Card issuers turn to algorithms to determine if you are about to leave them or are having trouble paying your bills — and allows them to step in to turn the tide
A growing number of financial institutions are using a powerful combination of computer algorithms and customer data to better tailor their services to credit card holders.Known as predictive analytics, this number-crunching technology works by taking the data a company collects on a segment of customers and entering this information into a predictive model. This model is comprised of sophisticated computer algorithms that then analyze the data to predict future events.
For some consumers, predictive analytics spells good news. For example, many financial institutions use predictive analytics technology to create “attrition models” — formulas based on purchase activity that flag customers on the verge of canceling their cards and taking their business elsewhere. In turn, these customers are more likely to receive special offers and targeted promotions that may or may not entice them to stay.
On the other hand, predictive analytics can alert a card issuer to when a cardholder is headed for trouble and can step in to either lend a helping hand or take more drastic measures such as cutting credit limits or canceling cards.
A helping hand
Take Premier Bankcard, for example. A South Dakota-based provider of First Premier credit cards, Premier Bankcard is a subprime credit issuer that extends credit to consumers who are more likely to default on their loans. Because of this, Premier Bankcard uses predictive analytics to get an early read on credit card holders who are struggling to make payments and could use a bit of hand-holding.
For customers “who are performing middle of the road and who need help” with their payment schedule,” Rex Pruitt, Premier Bankcard’s manager of the profitability and risk department, says the company offers account “maintenance” services such as financial counseling, as well as “rewards programs for customers who perform well.”
“There are a number of clear markers when people are losing control over their finances,” says Edwin Van der Ouderaa, a senior executive and financial services expert at Accenture, a global management consultancy. “If you can detect those markers early on, a responsible bank can go to those people and say, ‘Listen, we believe you may have a problem. If so, can we help you restructure loads or give you advice on how to manage your budget.'”
Making the grade
More personalized customer service is another positive byproduct of predictive analytics. For instance, Premier Bankcard uses predictive analytics to create a “good customer score” — a figure based on data such as a customer’s outstanding balance, the number of times a customer exceeds his or her credit limit and the longevity of an account.
By inserting these numbers into a predictive model, Premier Bankcard can forecast which customers are most likely to default on their loans, and conversely, which qualify for customer-focused programs such as special promotions and interest rate cuts.
So, too, are credit card companies using predictive analytics to tailor their marketing initiatives for increased customer satisfaction. For example, these days it’s not enough for a credit card issuer to simply identify a customer likely to accept a credit card application. Rather, by creating predictive models fed with data such as purchase history, IP address or even the percentage of time an individual pays a credit card bill online, financial institutions are determining how best to reach out to these customers, whether it’s via direct mail, an email campaign or telephone solicitation.
“More and more credit card marketers are using predictive modeling to determine not only the likelihood of someone responding [to a credit card offer], but what’s the best channel through which to reach a particular prospect,” says Ron Shevlin, an analyst at Aite Group.
A win-win situation
That’s not to suggest, however, that credit card companies aren’t making their own gains using predictive analytics. By feeding predictive models with data as precise as married men aged 30 to 35 living in an affluent neighborhood, or geospatial information, such as women who live within 30 miles of an Ivy League university, today’s financial institutions are better targeting frequent credit card users, reducing attrition rates and slashing delinquencies.
In the case of Premier Bankcard, the passing of the Credit CARD Act of 2009 places a cap on the first year fees it can charge customers in high-fee cards — a risky proposition given that Premier relied on those fees to offset high-risk consumers who qualify for subprime unsecured credit cards. Using predictive analytics, Premier Bankcard is now able to protect its bottom line by flagging delinquent accounts before they spiral out of control.
More and more credit card marketers are using predictive modeling to determine not only the likelihood of someone responding [to a credit card offer], but what’s the best channel through which to reach a particular prospect.
|— Ron Shevlin|
“In the past, we managed our risks by charging higher fees on the front end,” says Pruitt. “With predictive analytics, we’re able to identify pricing scenarios that will still deliver a return.”
No wonder research firm IDC predicts that today’s $1.4 billion market for advanced analytics, which includes predictive analytics, will grow 10 percent annually through 2011.
Predictions for lost privacy
There’s a price to be paid, however, for predictive analytics’ promises of better customer service, product discounts and targeted marketing campaigns. Some credit card holders grumble that the collection and crunching of information, such as geospatial data and marital status, verges on privacy infringement.
“If your bank wants to use all the information it has about you, you don’t really have much privacy,” says Tom Davenport, a professor of IT at Babson College in Massachusetts and a research director with the International Institute for Analytics. “They know where you spend your money, and they even have a pretty good idea of what your overall assets are.”
In fact, because of the many advantages banks derive from the collection of such confidential information, few are even willing to discuss their use of predictive analytics. Banks are “guarded around the impact of their [predictive analytics] results, particularly when they are substantial,” says Eric A. King, president of The Modeling Agency, a Pittsburgh-based predictive modeling and business analytics consultancy. More than there simply being “heightened security and sensitivity over data privacy,” King says that when it comes to predictive analytics, banks exhibit “fierce protection of this competitive advantage.”
In the end though, Davenport says there’s not much consumers can do about banks’ number crunching of confidential data other than embrace their tailor-made special promotions and interest rate cuts. Want to keep your purchasing details private? Then you’ll have to forsake the convenience of credit and use cash instead.