Customer Measurement Problem 8

STUDYING CURRENT CUSTOMERS TO UNDERSTAND DEFECTION

Jump To Page – Intro, 1, 2, 3, 4, 5, 6, 7, 9, 10, 11, 12, Conclusion

Consider the prevalent logic: measure and manage customer satisfaction to keep customers. Stated differently, measure and manage customer satisfaction to make sure we do not lose customers. These statements appear on the surface to say the same thing. But it does beg a question about keeping vs. losing customers. If we identify the set of factors on which strong performance helps us keep existing customers, have we by definition identified the set of factors on which poor performance leads to losing customers? I think an important distinction makes the answer no. I believe there is a fundamental flaw in the logic of studying current customers to understand customer defection. For one thing, sampling from the current customer base will provide no information from actual defectors.

A simple example may be helpful here. Consider a company that implements a substantial rate hike. Assume that their customer base is a finite mixture of two underlying segments – a highly price sensitive segment, and a quality-oriented segment. The rate hike is timed to occur a few months before the next wave of customer satisfaction measurement as an intentional attempt to create a kind of quasi-experimental design: measure, intervene, re-measure (a one-group pretest-posttest design). The data reveal that price ratings actually improve from the pretest to the posttest. Management pats itself on the back for a job well done: “we’ve raised rates and it did not affect customer satisfaction negatively. In fact, our scores on price improved!” Not so fast. Let’s consider the role of defection.

Just for sake of example here, assume all of the customers in the underlying price sensitive segment terminated the business relationship immediately after the rate hike went into effect. Having exited the customer base, they were no longer active/valid when sample was drawn for the posttest survey. That means in the posttest survey, we were only surveying the quality-oriented sub segment. Where the pretest was a mixture of latent classes, the posttest was essentially only one of those classes the one with more favorable ratings of price. The price sensitive segment is gone. Presuming the price sensitive segment was less positive in ratings of price from the start, we’ve removed a group of people who tended to score lower, thus leaving a group of people who tend to score higher. This accounts for the increase in price scores from pretest to posttest, and reveals why measurement of the current customer base alone can be misleading, at least with respect to issues that cause defection. Defection should be studied as its own subtopic. Churn/defection rates should be known and monitored. Lost customer research should be conducted to uncover systemic controllable root causes of loss.

For companies interested in deeply understanding the customer experience across the entire customer lifecycle, it is a mistake to focus only on current customers if we want to understand customer defection processes. Logically, how can customers who haven’t defected be good informants about defection processes? To understand why customers leave, we must treat lost customers as a separate sample frame. Only by talking to lost customers can we begin to tease out underlying root causes. Did a customer leave because of something the company did to drive them away, because of something a competitor did to pull them away, or something essentially uncontrollable from a marketing standpoint (e.g., homeowners insurance cancelled when someone moved into assisted living)?

One final caveat on this section. It may be possible within the current customer base to identify customers at risk for defection. In fact, early identification might allow for intervention. While this approach can be highly valuable (e.g., rescuing perishing accounts), by itself, it cannot offer a comprehensive picture of defection processes for all the reasons previously described. But, the identification of current at risk customers can be aided by the use of customer defection data when done in combination.

If in studying lost customers, we discover segments or types with particular profiles or particular experiences, we might use that as a pattern against which to screen existing customers. Almost like scoring higher probability prospects in direct marketing, now we are trying to score the existing customer base regarding probability of defection.

A second use for such profiling regards new customer acquisition. If certain kinds of customers, with identifiable characteristics, tend to come into the fold, feel their needs are not being met, and thus ultimately defect, it might be much better to try to screen out these kinds of customers before they ever enter the system! Rather than creating the profile of a target segment to acquire, here we are creating an anti-target profile – the kind of prospect we don’t want to acquire. Thus, lost customer research can have bearing on how we deal with existing at-risk customers, and how we think about certain elements of the customer acquisition process. It is an important, but often neglected area of study.