Net Promoter Score: What It Actually Measures, How to Calculate It Correctly, and Why Context Matters

NPS has one of the most interesting stories in market research. It started as a simple question: how likely are you to recommend us to a friend or colleague? It was published in the Harvard Business Review in 2003 by Fred Reichheld at Bain and Company with the claim that it was the single best predictor of company growth. The research world adopted it at extraordinary speed. Within a decade, it was embedded in CX programs at thousands of companies globally.

It is also one of the most argued-about metrics in the industry. The original growth correlation claim has been challenged by multiple independent studies. Yet NPS persists, because despite its limitations, when used correctly and in context, it provides a genuinely useful lens on customer loyalty.

Here is what it is, how to calculate it properly, and what to do about what you find.

The Question, The Scale, and The Three Categories

NPS is derived from a single question asked on a 0 to 10 scale: On a scale of zero to ten, how likely are you to recommend us to a friend or colleague?

Respondents are then sorted into three groups:

• Promoters (9 or 10): loyal and enthusiastic. Likely to recommend and to repeat-purchase. The most valuable customers from a growth perspective.
• Passives (7 or 8): satisfied but not actively advocating. Not counted in the NPS calculation but strategically important because they are the easiest group to convert into promoters with relatively small improvements in experience.
• Detractors (0 to 6): unhappy customers who may actively discourage others. The range 0 to 6 feels counterintuitive (surely a 6 is not that bad?) but reflects the finding that only genuinely enthusiastic customers actually recommend, and anyone below 9 tends to behave more like a detractor than a promoter in practice.

The Formula

NPS = % Promoters minus % Detractors

The result is expressed as a number between -100 (everyone is a detractor) and +100 (everyone is a promoter). Passives are excluded from the formula entirely but included in the denominator when calculating percentages.

Example: You survey 200 customers. 100 score 9 or 10 (Promoters = 50%). 60 score 7 or 8 (Passives = 30%, not included in formula). 40 score 0 to 6 (Detractors = 20%). NPS = 50 minus 20 = 30.

An NPS of 30 is not intrinsically good or bad. It depends entirely on your industry benchmark, your own historical trend, and what your competitors are scoring.

Interpreting Your Score Correctly

1. Always compare to an industry benchmark

NPS scores vary enormously by industry. According to Qualtrics XM Institute data, the grocery industry average is around 30, video streaming around 29, and consumer payments around -6. A score of 20 in consumer payments is genuinely strong. The same score in a premium consumer goods brand is a warning signal.

2. Track trends, not snapshots

A single NPS score is almost meaningless in isolation. Its value comes from tracking it over time. Is it rising or falling? What correlates with the changes? A score that dropped 8 points in a quarter needs an explanation rooted in operational events, not just a target to recover it.

3. Segment by customer group and touchpoint

Aggregate NPS hides important variation. A brand might score 45 overall while scoring 15 among first-time buyers. The problem is not with loyal customers. It is with onboarding. Segmenting NPS by customer cohort, tenure, product type, or geographic region surfaces the operational story behind the number.

The Limitations of NPS Every Researcher Should Know

Academic research has consistently challenged the claim that NPS is the single best predictor of growth. A landmark study by Keiningham et al. published in the Journal of Marketing found that NPS performed no better than other satisfaction metrics in predicting revenue growth. This does not invalidate NPS as a useful operational metric. It does mean it should not be treated as a uniquely authoritative growth indicator.

• NPS is a single question. It cannot explain the reasons behind the score. Always pair NPS with follow-up open-text questions asking respondents why they gave their score.
• Response rates to NPS surveys are declining as customers are bombarded with requests for ratings. Low response rates introduce non-response bias: the customers most likely to respond are often the most engaged, skewing scores upward.
• NPS scoring norms differ by culture and country. In the Netherlands, high scores are structurally less common than in the US, meaning a Dutch NPS of 30 likely reflects stronger loyalty than an American NPS of 30.

Relational vs Transactional NPS

Relational NPS is measured periodically across the full customer base, independent of any specific interaction. It tracks overall brand loyalty and is used for strategic benchmarking.

Transactional NPS is triggered immediately after a specific interaction: a purchase, a support call, a delivery. It measures the quality of a specific touchpoint and is used for operational improvement.

Most CX programs use both. Relational NPS tells you where you stand. Transactional NPS tells you which touchpoints are creating or destroying promoters.