There Is No Single Best Measure of Your Customers

Single-question customer metrics have become very popular. Companies all have their favorites. Some go for the traditional Customer Satisfaction measure, others the now famous Net Promoter Score. The most recent addition to this panoply is the increasingly popular Customer Effort Score, which tracks the amount of time and effort that customers have to put into solving after-sales problems.

Each claims to be the one metric that managers need to measure, monitor, and act on, because it promises a better correlation with business performance than any other existing measure.

So which of these is actually the best? A lot depends on what you measure performance with. In making the case for their Customer Effort Score, Matthew Dixon, Karen Freeman, and Nicholas Toman correlated the three measures with levels of customer repurchase and overall levels of spending. The results are shown on the graphic in figure (1) below.

But the picture is rather different if you correlate the three measures with customer loyalty. In a recent study, we analyzed 9,732 customer responses related to 102 different service companies in 19 industries. We looked at each measure to see if (a) it predicted loyalty in individual customers and (b) it predicted customer loyalty levels for companies as a whole. The results are presented in figure (2) below.

What's the Best Customer Measure?
(1)
(2)

What accounts for these huge differences? A lot, we believe, has to do with sample bias. The official Customer Effort Score question ("How much effort did you personally have to put forth to handle your request?") can be asked only if a customer has actually contacted the company after a purchase for some reason. The Dixon, Freeman, and Toman research only looked at those customers who could answer the question. What this means is that while customer effort scores have good predictive power for spending by one group it can say nothing about the loyalty and spending of all types of customers.

Our research, however, included customers who couldn't answer the Effort question: of the nearly 10,000 customer responses in our data set, only 1,741 had contacted the company with a request.

The takeaway from his comparison is that you need to be very careful what you use your metrics for. Customer Satisfaction is not, perhaps, a good way to predict the future expenditure of people who have to deal with your after-sales service but it does looks like a good way to predict your overall levels of customer loyalty, while NPS is clearly a good predictor of the individual customer's attitudes.

The implication of this is that the growing reliance on a simple single customer metric is a very dangerous trend and we strongly urge companies to adopt a more nuanced multi-dimensional approach to predicting customer behavior. After all, if we visit a doctor for a check-up, we wouldn't want the doctor to rely solely on our body temperature to assess our health, would we?

HBR Article: Thorsten Wiesel, Peter C. Verhoef, and Evert de Haan