10 new data plays for consumer screening

As we slowly emerge from the economic doldrums, there are dozens of these new data types available that can aid banks in surprising ways. In fact, this may be the most exciting era for applied analytics in decades. Because many bankers are unaware of the amazing progress made recently, we’ve assembled a few of the most impressive innovations here. You won’t likely need all of them, but you may find them interesting.  Better yet, you might discover there’s an attribute for your bank.

1. Is this a real person and is their ID genuine ID?

Checking IDs is nothing new. But technology is available today that can scan a person's ID with a mobile phone to verify that they are who they say they are and that their ID is valid. In addition to detecting signs of tampering, some systems can match the photograph to the state database to compare to the person applying for the account and ensure that the license is real.

Another option is checking data sources, like Homeland Security databases, to verify that the info on the ID matches what the actual ID should show. Not only does the technology combat fraud, but scanning the ID allows the application to be prefilled to create a seamless customer experience.

2. How stable is this consumer?

Let's say you have a group of consumers just on the fringe of your credit score parameters. If you needed one more piece of information that could verify their risk, stability would be a good choice.

Data that shows address quality and velocity is highly predictive of risk. The more often someone changes addresses the less likely they are to have a stable job and consistent income. On the flip side, if someone is moving from a low-rent apartment to buy a home in a more affluent neighborhood, they may present a better risk.

3. How does this person use their credit cards?

When a lender pulls a standard credit report, they see only a snapshot in time. Imagine the difference in assessing risk if you could see the same attributes over time and identify patterns. Trend data allows you to do just that—track balances and velocity over time.

TransUnion completed analyses of new consumer auto and credit card accounts that shows an interesting dynamic. The credit bureau discovered that consumers who were revolving on their existing credit card accounts in the month prior to opening a new account are two to three times more likely to become delinquent than consumers who were previously transacting on their credit cards (paying the full balance on all cards). This example verifies the highly indicative nature of trend data and the value it can bring.

4. I only get three questions on a mobile app, so how do I answer everything else?

This question that puts digital marketers at odds with credit risk. The good news is that both sides’ demands can be satisfied.

To create the best user experience on mobile devices you need to ask the fewest questions possible—the optimal number is between three and five before you lose a consumer. How can the credit risk team possibly make a good credit decision based on such limited information? If you can capture personally identifiable information such as a name, birth date, and Social Security number through the mobile app, the rest of the information can be populated through other data sources.

If you have a 40% dropout rate because your mobile app is too complex and you reduce that loss rate to 10% by simplifying your process, the extra data you pull is going to pay for itself.

The lesson here is that you can have your cake and eat it too—a simple, elegant front-end for the user and a complex back-end process for the credit risk group.

5. What price would this consumer be willing to pay?

Price optimization is a profitability play for banks. Using advanced analytics you have the ability to determine how much a customer would be willing to pay for a particular product and offer each person the highest price possible—based on the statistically likelihood of acceptance.

Knowing who prefers convenience and rewards over price helps determine the ideal offer to make. The key to this modeling is the ability to create a personal optimization model, rather than using a generic model applied across an entire segment. The financial benefits of optimization-of-one are huge. Lenders can achieve 10-20 basis points of additional profitability in their deposit and lending portfolios. For every billion dollars in the portfolio, that adds a million in profit.

6. Is this person really who they say they are?

Geolocation technology can be used to track an IP address or a cell phone and locate the user within a specific longitude and latitude. This minimizes the risk of third-party fraud if somebody is trying to take over an already established account or open new accounts in another person’s name.

Mobile devices can be narrowed down to a few hundred feet. If an application comes in for Joe Cash in California, but the IP address is located in South Africa, it is probably not really Joe. This works well for minimizing fraudulent online account opening, but works in a branch or retail location as well. Some vendors can send you the account information of the cell phone owner as well, shortening the application process.

7. How do you evaluate risk for people without established credit?

Whether they are classified as unbanked, underbanked, or thin file consumers, the question remains the same: How will you evaluate whether a person will pay without a sufficient credit report?

The first type of accounts that most people establish today are cell phones and utilities. Steadily paying these bills can be a good indicator they someone will pay an auto loan or credit card. Public records, professional licenses, and education history can also be applied to make a more informed credit decision. The first mover advantage will go to those providers who identify the low-risk, high-reward individuals with non-established credit before their competitors do, and turn them into loyal customers. Multiple providers now offer highly predictive models for thin-file consumers.

8. How much wealth does this customer have?

Back in simpler times, a banker might have known net worth and used that information to inform their credit decisions. Today, assets are broadly distributed among institutions and there are no central sources for deposit accounts. It is difficult to understand someone’s lifestyle and their capacity to repay.

New attributes actually answer this question with good predictability. This allows lenders to identify the products a customer is likely to accept—a high-end car loan or premium credit card. This information also serves to improve ability to pay verification by giving insight into a consumer’s assets.

9. What are new demographic trends?

Social media data is hot, but nobody really knows what to do with it yet. It's too early to utilize it for assessing credit risk and not just from a regulatory standpoint. Some alternative lenders are using it for identity verification, cross referencing LinkedIn and Facebook to verify employment information, or to assess small business risk through negative reviews.

However, the most feasible scenario right now is to use the data for building targeted marketing campaigns and segmenting consumers. Posting about life events can help identify the right products to market. If someone is talking about weddings, they might be interested in a personal loan and maybe buying a home soon. High school graduation photos might mean a student loan would be of interest. Someone travels a lot, then offer a mileage rewards card.

This is very new territory, but the interest is there and several startups are actively pursuing ways to tap into its still unknown value.  

10. What type of DDA risk does this person pose?

There are people who simply can't manage their money well and others who open DDA accounts intending to commit fraud and then disappear. Both are costly, but require very different analytics to identify.

Most banks rely on historical data for their fraud evaluation, but financial institutions are increasingly adding predictive data to the mix.

A top-10 bank recently completed a head-to-head comparison of historic and predictive fraud for DDA and found only a 40% overlap of populations identified. The additional cost to add this new analysis was a fraction of the fraud expense eliminated. Even for underbanked populations, information from cellphone and utility providers can provide rich insight into customer behaviors.

There you have it. Ten ways banks can make more predictive, lower-risk decisions during account origination. There are new data types available for nearly every part of the bank.

The good news is that these emerging standards are keeping up with a changing world. Are you?

About the author

Eric Lindeen is marketing director for Zoot Enterprises, a global provider of advanced loan origination, account acquisition, and credit risk management solutions located in Bozeman, Mont.