Fallen out of love with your credit card issuer? Chances are, the bank already knows about your wandering eye, thanks to an obscure credit scoring model.
Like credit scoring models, attrition-risk scores are used to predict a consumer's behavior, in this case the likelihood they could start a new relationship with a business' competition. "Attrition modeling is focused on which current customers are most likely to go elsewhere within a certain amount of time (three or six months is typical)," says Gregg Weldon, chief analytical officer with AnalyticsIQ Inc. in Atlanta, via e-mail.
That makes an attrition score an important tool for the payment card industry, as well as for the cell phone and insurance industries, where customer turnover is common. To prevent that turnover, businesses may try and retain customers using various perks or "deal sweeteners" that make for happy -- and committed -- consumers.
What's in an attrition model? Banks and other businesses use attrition models that combine internal corporate information and credit bureau data. Of those two data sources, the company's internal information is most important, Weldon says, since it answers key questions about the customer's behavior. Those questions may include:
- Is the customer currently visiting the store or using her credit card as much as in the past?
- Is the customer spending the same, more or less than in the past?
- Is the customer's card balance increasing or decreasing?
Some of the answers come directly from the credit bureaus, which maintain credit report information on recently opened credit lines and total revolving balances. That credit report data can raise red flags for companies, Weldon says, such as:
- The customer is suddenly doing business with a competitor.
- The cardholder's overall balances are increasing on competitor's cards, while the balance on the issuer's card is decreasing.
"These are warning signs that your customer may be getting wooed away by the competition," Weldon says.
One size model does not fit all Experts say that companies typically plug customer information into their own unique proprietary scoring models. Unlike credit scores, which are available to consumers, attrition scores aren't shared outside the businesses that use them. In fact, some companies won't even discuss whether they use attrition models at all. When contacted for this story, insurer State Farm, for example, said it wouldn't confirm or talk publicly about its use of such models, since that information is considered proprietary and could be potentially helpful to State Farm's competitors.
According to TransUnion, the first credit bureau to offer a generic attrition model in the mid-1990s, the reason why businesses use their own models is simple: Custom attrition models perform better.
However, TransUnion says attrition models are falling by the wayside in favor of models based on so-called triggers -- changes in a person's credit profile, such as a sudden credit inquiry or change of address -- which also indicate changes in borrowers' behavior. "Either an attrition model or a trigger does that -- it says a person is at risk," says Chet Wiermanski, group vice president of credit bureau TransUnion's analytics division. Triggers, however, do some things that attrition models do not. "In addition to just knowing a person is at risk, a triggers platform tells them what they need to do to retain that customer," Wiermanski says. When a low-risk cardholder applies for credit elsewhere, for example, the bank may respond by raising the cardholder's credit limit, lowering their interest rate or upgrading the account. Triggers also offer speed, giving real-time information on cardholder account changes so that a bank can react quickly.
Still, triggers have their weak points. For one thing, they focus on the short term rather than the long term, forcing an immediate response by the company. Additionally, some industries -- such as telecommunications and insurance -- don't have any triggers to guide them. Instead, they rely on attrition models to get an early idea of which customers are likely to be stolen by competitors, Weldon says.
What's in it for you? So what type of benefits might a customer reap when attrition models point to their planned exit? "Retention efforts may include discounts through the mail, offers for new or exclusive services, 'thank you' notes for past business or even happy birthday cards once a year. Anything that reminds the customer of the relationship they have with the company can work here," says Weldon.
Companies may go to great lengths to keep their best customers. In this tough economic period -- during which even the best credit card customers have seen credit limits slashed, APRs increased and accounts closed -- it may not seem that businesses always recognize just how valuable a great customer is. However, those who do can use an attrition-risk scoring model to get a heads up on when a good customer might be heading out the door.
"Smart companies use attrition models in good times and bad in order to keep the best customers with them and away from their competitors," Weldon says. "There are some companies that increase their use of attrition models in tough times, recognizing that they need every good customer they can get."
If a business views you as a lousy customer, -- such as if you've got bad credit and a history of late payments -- companies won't use attrition models on you. That's because they'd probably be happy to see you leave.
Good customers are a different story. With the high cost of finding new customers, attrition models offer businesses an attractive choice. Companies are forced to identify potential consumers, reach out to them via mail or telemarketing, for example, get them to respond to the offer and then approve them as a customer, Weldon explains. "These efforts may cost anywhere from hundreds to thousands of dollars per customer," he says, "so the idea that a competitor will come along and poach these customers drives some companies crazy."
The views and opinions expressed herein are the views and opinions of the author and do not necessarily reflect those of Nasdaq, Inc.