3
Rethinking return on investment

Banks are taking varied approaches to the persistent challenge of measuring ROI on AI.

Rethinking return on investment

Calculating the return on investment (ROI) of AI is difficult. Unlike traditional IT purchases and cost-savings initiatives, the benefits are sometimes not directly financial. Use cases can, for example, lead to lower risk, deeper insights or richer client experiences, and the financial impact of these can be hard to quantify.

Many banks also suspect that, as with the early days of online banking, immediate returns are overestimated but long-term returns are underestimated. A short-term return calculation might not capture the full benefits.

“When internet banking took off 20 years ago everyone thought there would be immediate cost savings through branch closures,” says Abhay Chauhan, EY Asia-Pacific Corporate, Commercial and SME Banking Leader. “But consumer habits only really began to change in the past five years.

The financial benefits materialized much later, and many believe history will repeat itself with AI.”

Complicating matters further, AI implementation is often accompanied by changes to workflows and processes, so isolating the impact of AI alone isn’t easy.

When internet banking took off 20 years ago everyone thought there would be immediate cost savings through branch closures. But consumer habits only really began to change in the past five years. The financial benefits materialized much later, and many believe history will repeat itself with AI.

Abhay Chauhan, EY Asia-Pacific Corporate, Commercial and SME Banking Leader

Faced with these challenges, banks have adopted a wide variety of approaches. At some banks, ROI is a crucial way to prioritize use cases. “ROI keeps AI honest,” says Lloyds Banking Group's Rowe. “If you can’t link it to business value, it’s just experimentation. If you can’t measure it, you’re probably not doing it right. It’s tricky, but you can use proxies to help.”

ROI keeps AI honest. If you can’t link it to business value, it’s just experimentation. If you can’t measure it, you’re probably not doing it right. It’s tricky, but you can use proxies to help.

Ulku Rowe, CIO, Commercial at Lloyds Banking Group

Other banks don’t measure ROI at all. “We just don’t measure ROI for AI investments,” says Bahadir Yilmaz, Chief Analytics Officer at ING. “We, of course, track project KPIs and OKRs, but ROI is difficult. We can measure the return, but the investment is often a joint effort encompassing many changes to processes alongside investments in data. For example, AI is enabling the digitalization of the loan processes. But we’re also changing the client journey and how we do client acquisition, so attributing improvements to AI is not possible.”

We just don’t measure ROI for AI investments. We, of course, track project KPIs and OKRs, but ROI is difficult. We can measure the return, but the investment is often a joint effort encompassing many changes to processes alongside investments in data. For example, AI is enabling the digitalization of the loan processes. But we’re also changing the client journey and how we do client acquisition, so attributing improvements to AI is not possible.

Bahadir Yilmaz, Chief Analytics Officer, ING

Even a rough estimate is better than nothing

Banks should not hold back on investing because of difficulties calculating ROI. At the very least, a crude assessment of ROI, perhaps using A/B testing, can help them to prioritize use cases. It’s also helpful to separate individual use cases and investments in building the foundational data, technology and human capabilities needed to scale AI. But investments in foundational capabilities must happen — with or without an ROI justification.

“ROI should not be ignored, even if it’s crudely estimated,” says James Sankey, EY EMEIA Corporate, Commercial and SME Banking Leader. “It helps you to go into this journey with your eyes wide open. Many assume the ROI will be positive, but when you take into account all of the technology costs, it may be negative in the first two years. It may be useful to ringfence some budget for experimentation and building capabilities without an ROI calculation, though.”

Useful AI-powered tools have emerged that help organizations to calculate ROI. By simulating future performance with AI-powered tools and comparing this with a baseline of performance, they can estimate ROI before scaling up deployment.

ROI should not be ignored, even if it's crudely estimated. It helps you to go into this journey with your eyes wide open. Many assume the ROI will be positive, but when you take into account all of the technology costs it may be negative in the first two years. It may be useful to ringfence some budget for experimentation and building capabilities without an ROI calculation, though.

James Sankey, EY EMEIA Corporate, Commercial and SME Banking Leader

Key actions

Always attempt to calculate the ROI of every AI use case.
Set aside budget to build the foundations. Don’t delay because of ROI frustrations.
Explore how AI might be able to help measure ROI.