend. This single question has launched a thousand board slides, an entire software category, and a number — "our NPS is 47" — that gets stated in rooms full of serious people with the confidence of a vital sign, despite describing something closer to a mood.

What the number is actually made of

Nine means promoter. Six means detractor. Seven and eight are discarded entirely, a full fifth of your respondents simply deleted from the math because their feelings were too moderate to be useful. The score is the percentage of promoters minus the percentage of detractors, which means a company where 60% of customers rate you a 9 and 40% rate you a 6 posts the exact same score as a company where everyone, unanimously, rates you a 7 — a number that does not exist anywhere in the final calculation, despite being the actual average opinion of every single person surveyed.

The question nobody asked for

"How likely are you to recommend this to a friend" assumes your customer has friends who need enterprise expense-reporting software, and that recommending it to them is a normal unit of social currency, on par with a restaurant or a movie. Nobody has ever said, unprompted, at a dinner party, "you have to try this vendor management platform, you'll love it." The question measures willingness to perform an act that occurs in approximately zero real human lives, and then treats the answer as a proxy for love.

The follow-up box nobody reads

Every NPS survey ends with an optional text field: "What's the reason for your score?" This box fills up with genuinely specific, often useful feedback — a feature request, a complaint about a renewal call, a typo in an email — none of which makes it into the slide, because the slide has one number on it, and the number came before the box, chronologically and spiritually. The box is where the truth goes to be stored, unread, as a kind of alibi: "we do ask for context."

What a swing actually means

NPS goes from 41 to 47 in a quarter, and someone presents this as a six-point win, attributable, confidently, to the new onboarding flow shipped in week three. It is at least as likely to be attributable to which twelve customers happened to open the survey email that week, a sample size small enough that a single annoyed customer forwarding it to a happy coworker can move the company-wide number more than any actual product change. The chart implies causality. The chart is mostly describing who clicked.

The benchmark nobody can verify

"Anything above 30 is considered excellent" is a sentence repeated at every company, sourced from no particular study anyone in the room has read, applied uniformly across industries as different as consumer apps and enterprise compliance software, as if a single number could mean the same thing for a tool people choose for themselves and a tool procurement forced on them. The benchmark isn't wrong, exactly. It's just doing the job of a real standard without ever having been held to the scrutiny one would require.

What we're not saying

We're not saying customer sentiment is unmeasurable, or that a long-term NPS trend tells you nothing — a number that reliably craters after a price increase is telling you something true. We're saying the number gets asked to do far more work in a board meeting than its actual construction supports, and that "our NPS is 47" is frequently presented with a confidence that would require ignoring everything above about the scale it's built on.

Our recommendation

Keep running the survey — it's cheap, and the free-text box really is useful, provided someone reads it. Just stop putting the single number on a slide by itself, trending quarter over quarter like a stock price, as if six points of movement in a self-selected sample of your most opinionated users constitutes a finding. Put the actual sentences next to it. They were always the data.

If your last board deck had an NPS slide with no sample size on it, email us at contact@b2bsaas.guru. We'll help you build one with a sample size on it. It won't mean more. It'll just be more honest about meaning less.