Customer Lifetime Value, Part III: “Leading Indicators” of LTV Change

This is the third and (probably) final installment of my essay on customer lifetime values.

In Part I, I suggested that a customer’s lifetime value (LTV) represents nothing more complicated than the value of the customer’s future business. And even though predicting the precise future is impossible, we can still use data and analytics to make intelligent, data-informed estimates of LTV, which allows us to prioritize our marketing activities and customer service budgets.

In Part II, I argued that a second, even more important role for LTV analytics has to do with tracking changes in lifetime value. When you treat a customer better, they will be more likely to do business with you in the future, generating an increase in LTV. So by evaluating customer LTVs at regular intervals, using comparable models, we can track these changes over time, with the long-term goal of raising every customer’s LTV (by, for instance, making the customer more loyal, or selling them more things). 

But now the obvious question:

Is it possible to identify LTV changes as they happen?

The short answer is: Yes. But to do it, your analytics people need to be able to look at current-period transactional and other data – including all the data that goes into modeling customer lifetime values in the first place – and estimate how current-period changes, updates, or trends in these data will affect your customers’ future-period behaviors.

Leading Indicators of LTV Change

If you have a reasonably workable set of LTV models, it shouldn’t be terribly difficult to identify the variables in these models that, when altered, lead to changes in lifetime value. These are your “leading indicators,” some of which will be more important than others. 

I would group these leading indicators into four general categories:

1. Lifetime value drivers. These are the elements of your LTV equation itself – the actual components that determine how much value a customer creates for the company, over time, such as average monthly sales, or perhaps churn rate. When churn increases among a segment of customers, the LTVs of those customers declines.
2. Behavioral cues. The number of contacts initiated, services or products contracted, complaints or comments submitted, referrals made, and payments made or not made – all these behaviors and transactions can serve as cues for predicting a customer’s future intent. If a customer refers another customer to you, their LTV will be higher than before. When a customer complains, their LTV is likely to drop, depending on how the complaint is handled.
3. Attitudes and VOC (voice of customer) data. Attitudes precede behaviors, so direct customer feedback as to satisfaction, willingness to recommend, and likelihood of buying from you again are important. If a customer suddenly changes from a promoter to a detractor on your NPS survey, it is clearly an indication that their LTV has declined. And with enough historically compiled data, you should be able to estimate by how much.    
4. Lifestyle changes. When a customer takes a new job, or gets pregnant, or retires, or gets married or divorced – when his or her lifestyle or personal situation undergoes any sort of change, their LTV may also be affected, and some of those changes will be predictable.

Importantly, leading indicators of LTV change represent the levers that can be used to improve the value of your customer base. Raise your CSAT or NPS score, and improve LTV. Get a spouse to sign on to the same credit card and see a higher LTV for the first cardholder, as well. Allow your customer experience to deteriorate, and see LTV decline.

No matter how good your data and analytics get, keep in mind that this will never be an exact science, and the best you can ever hope for is to make reasonable estimates as to the economic consequences of your company’s marketing and service policies. But the computer power and analytical tools are available today, and not prohibitively costly.

As an alternative, you could remain content with flying blind.

If you don't want to go to the trouble or expense of tracking customer lifetime values because they're too fuzzy, you should hope you never have to fly in a plane in which the pilot refuses to look at the instruments because they're not 100% accurate.