*** Updates to this page are likely to be especially significant while the page is being developed. Please do tell me when you find errors. I believe in being forthright but the good part of that is that I can cope it if you disagree or simply have spotted obvious errors that I have made. ***
The Net Promoter Score is a strange metric. It has broken through in ways that no other marketing metric that I can think of has. It is a very popular recommendation metric. But recommendations are complex. Looking at the photo you can ask whether I would recommend summer holidays in Atlantic Canada? Maybe we need more than just a single number to express the complexity of my thoughts; Atlantic Canada is beautiful and has some great sights — e.g., L’Anse aux Meadows is incredible — but I wouldn’t want to visit every summer given the weather was pretty appalling. To be fair NPS is now thought of as more than just a single question but it was initally (sometimes still is?) sold as a quasi-magical approach to marketing. Marketing research for people who know what they like and like what they know.
What is NPS? (Net Promoter Score/Net Promoter System)
I usually start by asking what a metric is for. From here we can create the metric. NPS is a little different. Unlike many other metrics it is very clear what NPS is. As Fred Reichheld created the metric he gets to tell us what it is. (He is the NPS pope and is infallible when he defines NPS — though not when he tells us what NPS does).
Net Promoter as a metric is very simple. The question asked is “How likely is it that you would recommend our company to a friend or colleague?” with responses on a 0 to 10 scale. Creating the metric involves comparing the % of customers who give high scores with % of customers who give low scores. The Net Promoter Score is just the high scores percentage — customers who answered 9 or 10 — minus the low score percentage — customers who answered 6 or less. (Customers responding 7 or 8 are termed passive and ignored). Any positive numbers show you have more recommenders than detractors. It is pretty intuitive that firms whose customers are happy to recommend are, all else equal, probably more succesful than firms whose customers aren’t.
What Is NPS For?
The challenge comes with getting a bit more detailed idea what NPS is actually for. What does it do exactly? Ironically, this marketing metric has suffered from its marketing. At least I should say that it has suffered in the eyes of academics. In many ways the marketing of NPS has been little less than brilliant. The brilliant marketing is, of course, one reason why I’d doing a page on it. NPS has been a marketing phenomenon. A simple Google Trends analysis with NPS in blue and the American Customer Satisfaction Index in red (another way of monitoring customer satisfaction) show interest in NPS dominating interest in ACSI in recent years.
The One Number You Need To Grow
Fred Reichheld called it the ‘one number you need to grow’ in a famous HBR article. The article has been a massive success. It has garnered 3,317 citations on Google Scholar at time of writing (09/22/20). For a practitioner focused article to gain such citations is incredible.
The ‘One Number You Need To Grow’ is a great title for a paper, though sadly it had more than a touch of sales flim flam. Reichheld didn’t really provide any compelling evidence for his claim. There were a few pretty pictures showing positive correlations but Reichheld didn’t share any real proof.
Is NPS Just The Number?
NPS supporters, to be fair, nowadays tend to emphasize that NPS is a system — including feedback loops, follow up questions etc… It is the whole system, not just the one number, which is what matters. The system, including the actions one takes in response to the numbers, not the score per se, is the thing that brings success. This makes more sense than a single number having magical properties. Unfortunately, whether the system is driving growth is much harder to test statistically than testing a single metric as you need to make sure that the system is being implemented effectively.
If a manager doesn’t get growth when using NPS maybe it is the manager’s fault, not the metric’s. This might be because the manager didn’t do the system properly. Did you hire the experts on NPS to help with implimentation? Maybe, Satmetrix? Of course, this is more great marketing — honestly I have to say I am impressed — this time selling services. Still the need to hire in experts limits one of NPS’ most appealing features, namely that it is non-proprietory.
