top of page
  • Writer's pictureJohn San Filippo

CUs Can Make More Loans by Mastering Small Data

By John San Filippo


Credit unions have known for years that they possess a wealth of so-called “big data.” The struggle has always been in figuring out how to turn that big data into actionable initiatives. One increasingly common use case for big data involves using it to drive alternative loan approval methods that don’t rely on traditional credit scoring. (Finopotamus recently wrote about one such method.)


Jeff LoCastro

Now a company called Neener Analytics claims that the key to approving more loans is mastering what it calls “small data.” Specifically, the company says it can accurately decision a loan by programmatically interrogating a member’s social media accounts.


Finopotamus spoke with Neener Analytics co-founder and CEO Jeff LoCastro, a previous co-founder of Classmates.com, to learn more about this innovative approach to loan underwriting.


Big vs. Small Data


According to LoCastro, all big data models rely on aggregating information and grouping people by commonalities. This is in sharp contrast to bygone days when borrowers were evaluated solely on their own merits.

“Back in 1940, when my grandfather bought his first house, he walked into the bank and the banker said, ‘Hey, I know you. Good to see you. How have you been?’ That banker literally meant, ‘I know you,’” said LoCastro. “Now my other grandfather who didn't live in the same city or state, if he were to walk into that same bank, the banker would say, ‘I don't know you, but I know people like you.’ What a horrible thing to say to someone.”


LoCastro continued, “My second grandfather was big data – bins, aggregations, affinity groups lumped together to create a correlation that masquerades as something specific, but it's not. It's simply indicative of whatever bin you pluck that person from.


“My first grandfather was small data – this unique matrix for each and every consumer that manifests in binary outcomes. How was that first banker able to say to my grandfather, ‘I know you’? It was based on the conversations and communications he had with him over time. It wasn't based on him living in Chicago, or being a Cubs or a Bears fan, or what his favorite color was.”


LoCastro refers to aggregating people into different bins as “big data nonsense.” He noted that the current credit scoring models (e.g., FICO) represent the prime example. Each consumer has a credit score attached to them, the assumption being that anyone with a given credit score represents the same risk as anyone else with that same credit score, which is demonstrably not the case.


How It Works


“We don't even look at financial data,” LoCastro told Finopotamus. “We're the only social media-directed company in the world that Facebook allows to do what we do to gather the data like this, because we've cracked small data, because we do not have to bin or aggregate their users. [Aggregated data] is nowhere on our servers.” He said Neener Analytics’ software only looks at an applicant’s data; it does not compare that data to some larger group.


LoCastro doesn’t propose replacing a credit union’s underwriting process, only supplementing it. He added that one logical use is as part of the rejection flow. “The easiest place to engage initially is always going to be at a rejection flow,” he explained. “Hey, we had to reject you, but here's a second chance. They're taken to – let’s say we're using Facebook – they’re taken to the Facebook opt-in, they make that one-click opt-in and that's it. They're redirected back into the application and given some kind of a decision. Thank you. You've been approved.”


LoCastro made it clear that Neener Analytics’ platform is neither looking for nor judging and particular type of behavior. “No matter how hard we try to hide from our unique signature, our unique semantics, we can't do it,” he said. “That's what we're looking at. I had a lender ask, ‘What if I have a lot of pictures of me drinking, or I say I went to a party and I got drunk?’ None of that matters. We’re maybe listening to how you're talking about that party, not that you went to a party or what you’re saying about it.”



Because Neener Analytics also serves retailers with point-of-sale solutions, the platform must be very fast. “We get asked that all the time by retailers,” said LoCastro. “It adds half a second to the consumer experience. As part of a rejection flow for a lender, the decision may take 15 seconds.”


Easy Integration and Deployment


The Neener Analytics’ platform is application programming interface (API) driven, making it easy to integrate and deploy. “It couldn't be simpler,” said LoCastro. “Our documentation is literally three quarters of a page. There are only two API calls and we can reduce that to none, depending on how you deliver decisions to your members.


He added that all integration is through the credit union’s loan origination system (LOS), with no connection to the core data processing platform required. If the Neener Analytics solution is deployed as part of a rejection flow that runs outside the normal workflow, even that integration isn’t required.


More Inclusive


“The problem certainly in the U.S. is that 56% of the population alone is thin file, no file or credit challenged, which is just a crime,” said LoCastro. “The wealthiest country in the history of mankind, the ones who invented the bureau system, essentially can't decision or have trouble decisioning 60% of their population.”


He noted that one Neener Analytics customer in Brazil typically rejects about 95% of its applications. “The lowest rejection rates that you'll see are, are in the 50% range,” ne noted. “Do you really think that 50% of your applicants aren't going to pay? It's insane. This is a problem that everyone has. They've all got it. They're all saying no to too many people, and they have no idea who the good ones are.”

Comments


bottom of page