Banks Turning to Predictive Analytics

In a recent BizTech article, journalist Nathan Eddy examines how banks are extending the use of predictive analytics beyond traditional fraud and credit applications into customer engagement, liquidity management, regulatory compliance, and increasingly automated decision-making. 

Shanker Ramamurthy, Managing Partner for Global Banking and Financial Markets at IBM Consulting, describes predictive analytics as “how a bank learns to see around corners,” allowing institutions to move “from managing by rearview mirror to managing by headlights.” Banks are increasingly using transaction histories, cash flow patterns, and behavioral data to develop more current assessments of customer risk, anticipate funding needs, help customers avoid financial distress, and detect potential compliance problems earlier. Forrester Principal Analyst Aurélie L’Hostis calls predictive analytics “one of the most mature and valuable applications of AI in financial services.”

The expansion raises significant governance questions. IDC Research Director Sam Abadir said banks can use predictive models to assess both the likelihood of a regulatory violation and the possibility that an internal control failure allows it to go undetected. Jerry Silva, Program Vice President at IDC Financial Insights, argues that “the single biggest challenge is not a technology issue; it’s the governance of AI, data, models and analytics at the enterprise level.” Effective use therefore depends on reliable data, explainability, monitoring, and the ability to justify consequential decisions.

Those demands are likely to intensify as predictive systems begin taking action themselves. Agentic AI could connect activities such as liquidity checks, compliance controls, and transaction execution, enabling what Ramamurthy calls “just-in-time finance,” where credit, liquidity, or advice arrives as a need emerges. Banks are expected to start with reversible actions, such as temporary fraud holds, while giving each agent a defined identity, scope of authority, and audit trail.

As systems become more autonomous, governance will also need to capture why an action was taken, not simply what happened. “Start with the decisions worth improving, not the technology worth buying,” Ramamurthy advises.