The ROI of AI Conundrum // What's Going On In Banking

Hey everybody. Ron Shevlin here, chief research officer at Cornerstone Advisors and author of the FinTech Snark Tank on Forbes. Welcome back to another episode of What’s Going On in Banking, which is kind of a joke because it has been about a year since the last episode.

I probably owe people an explanation, although you may not have even missed it. My marketing team did, and they’ve been giving me grief for months, if not a full year, about bringing the podcast back.

I’ll be honest: I didn’t like doing it. I didn’t enjoy it, and that’s why we stopped.

When we started this podcast, I was adamant that it would not become another talk-show format where people sit around and say, “Hey, what’s going on? What’s happening in your world?” I wanted it to be hard-hitting, timely, forceful, and focused on the truth of what’s going on in the industry.

So we leaned toward newsworthy developments. Something would happen in banking, and we’d respond to it. The problem is that there’s a lot going on in banking, and not all of it deserves a 15- or 20-minute treatment.

I also found that even though I had some amazing guests, including Nigel Morris, co-founder of Capital One and managing partner at QED Investors, the show started to feel too much like an interview. I didn’t always feel comfortable looking at a guest and saying, “You’re wrong. You’re completely wrong about that.”

Then we thought, “Okay, Ron, why don’t you just do it yourself?” I could easily talk for 15 or 20 minutes by myself, but I didn’t want that format either, and I’m not convinced anyone wants to listen to me monologue for that long.

What I needed was someone who met a couple of criteria. First, that person had to have strong opinions about the industry and what’s going on. Second, they had to have the personality to be forceful, take me telling them they’re wrong, and be willing to look at me and say, “Ron, you don’t know what you’re talking about. Let me set you straight.”

There was really only one person at Cornerstone, or frankly in the industry, who I thought could do that: my colleague Stacey Bryant.

Stacey, thank you very much for being the co-host. Welcome to What’s Going On in Banking. I’m excited to relaunch the show with a new format and have you here.

I’m excited to join you, Ron. I never thought a girl from Brooklyn, New York, would make it all the way to co-hosting with you. I’m honored to be part of a team that gets to be nerdy about the industry while also being spunky and opinionated. It is absolutely my pleasure.

In the true spirit of What’s Going On in Banking, we’re recording this in early May 2025. I also want everybody to know that we’re recording on my 10th anniversary at Cornerstone. I’ve been here 10 years today.

I think the only reason I’ve made it this long is because I work with a team of people who tolerate me. I know how difficult I am to work with, so I’m genuinely grateful to work with all of you.

So, what’s going on in banking?

The first thing I think we should talk about is the FIS-Worldpay deal announced a couple of weeks ago. Here’s the background.

This is almost like a complicated asset swap. FIS is selling its stake in Worldpay to Global Payments while acquiring Global Payments’ issuer-solutions business. That fundamentally reorients the strategic direction of everyone involved.

What I really want to focus on is the FIS side of the deal. FIS is effectively doubling down on the issuer and financial-institution market. When it acquired Worldpay a few years ago, that marked a strategic move into the merchant side of payments. I think a lot of financial institutions worried that they were becoming second-class citizens in the FIS portfolio.

This deal now pushes FIS further into credit-card processing by bringing in what used to be TSYS. It reflects an important strategic shift.

I’ve read a lot about the deal, including what some of our colleagues have written, and there seems to be a consensus that this is good news for financial institutions because FIS will be focusing more heavily on them.

Maybe that’s true, but I still have some unanswered questions and concerns.

The first question is simple: what is FIS betting on in the credit-card market that the rest of us don’t know?

Then there are the economics. FIS has said it expects roughly $150 million in net EBITDA gains over three years, along with significant revenue synergies. Clearly, it sees considerable operational efficiencies and cross-selling opportunities.

But I wonder about the pricing assumptions behind those projections. If you’re already a large debit-processing customer of FIS and they come to you saying, “We can now handle your credit card business too,” you’re probably going to ask for concessions. That could negatively affect the economics of the cross-sell.

Second, doubling down on issuer processing could expose FIS to regulatory changes affecting the card networks and interchange fees. There’s also continued consolidation among large financial institutions, which could put pressure on the business. More broadly, reduced transaction volumes tied to economic conditions could also matter.

Third, FIS is taking on roughly $8 billion in debt to do this deal. I wonder whether that could restrict capital available for other technology investments, expose the company to interest-rate fluctuations, or simply reduce financial flexibility if the economy weakens.

Those are some of my concerns. What was your take?

First, Ron, I agree that the FIS, Global Payments, and GTCR arrangement is a very complex asset swap. Your “wife swap” analogy is not entirely wrong.

