Banks are rapidly adopting AI, but most are still struggling to prove what the technology is actually delivering to the bottom line.
Research from nCino’s AI in Banking Benchmark found that 91% of banks have an AI strategy in place, while 71% track clear key performance indicators (KPIs). However, only 21% measure whether AI is contributing to increased revenue. The findings highlight a growing gap between AI adoption and the ability to demonstrate a financial return.
The disconnect comes as 81% of banking executives say their organisations prioritise AI adoption over return on investment. This suggests banks have made significant progress in deploying AI and monitoring its use, but many have yet to establish how those deployments translate into measurable commercial outcomes.
The benchmark groups AI KPIs into three layers: Activity, Capability and Outcome. Activity measures what AI is doing, Capability measures how it improves existing processes, while Outcome focuses on the business results generated by those improvements.
Employee productivity and workflow automation are the most widely tracked measures, with 45% of banks monitoring them. By comparison, only 29% track cross-sell, 26% measure cost reduction and 21% track increased revenue.
This creates a measurement gap for banks. Activity and Capability metrics can show that AI is being adopted and improving processes, but Outcome metrics are needed to establish whether those improvements are creating financial value.
The challenge is not simply a lack of KPIs. The research points towards data and technology architecture as a major barrier to connecting AI deployments with business results.
Deloitte’s State of AI in the Financial Services Industry survey of more than 570 financial services leaders found that 84% of firms have not yet redesigned their workflows around AI. Only 18% are currently generating revenue from AI, despite 75% expecting to do so.
Siloed data adds to the problem. nCino found that 52% of banking leaders identify siloed data as their biggest data governance challenge.
This can make it difficult to trace a result from one part of a bank back to an AI capability elsewhere. A faster loan origination process, for example, could ultimately contribute to additional product sales, but those outcomes may sit across separate systems and teams.
Banks therefore need to move beyond measuring individual AI use cases and develop a more connected view of how deployments affect the wider business.
The Activity layer provides the most immediate evidence of AI adoption. Metrics such as deployment numbers, workflow automation and employee productivity show whether AI is becoming embedded into everyday operations.
Capability metrics take the analysis further by measuring improvements to existing processes. Fraud detection, customer experience and loan origination speed can demonstrate that AI is making a particular function faster or more effective.
Outcome metrics represent the final stage. Revenue growth, cross-selling and cost reduction provide a direct link between AI investment and commercial performance.
However, each layer requires more sophisticated measurement. Activity can often be captured within the AI system itself, while Capability requires comparisons against existing processes. Outcome measurement requires banks to connect AI activity to results that may occur in entirely different systems.
The findings suggest the next phase of banking AI investment will focus less on proving that the technology works and more on proving what it is worth.
Banks have already established much of the infrastructure needed to track adoption and process improvements. The challenge now is connecting those metrics to revenue, cost savings and other measurable outcomes.
nCino’s research suggests this could become a defining priority as banks move from AI experimentation towards demonstrating a return on increasingly significant technology investments.
Copyright © 2026 FinTech Global
