Predictive Models For ACH Payments

Real-time access to data is important for many reasons but perhaps never more important than being able to make a quick — and accurate — decision about whether a customer applying for a loan is legitimate or fraudulent or about whether a borrower should be offered Loan A or Loan B based on their credit score.

Access to real-time data analytics can have the power to change the way these types of decisions are made, making it no surprise that businesses are seeking real-time decision-making capabilities for a variety of different use cases.

Enova Chief Analytics Officer Joe DeCosmo said the company’s Analytics-as-a-Service brand, Enova Decisions, powered by its real-time decision-making platform, Colossus, is designed to meet that need and help lenders use those insights and capabilities for customer acquisition, fraud risk assessments, customer operations or even payment processing.

“Having a robust, real-time platform to make these decisions is crucial. You can hire a statistician or a consultant to build a model, do one-time studies and batch scoring, but none of that gets to the experience that lenders really need today,” he explained.

Instant answers — that are also accurate.

DeCosmo said the demand to have answers decided upon instantaneously is more compelling for lenders than some of the more specific use cases available, such as credit risk, fraud risk, customer operations or fraud defense. Enova Decisions helps lenders to make decisions about everything from fraud to customer retention and growth, but DeCosmo emphasized that the main objective truly surrounds the real-time availability of the data for actionable decision-making.

With Analytics-as-a-Service, the platform supplies the analytics, runs the models that are built by the service and deploys everything in real time, enabling its clients to “make a decision, while the customer is interacting with you on your website, on the phone or in your call center,” DeCosmo explained, adding that this is the piece that resonates most with clients.

But, of course, having access to data that is convenient, useful and available in real time doesn’t come easily.

Real-time systems are hard to build, DeCosmo points out. It takes time, money and the right talent to create a solution that’s truly going to provide value. DeCosmo also noted that, in many cases, a financial institution or payment processor may have a homegrown solution but may have challenges getting it to scale or to support the range of models and data required. Enova Decisions can either be used to replace or add on to what a client already has in a current platform, giving companies the option not to invest heavily into building an in-house platform and hiring the analytics staff needed to run it, he added.

The demand for quicker access and faster information doesn’t just stop at data and analytics for lending; it’s a hot topic for payments as well.

Karen Webster pointed out in her discussion with DeCosmo that the momentum towards faster or real-time payments is ever-growing, with Phase One of Same-Day ACH capabilities rolling out in the fall. She wondered how this might impact, if at all, Enova Decisions’ ACH product offering.

DeCosmo explained that Enova’s real-time data and models provide the ability to know exactly when an ACH transaction needs to be presented, a decision that not only optimizes clearance but also minimizes the risk of customer impact.

“You can eliminate a lot of unnecessary debits, while, at the same time, improving your overall rate on the ACH transactions with our Smart ACH solution,” he said. “The application of predictive analytics is relatively new to the ACH and transaction processing world. We are at the forefront and understand the data and modeling required to achieve it.”

As use of advanced analytics for ACH transactions continues to mature, DeCosmo emphasized the importance of having a platform that can keep up with, and take advantage of, these advances in data.

“These days, we are seeing more machine learning and other types of advanced analytics models that are very powerful and predictive,” he stated. “They aren’t perfect in all situations or for all use cases, but they absolutely have a place in the world of predictive analytics.”

This is why DeCosmo noted the importance of constantly testing new algorithms and modeling methodologies, as well making sure Enova is utilizing the best and most effective tools.

He pointed out that having an effective transactional decision-making service is not a “one-and-done” type of operation; data and analytics models can’t just be built and forgotten about.

“You create the best model you can given the data you have at the start, and then, you go through a series of live testing to gather new data and recalibrate the model,” DeCosmo said. “It’s about iteratively improving it over time.”