What Are the Problems With The Claims For NPS
Correlation is Not Causation
This is a big one. Are you surprised that happy customers tend to say they would recommend? It isn’t clear though whether managing recommendations, i.e. knowing and acting on the NPS results, are really the thing that will drive success. This idea is plausible but it isn’t shown properly in anything I have read. A few regressions are presented in various NPS supporting pieces but that doesn’t show what is driving success. It just shows NPS tracks success not that NPS drives success.
Of course, confusing correlation and causation has a long tradition. Doing so leads to weird conclusions — see Tyler Vigen’s site for some excellent fun examples. (The classic being Nicholas Cage movies correlates with number of people drowning by falling into pools. Hopefully these do not cause each others).
What Is Growing? It Matters Whether It Is Revenue Or Profits
Often it isn’t clear what exactly is growing. NPS work often seems to discuss profits in the text but it is usually revenue used in any analysis. The difference matters. It is easy to imagine how you can increase revenue by being more generous with customers. Such generosity might also increase NPS, Who wouldn’t recommend a free service? That said, this might not be the best business idea as you might gain customers at a loss. Profits are not revenues. A lot of the NPS advice is not clear onyou should always be clear on the difference.
I don’t see this as a major problem here but always remember that it is hard to compare across industries. Some industries are easy to generate recommendations, others much less so. When was the last time you recomended a smart phone versus when did you last recommend toilet roll? If people rarely discuss toilet roll, they may be less likely to recommend it however great it is. Just remember that a score from one industry is likely to be quite different from the score in another industry. This is true even if both companies are both really great (or really terrible) at what they do.
The arbitrary nature of the system — 9 and 10 are promoters, but not 8s — is a challenge. Some people are just more effusive than others. 8 is a good score to an English person but a much worse score to many US people. 8 probably isn’t a pasive in England but may well be actively promoting the firm. I have seen little justification of categorizations in NPS. As such, we have little idea when NPS as written might be expected to work well and when it isn’t suited to the context at hand.
Why Throw Away Information?
There is little theoretical reason why a metric that throws away data is better than one which uses all the data. Surely someone who answers six is less of a detractor than someone who asnwers zero? We just throw away information on this difference. Why would ignoring the information that there are many 7 and 8s, who are called passives, be better than using this information in some way?
It isn’t clear that NPS is better (or to be fair) worse than other similar approaches.
“We find that all the metrics [including NPS] perform equally well in predicting current gross margins and current sales revenue growth and equally poor for predicting future sales growth and [future] gross margins as well as current and future net cash flows” (Van Doorn et al., 2013, p. 314).Van Doorn, J., Leeflang, P. S. H., & Tijs, M. (2013). Satisfaction as a predictor of future performance: A replication. International Journal of Research in Marketing, 30(3), 314–318. doi:10.1016/j. ijresmar.2013.04.002
While I do appreciate the NPS people highlighting the issues of customer satisfaction/loyalty/recommendations (broadly speaking) I sympathize with critics. Given NPS was very much sold as a single number was all you needed saying you really need a full system seems a bit rich. The famous NPS Harvard article was called the one number you need to grow, not the system you need to grow. (Reichheld, to be fair, in the article does share some sensible implementation advice). NPS simply is not the one number you need to grow.
Simple Is Good
Utlimately, NPS might be useful because it is simple but I haven’t seen any plausible argument why it is the one number you need to grow. We need to have theory — by which I mean just a reasonable causal path. As it is NPS is a little too magical thinking for my liking. It seems like a plausible metric but is it really the one number you need to grow? I don’t think so.
Summary And A Helpful Flow Chart
If you think of NPS as a potentially useful approach to understanding what your customers think then that is great. Yet, if you think NPS is the one number you need to grow (or do anything else for that matter) you are buying into rather more magical thinking that I am keen on.
We might compare NPS to Guinness, the popular Irish stout, which was once marketed as being “good for you.” Guinness and NPS may both have wonderful qualities, but that doesn’t mean one should believe everything said about them.Neil Bendle and Charan Bagga, 2016
That said, NPS is a clear metric and if you can help you focus your organization on customer service maybe it can be a good thing.