Credit Unions Battle for the Right Data to Fight Fraud

Watch more: Searching for Reliable Signals in Banking’s New Data Reality

Financial institutions, especially credit unions, are finding themselves in uncharted territory. The promise of artificial intelligence (AI) to unlock strategic insights is often juxtaposed against a backdrop of uncertainty regarding data provenance, bias and reliability.

As Jeremiah Lotz, senior vice president, enterprise data and experience design at Velera, told PYMNTS, at the same time, government agencies are facing pressure, defunding and the tweaking of statistical programs. 

That means that the traditional bedrock of economic data — historical series — is becoming a bit more limited and presents a challenge for AI applications designed to aid decision-making.

“Data is critical to nearly every decision a financial institution is making,” Lotz said in a recent interview with PYMNTS, part of the “What’s Next in Payments” series on data reliability in banking. “And access to that data and reliance on the accuracy of that data is important.” The shrinkage of data from the traditional source of the government agencies necessitates a strategic pivot for these institutions, he said.

Lotz maintained an optimistic outlook, stressing the ability of well-managed data strategies to compensate for statistical gaps.

“Those of us who are able to bring in first-party data, commercial data … [have] the ability to make up for that [shrinkage] with our financial institutions and give reliable signals from what’s happening within our own networks” and fill in any data gaps, he said. 

With the aid of Velera, a credit union service organization (CUSO), the data can be strategically leveraged, especially when augmented with carefully selected external sources. The direct insight into billions of transactions offers a granular view of consumer behavior and market trends.

Beyond proprietary data, institutions can significantly enhance their analytical capabilities by incorporating other information through third-party sources. “The third-party data sets that you bring in,” he cautioned, “have got to be vetted for accuracy and bias.”

He advocated for a multi-faceted approach, combining internal data with commercial data and consortium-based datasets. This holistic approach allows for the creation of proprietary indexes to forecast trends. Lotz further elaborated on the power of combining data types for strategic decision-making, including behavioral and geolocational data.

“My advice is not to be afraid of consortium data,” Lotz said, as Velera shapes that shared data across 4,000 credit unions (CUs). “Consumers know that [their financial institutions] have their data, and [that] they … have access to their data.” This general understanding among consumers creates an expectation for data interoperability to enhance their experience and security.

Discussing the application of AI with this data, Lotz underscored the critical role of governance. “You can take data and AI and you can make it smart and you can make it do great things, but it can’t just be smart,” Lotz said. “It literally has to be accountable for what it just did.”

The ultimate goal is to build trust in both the data feeding AI and the models themselves, ensuring they are ethical, bias-free and effective in spotting anomalies.

The effective use of data in financial services is not about choosing between historical and real-time information, but rather about their intelligent combination. Both types of data play distinct yet complementary roles in strategic decision-making and immediate operational needs, such as fraud prevention. Lotz articulated this dynamic: “Real-time data is certainly going to ‘win the moment’ in many cases, but historical data is going to ‘win’ the strategy, and you need both of them to compete” as open banking, instant payments and new use cases continue to emerge.

In fraud management, for instance, real-time transactional data is essential to understand immediate events, while historical or “retro” data provides crucial context by revealing patterns and outcomes of past models. “We want to use real-time data and real-time transactions to be able to understand what’s happening in that moment for that consumer,” Lotz said. “But we also want to look back at ‘retro’ data” to spot trends.” This combination allows AI models to inform immediate actions based on learned intelligence, optimizing responses in real-time fraud scenarios.

For Velera, this commitment to governance is paramount. Lotz simplified the core principle: “If you can’t explain where it came from or how it’s used,  or why it matters ultimately to the member, then you shouldn’t use it.”      

As he told PYMNTS, “shared data works when it’s secure, when you can measure it, when you [can] track it back from where it came and when you can provide some type of positive impact for everyone involved.”

The post Credit Unions Battle for the Right Data to Fight Fraud appeared first on PYMNTS.com.