That is akin to the experience of some credit card customers recently when they “reactivate”—suddenly using a card they have held for a long time but rarely used—and the issuer responds by reducing or restricting the card’s credit line. It’s not a perfect analogy, of course—credit lines are not airplane tickets—but viewed through the eyes of a credit-worthy customer, the practice is confronting if not offensive. Viewed through the eyes of forward-looking lenders hungry for good customers, a reactivating customer would seem to be just as much cause for rejoicing as for alarm.
From October 2008 through March 2009, “the country’s three largest credit card lenders…cut unused lines by almost one-quarter, to about $2 trillion," according to American Banker. As an exercise in sheer risk reduction in the midst of unprecedented uncertainties, this was prudent. These and other lenders reduced their reserve requirements for unused and latent risk credit lines and wiped billions of dollars of exposure from their balance sheets, much of it from the millions of dormant or inactive credit cards.
Review the likely scenarios
But reactivated cards are by definition not inactive. They are customers suddenly acting like customers, if belatedly. When that happens, lenders whose risk antennae have been newly sensitized need to keep two faces of risk in mind. The risk of not getting paid back is one face. The risk of running off a hard-won customer is the other. Thanks to predictive analytics today and refined preemptive risk management models, lenders can quickly learn which risk they are looking at when a customer suddenly reactivates.
If they focus exclusively on the first face of risk, they may find themselves asking, when the recovery comes and they begin actively extending credit again, “What happened to all our good customers? Why did we fire all these customers we are now trying to bring back into the fold?”
After all, what does it mean when a cardholder uses a second-in-wallet credit card after not using it for months at a time?
It might mean that the customer prudently used the card for a purchase that in a better economy he might have paid cash for. Or the customer has just read and taken the advice of any of several finance columnists who have recently counseled that cardholders should make occasional, token purchases to avoid having their cards being deactivated.
Maybe the customer has become unhappy with the service provided by another credit card company and wants to try a different one. Or the cardholder has launched a business and wants to keep that activity separate from the family’s personal purchasing activity.
Or a promotion from the issuer struck a chord, prompting a purchase. Or the customer is nearing the limit on the first card in wallet. Or a child home from college needed a card for a day of shopping.
Those are signs of a good, maybe great customer in the making.
Certainly, it could also mean the worst case: the customer, finding himself in serious financial trouble, desperate to access all available credit, plans to consume the entire line on the card as fast as he can.
With unemployment high and credit tight today, the worst case is true more often than it used to be. But lenders may be over-emphasizing the recession and forgetting the future if they presume the worst case when so many other positive possibilities exist and are more or less knowable. Particularly when all the positive possibilities demand the precisely the opposite course of action.
You only have a short window
The fact is, most lenders have yet to take advantage of data and tools that would allow them to predict, with a high degree of accuracy, whether a reactivating customer is a great danger or a great opportunity. The customer was probably credit scored long ago in a different personal situation and under different economic circumstances. Behavior scores would serve better, but a single isolated transaction yields little to no insight for most lenders. And few lending organizations are structured to take a broader view of the customer’s other accounts, balances, and history.
Most worrisome for these lenders, time is not on their side. Our research shows that if the bad case is true—if that reactivating customer is a big write-off in the making—he or she will race through the entire credit line rapidly—typically within 30 days. The lender has a short window in which to answer the burning question, “Is this the beginning of a highly profitable relationship, or a huge write-off?” It’s a binary question—a mystery that has to be solved. If the answer is, as it is for most lenders, “We can’t tell,” the default position in today’s economy is the conservative one: curtail and constrain spending.
But what a costly mistake the default position is if it might have been the beginning of a profitable relationship. No customer-focused business can survive doing business that way—treating with suspicion hard-won customers whose loyalty they have subsidized for years—no matter how risky the environment has become.
Instead, lenders need to be able to answer that binary, burning question quickly and accurately. The tools exist for doing so. For most lenders, it is a matter of first committing to a new philosophy about this aspect of risk management, and then deploying the tools that will serve the particular needs of their customer segments and strategies.
Three rules, three tools
The philosophy rests on three rules of preemptive risk management. 1. Use analytics early and often to know your customer well and see signs of trouble as soon as possible. 2. Be alert for the information that every customer interaction reveals, capture it, and apply it to make better decisions. 3. Know the whole customer, not just the account; it is the customer who yields the accurate risk profile.
The tools to support those rules include data, analytic models, and an analytics platform. Banks can build their own, draw on a wide array of market solutions, or deploy a combination of both.
Here’s how each tool contributes to the answer:
Data—To compensate for scarce data on accounts with little activity, some lenders are gathering more data from nontraditional third-party and internal sources, and they are learning how to dig deeper into thin transaction data to see specific purchase details that help them understand more about the implications of a reactivating customer’s transactions. And they are settling for shorter horizons, too, not trying to predict risk using stale data, but leveraging recent, relevant information.
Analytic Models—Special situations require specific analytic models aligned to those precise circumstances. In the case of reactivation, models that require significant activity or past history are useless. Instead, some lenders are learning to apply a model that has proven effective in predicting high-risk accounts based on specific transactional activity. In a matter of hours, lenders can update the score on the customer, determine the appropriate strategy, and take direct action at the customer level. Clearly, this is a powerful tool for preventing high-risk customers from building large delinquent balances, but it is equally powerful for revealing to the lender a customer that should receive a promotional offer or other outreach reserved for desirable customers.
Analytic Platform— Most lenders’ legacy systems are simply not designed for the speed and nimbleness today’s analytics require. Instead, many are incorporating new event-based analytic platforms that make it possible to aggregate many sources of data quickly, execute their predictive analytics, and facilitate decisions about appropriate treatments, customer by customer. They are better able to identify high-risk changes more quickly, take actions to prevent losses, and recognize profit-generating opportunities before their competitors.
In the same news article referenced above, an industry advisor said, “If a cardholder starts using a dormant line, the issuer should be worried. You’ve got to get him before he gets you.” Perhaps. But if he is a good customer, it is equally important that you “get” and hold him before a competitor does.