Introduction

The Net Promoter Score or NPS was introduced by Fred Reichheld, a senior Bain executive, twenty years ago with the claim that it is the ‘one number you need to grow’ (Reichheld 2003), i.e. the explicit claim is that if a firm improves its NPS score then it will grow revenue. The NPS is derived from asking clients of a firm if they would recommend it to friends and colleagues.  The answers to this willingness to recommend question are condensed into two groups: so-called ‘promoters’ who give a score of 9 or 10 out of 10, and so-called ‘detractors’ who give an answer from zero to 6 out of ten.  Those who give 7 or 8 out of 10 are (bizarrely) considered ‘passive’ clients and cut from the analysis. The percentage of detractors are then subtracted from the promoters to give a ‘net’ score. In Report 104 we highlight that this counting procedure results in excessive variation in scores over time and across competing brands, making the NPS a volatile and problematic tool. Here, we focus on the presumed link between NPS and revenue growth.  Understanding if NPS really does predict future growth is important for two reasons.  One is simply to understand if a widely used metric is actually useful.  The second is that NPS strongly suggests growth can occur without necessarily building more mental and physical availability – the NPS logic is that growth can happen (simply, or solely) from customers recommending the firm.

Background to the origins of NPS

Reichheld launched the NPS in a Harvard Business Review article and a subsequent book. He claimed to have evidence that it was a driver of growth, based on consumer surveys in 14 industries for around 400 firms, that apparently found the NPS was the best indicator of revenue growth in the majority of cases. It’s telling that this relationship between NPS and growth was only ever shown for approximately 35 firms out of the 400, raising the question of whether the evidence shown in Reichheld’s book and article was selective. Also, US researchers (Keiningham et al. 2007b) pointed out the NPS scores in many of the examples at least partly post-dated the growth figures, casting doubt on whether NPS predicted growth or whether growth caused high NPS scores.

We start our examination of the evidence for NPS by reviewing past work. It is good practice to find out what is already known or not known about a research topic by examining what has been done to date. So, we first reviewed past studies on the link between NPS and growth or closely related performance metrics. The results of these studies are summarised in Table 1. This represents all relevant studies that were able to be identified. We highlight the main findings of these studies, but also critique either their method or interpretation in some cases. Some studies certainly show a positive association, but many show the opposite. We classify the overall result as a yes, a ‘yes but’ (if the result was supportive but there is a method issue), mildly supportive, or as a no. There are not many results for which we can say ‘yes, this supports Reichheld’s claims’. Overall, this summary casts doubt on the idea that there is a reliable link between NPS and future growth.

Table 1: Summary of work to date on NPS and revenue growth (or related measure)

Discussion – the possible biasing effect of persistence in firm growth over time

Two studies that showed quite strong support for the NPS-growth link were by Owen and Brooks (2008) and Sauro (2018, 2019) that showed the relationship between NPS and growth in scatterplots, while also paying attention to the fact that the NPS metric should be for a time period before the growth occurs. And, both of these studies do show the higher the NPS score, the higher the revenue growth in the following year. However, this doesn’t necessarily mean the NPS is a leading indicator of growth. First, there is some evidence that company growth persists from one year to the next (Coad et al. 2018 p. 67; Chandra & Ro 2008). That is, if a firm grows in one year, it’s a bit more likely to grow again next year. So, a firm could grow, and as a result, see a higher NPS (new customers could be more likely to say they will recommend) – then the company carries on growing the next year, giving the impression the NPS is a lead indicator of firm growth. So, the effect could be simply some persistence or year-to-year momentum in growth that is driving this year’s NPS scores, as well as next year’s revenue. Therefore, a stronger test of NPS is to look at specific firms over longer periods of time (5-10 years), to identify if NPS scores are related to future growth for each specific firm, over time. Then, to calculate the average over-time relationship between firm’s NPS and their future growth in the form of a correlation. That is what we did, which we report on now.

Empirical study

To construct this stronger test for the NPS we searched for publicly available information on NPS for firms. We identified data for US airlines and supermarket chains, as well as a multi-year dataset of scores for US insurance companies. We then matched the firm’s NPS scores to their revenue, either from their annual reports, the US Bureau of Transportation for some airlines, or in the case of the insurance company, from industry reports. We were able to assemble matched NPS and revenue data for six US airlines for periods between four and ten years; and for four US supermarket chains for between six to eight years. For insurance, we obtained data for the top ten US car and insurance companies for 2017-2020.

For the airlines and supermarkets we first calculated the correlations between NPS and current revenue. Since this is an over-time analysis, it would be a fair test to examine whether higher or lower NPS for a firm over time is related to it having higher or lower revenue over time. We then also examined the over-time association between NPS and current growth (this year’s revenue minus last year’s revenue); and future growth (next year’s revenue minus this year’s revenue). The correlation between NPS and revenue was r=0.21, which is positive, but with a wide margin of error such that we can’t say it is actually greater than zero. The correlation between NPS and current revenue growth was in fact negative (r=-0.37). The correlation between NPS and future growth was also negative (r=-0.14), but again had a very wide margin of error.

For the insurance companies we found positive, albeit weak, correlations between NPS and future growth of r=0.17 and r=0.21 for car and homeowner insurance respectively. But we also noticed that the NPS scores correlated more strongly with past growth (r=0.31 and 0.33) than future growth. In fact, we found past growth accounted for future growth far better than the NPS did, with correlations of r=0.34 and r=0.58 between past and future growth.

Conclusion

Our overall conclusion is that NPS does not predict future growth. It seems to be more of a reflection of past and current growth. While obtaining a high NPS score is certainly not a bad thing, expenditure on NPS measurement, analysis and related managerial interventions could very easily simply reduce profits. This is because if it does have a positive effect on revenue growth then that effect must be very mild, because it’s so hard to see.