Uncovering Hidden Card Portfolio Data Secrets

Debit and credit card portfolios generate interchange income, but savvy credit unions know they also generate a digital gold mine of data about members. That data can detect more than just fraud, too – cut it the right way, and card portfolio data can reveal everything from regular shopping habits to which members most likely need a car loan or other product.

There’s just one problem, though: The amount of data available to credit unions nowadays is truly massive, and figuring out how to access it, organize it and tap into its insights can be intimidating and expensive.

That doesn’t have to be the case, though. Four card data experts offered a peek at some of the hidden data treasures inside those credit union card portfolios, plus a few ways that data can boost member relationships – and the bottom line.

Casting the Net

For most credit unions, the first step in the data-mining process – getting access to data – begins with their core processors. They typically provide a basic set of authorization data that’s used to allow or deny transactions, according to David Wallace, global financial services marketing manager at business analytics firm SAS.

That’s good information, but it’s not the only data out there. For example, credit unions can also pay processors for what Wallace called “enriched data,” which provides even more detail. Plus there’s a long list of other data providers. Credit unions can get enriched data from credit bureaus, for instance, which Wallace said supplies more information about what members do with their credit union cards, as well as details about other payments or consumer credit transactions.

Relatively speaking, finding data is the easy part. Figuring out what to do with it all is where the work often begins, Wallace noted. But it’s also where the magic happens.

“If data is the oil or the fuel, analytics is the engine,” he said. “The vehicle can’t move forward without the engine that’s really doing the processing of the data.”

Once credit unions link the processor information to their member IDs, they could link the members’ products and services with that core information to get a fuller picture of what members are doing and have been doing over time, he explained. Pair that with other information, such as data about utility bill payments or social media, and things really get interesting, he said.

The Art of Slicing and Dicing

Many credit unions may not have the tools, training or staff to analyze enormous amounts of data, but CO-OP Financial Services SVP of Fraud Products Fotis Konstantinidis said those things aren’t necessarily deal-killers.

“There are a number of companies and offerings that actually target the non-data scientists and nontechnical [people],” he said. “You drag and drop nice [user interface] components, and then you can come up with some quick conclusions about that data.”

Konstantinidis added, “One or two great people can lead that from a credit union standpoint in-house because the tools are there.”

Jim Patterson, a member of the Advisors team at Fair Isaac Corporation, which offers cloud-based data analytics programs to credit unions and other financial institutions, said every credit union should begin by creating macro-level portfolio trend metrics in order to understand card portfolio health and uncover growth opportunities.

“This generally includes metrics such as account and balance growth rates, active rates, account and balance attrition rates, average active and delinquent balances, delinquent account and balance ratios, average monthly purchase volume, average credit limit and limit utilization,” he noted.

Mary Du Pont, also a member of the Fair Isaac Advisors team, said many clients have used card portfolio data to improve authorizations.

“For example, 97% of those [creditors] we study use days delinquent, not cycles delinquent, in their authorization strategy. This allows slow payers to be approved for authorizations but not potential delinquents when used in combination with a behavior score, which is used by 85%,” she explained.

Lenders are also using data-driven strategies that allow authorizations of up to 120% of the limit for their best accounts, which protects member relationships, she added.

“Time on books” is another popular card data point, Du Pont said. “This allows the lender to take relationship value into consideration, knowing that older accounts in good condition generally behave better than newer accounts in good condition.”

Fraud prevention is another target for data analytics. “Bust-out” payment fraud (which occurs when people run up a credit card, pay the balance in full with a bad check and then run the card up again before the check bounces) is getting particular interest, Patterson noted.

“Lenders are faced with the delicate balance of protecting against this type of fraudulent behavior via payment holds restricting the available credit and ensuring that ‘good’ members are not adversely impacted,” he explained. “To be successful, credit unions must be adept at analyzing spend and payment characteristics to detect correlations with bust-out fraud. Just like other types of fraud, these traits tend to be a moving target, as fraudsters continually attempt to outwit card issuers.”

One of the most promising features of card data analytics, however, is its potential to unlock cross-sell opportunities. That can in turn make relationships with the members stickier and avoid spammy or costly mass-marketing campaigns.

“Inactivity or light usage by members often reveals a product-feature mismatch such as an inadequate or uncompetitive credit limit, pricing or product features,” Patterson explained. “Careful examination of spending and payment trends can help ensure that members get the right product for their needs – high-volume transactors naturally prefer rewards-based products, whereas revolvers tend to be rate-shoppers.”

The idea is to iterate toward accurate predictions of who’s looking for certain products, Wallace said. That makes offers to members – and even simple suggestions about ways to better manage their finances – far more appealing.

“The member sees that and they say, ‘Oh, you know, I’ve been thinking about that, but I haven’t told anybody.’ If you can get to that point, then you are communicating that you really do understand that individual as a member.”

It’s not an unusual idea, Konstantinidis said.

“Netflix, Amazon, Google – they understand certain trends about all of us. Each one of us, actually,” he said. “They make it very personalized, and in many cases they help us even understand ourselves, funny enough.”

Four Keys to Making Use of Card Portfolio Data

1. Choose technology only after you know what you want. “Don’t start from the technology. I would even dare to say disregard the technical component or even the company names and what they’re providing,” Konstantinidis said. “Start from what you’re trying to solve … test it with a small sample of your members, and then go forward and choose technology.”

2. Systematize the data-gathering. Consolidating data from several different sources can be hard, Du Pont noted, but regimenting that process early on can eliminate the heavy-lifting. “A best practice would be to define the analysis that should be performed on a regular basis, and standardize and document the process to make it repeatable,” she said.

3. Embrace anomalies in the data. “From a data science standpoint, what is more important is some spike – some event of the data that they call ‘outliers,’ if you will. They make me, as a data scientist, understand the member better,” Konstantinidis said. “The prediction power comes when you understand somebody fully as a person, and then you understand those spikes.”

4. Stop trying to be perfect. “We have seen credit unions spend years developing the perfect data warehouse, trying to use values that balance across many systems. Typically, the challenge is data from the financial systems versus other lending systems,” Du Pont explained. “Do not let the perfect be the enemy of the good.”