What Is Net Promoter Score (NPS)?

In December 2003, Fred Reichheld published "The One Number You Need to Grow" in Harvard Business Review. Reichheld is a fellow at Bain & Company, and Bain describes the calculation today as follows: start with the question of how likely the customer is to recommend the company to a friend or colleague, and score the answers on a zero-to-ten scale.

Bain splits the answers into three groups. Promoters give 9 or 10, passives 7 or 8, and detractors 0 to 6. The name comes from this split: the net figure is the promoters left over once the detractors are subtracted.

According to Bain, Net Promoter, NPS and Net Promoter System are registered trademarks of Bain & Company, NICE Systems and Fred Reichheld, and Net Promoter Score is a service mark. Anyone may apply the calculation itself; for commercial use of the marks, Bain requires a licence.

The calculation only needs percentages. An example with made-up figures: of 200 respondents, 90 give a 9 or 10, 60 a 7 or 8 and 50 a score from 0 to 6. That is 45% promoters and 25% detractors. NPS is 45 minus 25, which is 20. It is reported without a percent sign.

The value can range from minus 100 to plus 100. Minus 100 means every respondent is a detractor; plus 100 means everyone is a promoter. Passives count only in the total number of respondents, not in the numerator.

The calculation has a quirk: very different distributions give the same value. 20% promoters, 80% passives and no detractors also give 20, just like 60% promoters and 40% detractors. NPS compresses eleven answer levels into one number, and information is lost in the process. Reporting only the number hides how it came about.

The question is asked, for example, after a purchase, after a support case or at regular intervals, by email, in an app or on the website. The answers often end up in the CRM, where they are linked to each customer.

NPS is easy to collect and easy to explain: one question, one number. Bain states that thousands of companies calculate it, including two-thirds of the Fortune 1000. That is a claim by the trademark owner, not an independent count.

Whether the figure predicts growth better than other metrics is disputed. In 2007, Timothy Keiningham and colleagues tested this in the Journal of Marketing with longitudinal data from 21 firms and more than 15,500 interviews from the Norwegian Customer Satisfaction Barometer. In the industries Reichheld cites as exemplars, they could not replicate the "clear superiority" of Net Promoter over other measures.

A second study by the same team, published in 2007 in Managing Service Quality, used a two-year panel of more than 8,000 US customers from three industries. Its finding: recommend intention alone does not suffice as a single predictor of customers' future loyalty, and models with several metrics predicted recommendation and retention better. The authors themselves name as a limitation that only three industries were studied.

In practice, NPS works as a KPI for tracking your own development over time, not as sole evidence of growth or of customer lifetime value.

• Always word the question the same way and ask it in the same channel and at the same moment, so values stay comparable over time.
• Report the split between promoters, passives and detractors alongside NPS.
• Add an open follow-up question about the reason for the score; those answers show what to change.
• State the number of responses, because small samples cause large swings.
• Combine NPS with behavioural data such as repeat purchases or churn instead of reading it alone.
• Compare your value mainly with your own earlier values, not with other companies' figures whose collection you do not know.

• Calculating NPS as the average of the answers instead of the difference between two percentages.
• Comparing values from surveys with a different question, scale or timing.
• Adopting industry benchmarks with no source and no collection method.
• Paying staff according to the score and so prompting them to ask customers for good ratings.
• Concluding from a rising NPS that revenue is rising, without checking.