To me, a lot of this feels driven by scale and headlines. You asked what they may know that the rest of us don’t. The biggest thing I see is scale that could help them outpace competitors such as PayPal and Stripe. That creates a much more dominant position in global payments.

Outside of that, I see integration risk. Worldpay has gone through multiple ownership changes, and the results from past acquisitions have been mixed.

From a community bank perspective, the broader product suite may become more attractive, but I still question how well they’ll be able to integrate and scale all of this. When we think about return on investment, I’m not convinced the story is as simple as the headlines suggest.

So I think we agree there are risks here. It may not be as rosy as some of the early commentary suggests.

Maybe this ultimately is good news for financial institutions, but mergers and acquisitions of this size are always difficult. I also find it interesting that FIS is stepping back from the merchant market. From a diversification and revenue perspective, that’s a meaningful decision.

The merchant community has been trying for years to reduce its dependence on the traditional card networks. I’m not sure this transaction changes that dynamic. Maybe the processing opportunity remains regardless of which network handles the transaction, but it still feels like FIS is making some important bets on where the market is going.

The other question is how competitors respond. Stripe, PayPal, challenger banks, and fintechs are not going to stand still. FIS may gain scale, but we still have to see how that scale translates into execution.

I also think about cross-selling opportunities in treasury and embedded finance, but I keep coming back to integration. I’d be very interested to have this conversation again in a year.

That’s a good point because they’re projecting meaningful impact within a three-year timeframe. If the organization spends 12 to 18 months just absorbing the deal, that cuts significantly into the period they’re using to justify those projections.

I understand why companies do this. They have to communicate a strong story to Wall Street and investors. But I think we both share some concerns.

Let’s move on.

Another topic I wanted to call your attention to was a Wall Street Journal article last week titled “Companies Are Struggling to Drive a Return on Artificial Intelligence.” The article argued that AI adoption is extremely high, but many companies are struggling to put it to productive use. They believe AI is essential to their future, but they aren’t sure how to unlock its value.

What caught my attention is that I don’t think companies are necessarily struggling to drive a return on AI. I think they’re struggling to measure it.

That’s going to be a huge challenge for financial institutions, especially midsize banks and credit unions, for a few reasons.

First, many institutions don’t have baseline measurements to compare against. A lot of the impact of AI today is productivity-related, whether that’s contact-center work, loan processing, underwriting, or something similar. If you don’t already know how long those processes take or what they cost, it becomes difficult to prove how much improvement the technology created.

Second, productivity improvement doesn’t always create an immediate financial impact. Unless you’re reducing headcount or avoiding future hiring, you may not see a clear dollar amount tied to the improvement.

Third, I’m not sure every AI investment is supposed to have a direct ROI. Infrastructure investments often enable something else that eventually produces the return. A lot of what institutions are doing today with AI looks more like infrastructure: creating new data, processing, and decisioning capabilities that other applications can build on later.

There’s also a personal-productivity side. I tell institutions to think about AI in two categories: personal AI and enterprise AI.

Personal AI includes things that help automate email, scheduling, presentations, spreadsheets, and other individual work that used to take much longer. Enterprise AI includes systems embedded in business processes and customer-facing operations.

So I don’t think the main struggle is necessarily generating a return. I think it’s measuring the return.

I agree with you. I was actually looking at your article, and you referenced the old joke from the early days of social media. When someone asked, “What’s the ROI of social media?” advocates would respond, “What’s the ROI of your mother?”

That line has been around forever, but it still makes the point.

The bigger issue is that AI is incredibly broad. In the latest What’s Going On in Banking report, you broke it down into conversational AI, generative AI, machine learning, robotic process automation, and other categories.

When I’m out speaking with credit union CEOs, CIOs, community bank executives, and other leaders, I ask the same question: “What are you doing with AI?”

Some organizations look at what larger institutions are doing and jump on the same bandwagon. Others are actually talking to their team leads, identifying pain points, and asking what objective they are trying to solve.

Before talking about ROI or jumping on somebody else’s AI initiative, I think institutions need to test, learn, and scale.

Pick one area. Maybe it’s the call center. Maybe it’s payment collections. Maybe it’s a chatbot. Maybe it’s internal presentation creation. Focus on a defined use case, learn how it performs, and then decide whether to expand it.

It’s difficult to copy what works at another bank if the business models are different. A commercial-heavy bank shouldn’t automatically assume the same AI priorities as a bank focused heavily on direct-to-consumer business.

I think that makes sense.

Another thing that bothers me about this conversation is terminology. About a year ago on LinkedIn, I said I wanted to stop using “AI” as an umbrella term because it hides the differences between technologies.

We need to distinguish machine learning, conversational AI, generative AI, agentic AI, and robotic process automation.

