Data has become less about accumulation and more about judgment. The institutions that extract value from it are those that treat it as a living input to decisions rather than a static record.

That principle frames how Synchrony approaches credit, fraud and customer engagement. In a recent interview for the “What’s Next in Payments” series focused on the “data game,” Anita Chalkley, chief credit officer for the company’s home and auto platform, described a framework built on experimentation and disciplined measurement rather than fixed rules.

Chalkley placed emphasis on culture before technology. “What it comes down to is curiosity and then having a test-and-learn culture,” she told PYMNTS. That approach requires two conditions. Firms must invest in data and analytics, and they must maintain systems that can support rapid experimentation.

She pointed to the company’s internal platform as an example of how that culture operates in practice. The system can evaluate large volumes of inputs and run parallel tests in real time.

“The system that we have built is able to ingest thousands of attributes from multiple different data sources and make a decision in less than six seconds,” Chalkley said. It can also “test and learn across hundreds of tests simultaneously in real time.”

That structure shifts decision-making away from static scorecards toward continuous refinement. It also places pressure on governance, since rapid testing must still align with risk standards and regulatory expectations.

Matching Products to Customers

At the center of the strategy is a simple objective.

“Our entire mission is getting the right credit product to the right customer at the right time,” Chalkley said. Data serves as the mechanism for that alignment across the full customer lifecycle, from origination to ongoing engagement.

Underwriting thus becomes a sequence of decisions informed by changing signals. The same data that supports approvals also informs how institutions interact with customers and how they deliver value to merchant partners.

Testing extends beyond product design into the structure of decisioning itself. Chalkley described an environment where multiple hypotheses can be evaluated at once, allowing institutions to compare outcomes and refine models continuously.

This approach reduces reliance on single-variable decisions. It also introduces a more granular view of risk and opportunity, where outcomes are assessed across segments rather than averaged across broad populations.

Fraud prevention has become a parallel application of the same data discipline. Chalkley noted that adversaries operate with access to extensive information, which forces institutions to respond in kind.

“The fraudsters have a lot of data, right? So the only way to fight that is to use data,” she said.

The focus has shifted toward behavioral analysis. Institutions examine what constitutes typical activity at the individual level, drawing on device identifiers, transaction patterns and external data sources. “We use data to try and understand at an individual customer level or applicant level what is typical and what’s typical behavior,” Chalkley said.

At a broader level, aggregated data helps identify emerging threats. Unusual patterns can signal coordinated activity or new attack methods, allowing institutions to adjust controls without imposing unnecessary friction on legitimate users.

Alternative Data and Financial Inclusion

Chalkley pointed to cash-flow underwriting as an area of growing importance. Traditional credit bureau data reflects past borrowing behavior, but it does not fully capture how consumers manage income and expenses.

“By having cash flow data … you gain a lot more insights into customer behavior and how they manage the inflows and outflows of their money,” she said.

This broader view enables institutions to extend credit to individuals with limited or nonexistent credit histories. “It’s allowed us to underwrite people that were traditionally locked out of the system,” Chalkley said.

The use of alternative data introduces new considerations around data quality, privacy and model transparency, particularly as regulators examine how such inputs affect lending outcomes.

AI as an Accelerator, Not a Substitute

Artificial intelligence (AI) operates as an extension of these data practices rather than a replacement for them. Chalkley described a “human in the loop” approach that combines machine learning with oversight.

“We’ve been using AI for over 10 years in responsible machine learning decisions,” she said. Current applications range from fraud detection to marketing and customer service.

The primary advantage lies in speed and scale. “AI really unlocks the ability to use more data faster,” Chalkley said, noting its strength in pattern recognition and handling large data sets. That capability shortens the time between insight and action, which is central to maintaining relevance in credit and payments.

Measuring Success in the Data Game

Success, in Chalkley’s view, is measured across several dimensions. Data should improve decision quality, support a more seamless customer experience and generate value for partners. Each of these outcomes requires continuous measurement, reinforcing the importance of the test-and-learn framework.

Chalkley summarized the approach by returning to the underlying mindset, giving rise to “a data-driven innovative culture where the employees feel like they are making a difference … and really values curiosity and innovation,” she told PYMNTS.