The Net Promoter Score consists of one question on a scale from 0 to 10. Anyone answering 9 or 10 counts as a promoter, 7 and 8 as passive, 0 to 6 as a detractor. The score is the percentage of promoters minus the percentage of detractors — it runs from −100 to +100. Short answer to the question in the title: as the only question no, as one of eight yes. But the reason lies elsewhere than is usually claimed.
What the research actually says
Two extremes circulate online: that NPS is the only number you need — or that it has been debunked. Neither is true.
The much-cited study by Keiningham and colleagues (2007) in the Journal of Marketing found that intention to recommend is not superior: it does say something about future recommendation behaviour, but not as a single indicator, and it does not reliably beat other metrics.
Later work confirms this picture of equivalence. Van Doorn and colleagues (2013) found NPS correlated with revenue growth (r = .46) — but every other measure they tested landed in the same range, at r = .42 to .49. De Haan and colleagues (2015) found practically identical values for two-year retention: NPS (r = .170) and plain satisfaction (r = .184).
The fair summary: NPS is no worse than other satisfaction measures. But it is no better either — and all of these relationships are moderate. Whoever collects NPS is essentially collecting satisfaction on a particular scale.
The bands are not set in stone either
The cut-offs at 6 and 8 look precise, but they were set rather than derived. Kristensen and Eskildsen (2011) found groups in their data that did not fit this split, and proposed 0–4, 5–7 and 8–10 instead.
On top of that comes a statistical problem that hits trainers in particular: a difference between two proportions needs larger samples than an average does to be stable. With twelve participants, a single person shifts the score by more than eight points. An NPS from a seminar with fourteen people is not a metric, it is noise.
Why it misses the point in a seminar
NPS comes from consumer business with repeat purchase: you recommend a mobile provider, a hotel, a piece of software. There, recommendation is a sensible proxy for the value of a customer relationship.
A seminar is something else. It is not meant to please, it is meant to change something. And the research on training evaluation is clear on this point: participant reactions correlate only moderately with immediate knowledge gain and practically not at all with delayed knowledge (Sitzmann et al., 2008). Affective reactions do not correlate with transfer (Blume et al., 2010).
A day can have an NPS of +80 and change nothing. An uncomfortable day that catches somebody on a sore spot can score low and be the most effective one of the year. Optimise for recommendation and you optimise for likeability. In professional development that is a dangerous target.
When it does make sense after all
When your client collects it anyway. Many large companies run NPS across every initiative. Then the question is not whether it is any good, but whether your seminar shows up in their system at all. Include it — as one question among several.
When you compare across many runs. Across thirty seminars the noise turns into a signal. A single score says nothing; a series does.
When you sell it yourself. With open enrolment, recommendation really is a sales channel — there NPS measures something that affects you directly.
The better question
If you were only allowed one question, it would not be the one about recommending, but this:
What will you do differently in the next two weeks, concretely?
It is open rather than scaled, so it takes more effort to evaluate. In exchange it measures intention rather than mood — and intention is what comes closest to transfer. The answers also tell you which parts of your day actually landed. No number tells you that.
A usable form needs both: one or two scaled questions for comparability and two open ones for insight. How such a form is built is described in Feedback forms for seminars: questions and scales. And why the spoken round at the end does something completely different from the form is in The feedback round in a workshop.
In summary
- NPS is not debunked, but it is no better than simple satisfaction questions
- The split into promoters and detractors was set, not derived
- Below about twenty participants the score is unstable
- Satisfaction says little about learning and nothing about transfer
- As one question out of eight: useful. As the only question: misleading