What’s funny is that I sometimes get pushback from people who say robotic process automation isn’t really AI because it’s rules-based. Then someone else will tell me machine learning isn’t really AI because it’s “just models.”

At that point, you have to ask what they think AI actually means, and too often the answer is essentially “generative AI.”

That’s a definition problem. Organizations can’t manage technology effectively if they aren’t clear about the terms they’re using.

I don’t care what terminology an institution chooses internally, but it needs to be specific. If a vendor comes in and says, “We have AI embedded in our platform and it creates a 137% ROI,” the first question should be, “What kind of AI? Show me exactly what it is doing.” If they can’t answer that, that’s a problem.

That’s a good segue into data.

You recently released a commissioned paper on Data IQ, and it reminded me of the conversations I’m having with institutions of all sizes. Whether they have $400 million, $1 billion, or $5 billion in assets, a common question is: “We have all this data. What do we do with it?”

How do we become a smarter bank or a smarter credit union? How do we actually use the data we already have?

The problem is that many institutions also have a lot of poor-quality data. You start with data strategy, then data governance, then warehousing, and suddenly the work becomes much larger.

I’m working with one institution in the Midwest that is supportive of developing a data strategy. But when that conversation leads to data warehousing, they get nervous because many modern data platforms are cloud-based. Their board is extremely conservative and wants to keep things on-premises.

At the same time, even core providers such as Jack Henry and Fiserv are increasingly moving parts of their modern banking platforms into the cloud. So how do we help boards and other influencers understand that refusing cloud technology entirely may actually limit the institution’s ability to innovate?

It’s a mess.

Too many senior management teams still treat data as an IT problem. They delegate it to technology and say, “Figure it out.” Maybe they’ve created a chief data officer role, but sometimes that just means another executive title without the budget or team needed to make anything happen.

I think the problem starts even earlier. Unless you have strategic clarity about where the business is going, what capabilities you need, and which products or services you’re trying to build, data prioritization becomes nearly impossible.

Everybody agrees the institution needs better data governance and strategy, but nobody knows where to start because the business hasn’t decided what matters most.

Strategic direction gives you that prioritization. It lets you say, “Yes, Stacey, your business unit is important, but this is the problem we have to solve first.”

That’s difficult for community financial institutions because they’re dealing with regulatory pressure, technology changes, competition, generational shifts, and new customer touchpoints all at once.

I joke that if I had a nickel for every consultant who said “data is important,” I’d be rich. Then I realized I’ve said it enough times myself that I’d be rich from my own comments.

It’s a sticky problem because there are so many dependencies, but it becomes even more important as institutions adopt machine learning, generative AI, and other technologies.

We recently completed research on the impact of AI on productivity in banks and credit unions. We asked every institution we interviewed what they would have done differently if they could start over. Nearly everyone said they would have gotten their data in order earlier.

When we asked what that specifically meant, many of them still struggled to explain it. So there’s heightened awareness that data matters, but also uncertainty about what to do next.

The quality of the data is going to determine how useful these tools can be. I think the successful institutions over the next few years will be the ones that not only clean up their data, but also establish the strategic clarity needed to decide what should be fixed first.

Let’s make that a little less scary.

People who follow this conversation already understand that data and AI matter. But given everything else happening in the economy, how do we simplify the next step?

When people ask me where to start with data and efficiency, I usually bring it back to people, process, and technology. Start with a technology assessment, but also look at the people and processes around it.

I like that framework, but I think there’s another element: organizational structure.

When I look at some of the leading banks and credit unions that are focused on growth, I see common patterns. One is that they’re creating new products for very specific target markets.

Think about AlumniFi from Michigan State University Federal Credit Union, Hustle from Vantage West, or similar niche offerings. What characterizes those efforts is not simply a new product or a new piece of technology. The institution creates a dedicated group and says, “You’re the team that owns this.”

That team may still need internal resources, external partners, data, and technology support, but there is a clearly accountable organizational unit driving the initiative.

I think that organizational commitment is often the key step. It gives the institution a path toward new revenue, greater efficiency, or a strategic pivot instead of leaving the initiative spread loosely across the existing structure.

That reminds me of the importance of sticking to a niche or finding the right opportunity in your market and then really committing to it.

Exactly. For some institutions, the challenge isn’t just doubling down on a niche. It’s figuring out what the niche should be in the first place.

Stacey, I’m looking at the clock, and we promised to keep this short and sweet. We have a million other things to talk about, so we’ll save them for the next episode.

Thanks a lot for doing this. I’m already enjoying the new format, and I think this is going to be good.

For everybody listening, I hope you enjoy the relaunch. Please follow the show, share it, or whatever the appropriate podcast call to action is these days. Thanks for listening, and we’ll see you next time.