Banks spent $650 billion on technology last year — roughly the equivalent of the gross domestic product of Belgium or Sweden — but have little to show for it, according to a report McKinsey released Wednesday.
“Banks have been investing in technology, but for different reasons, they haven’t been able to, at least at the industry level, monetize the outcomes of all these investments,” said Xavier Lhuer, McKinsey partner, in an interview.
Not everyone takes such a dire view, and of course banks have to invest in technology in order to function. But most analysts and industry observers agree that banks need to be more transparent about tech spending and show more of a return on it.
“The marriage of banking and tech has been powerful over the long term,” said Mike Mayo, managing director and head of U.S. large-cap bank research at Wells Fargo Securities, in an interview. Mayo
Banks have shifted customers from branches and tellers to ATMs, call centers and digital banking, he pointed out. The largest U.S. banks now have 150 million digital banking customers, about half the adult population. Bank of America’s Zelle transactions have doubled over the past three years. New tools have enabled better credit underwriting, anti-money laundering and cybersecurity protection.
Yet despite all this, banks’ average expense-to-revenue ratio is still 60%, Mayo said.
“As investors, when do we see this drop to the bottom line?” Mayo said. “I’m in the camp of, enough already. Tech firms have their big promises. Let’s see these reflected in the expense-to-revenue ratios for banks.”
Banks’ boards and
“There is this general view that the technology spending is opaque and often the value enabled is hard to quantify and so that increases skepticism,” Lhuer said. “Very few banks are actually able to articulate how much is spent on technology and what they’re getting from it.”
What banks get wrong about technology purchasing
Global technology spending in banking has been increasing 9% a year on average, outpacing revenue growth of 4%, according to the
The reasons for this apparent disconnect include declining productivity, unclear competitive differentiation and increasing cost of complexity, according to the report.
Some financial institutions are in what McKinsey refers to as a “negative loop”: “they have limited discretionary capacity for tech spending but determine they need to build certain solutions themselves, often because vendors’ offerings don’t meet their needs,” the report said. They end up doing small projects “whose returns are often unclear.”
In Mayo’s view, banks don’t reap as much benefit from technology as they should because when they modernize, they tend to automate bad processes.
“Most banks still advertise for COBOL programmers,” Mayo said. “COBOL is a programming language that first appeared over half a century ago in 1959. That’s just one data point that shows that modernization still has a way to go. The future question is, when do banks take the new technology from pilot to production to profitability?”
Ryan Favro, a former Capco consultant who recently started an AI company, Graivy, pointed out that sometimes companies have to invest in tech, even if returns won’t come for some time.
“If their spending was higher than their revenue ratio, I would ask, Is there a strategy behind that?” he said in an interview. “Meaning, are we testing something that we know is going to pay off a few quarters away?”
One mistake banks make, in Favro’s view, is they take on the most complex and toughest problems first, assuming they will yield the highest returns.
“Unfortunately, this approach frequently leads to delays and project failures, as high-complexity initiatives are difficult to execute, especially within legacy system constraints,” Favro said. They risk getting stuck in proof-of-concept cycles that don’t produce meaningful results, wasting time and resources, he noted.
“By the time they recognize the missteps and inefficiencies they’ve created, they’ve often exhausted their budgets, leaving little room for course correction or further innovation,” Favro said.
They also tend to want to build their own AI models, leading to unnecessary expenditure and slow adoption, in lieu of leveraging billions of dollars already invested by tech giants like Microsoft, OpenAI, Google and AWS, Favro said.
AI spending
Spending on AI projects, especially generative AI, is the latest focus of consultants’ and
“I want to see concrete use cases,” Mayo said. “In this case, talk has been anything but cheap for the banking industry. Where is the use case that gets customers, employees and investors excited in the way that big tech product releases get their stakeholders excited?”
Teresa Heitsenrether, chief data and analytics officer at JPMorgan Chase, who oversees AI initiatives across the bank and has been rolling out a generative AI portal to all employees, acknowledged that a lot of people are asking this question.
“It’s the question: Where is the commercial value?” she said during an AI panel on Tuesday at the Most Powerful Women in Banking conference in New York. “There are people who are disappointed.”
She looks at this as a three-phase journey.
“The initial phase is just having this thing on your desk and making you more effective for an hour or five hours a day,” Heitsenrether said. “It adds up, but it’s very hard to quantify because it’s a small portion of a lot of people’s time in a day.”
The next phase is supplementing generative AI with proprietary bank data. So in JPMorgan Chase’s call centers, for instance, where contact center reps serve 80 million households, “if they can answer the question more quickly, if it’s a better client experience, if they have access to information across all of those products, that’s a real savings for us,” she said. “Every second [saved] on those calls is actual real bottom line impact.”
Achieving this takes a lot of work, however, because product policy documents and brochures aren’t written for this purpose. “It’s a question of now adapting all of that knowledge so that it can be useful and making sure that it’s super accurate and it’s up to date.”
The third phase is
“The models are getting better at reasoning, they can do multiple steps,” she said. “You can think of the tool as being this capable analyst who can do a lot of work for you, as long as you explain to the analyst these are the steps that you should take to get something done, that’s where we think you’re going to see a lot more productivity.”
Lhuer agrees that the concept of agentic AI “has a lot of promise,” Lhuer said. But in his view, the negative loop concept still applies.
“If banks are not rewiring to unlock value from their tech investments, be they in AI agents or in traditional AI automation, they will still fall into the same negative growth,” he said, By “rewiring to unlock value from tech investments,” Lhuer means making sure that tech dollars are applied in a few areas where outsize value can be enabled.
Instead of investing in dozens of pilots, banks would be better off picking a few and focusing on them. For instance, they could use generative AI to make developers more productive (several banks, including Citi and Goldman Sachs, do this today).
With AI projects, “the key to success is to bring the chief financial officer, the chief information officer, the chief risk officer and the business executives to the table to define how their business is going to get transformed and then commit to realize the benefits,” Lhuer said. “That’s what banks are often grappling with, how do you drive that cross-functional collaboration because you need all those parties to come together to be able to realize the value.”
Favro recommends that banks focus on low-complexity, high-value initiatives to show quick wins and demonstrate the potential of new technology. “This can build internal confidence and gradually reduce resistance,” he said.
The McKinsey report offers several overall recommendations to banks on tech spending. One is that they should increase their discretionary spending budgets by 50% or more by increasing engineering productivity and “optimizing” run-the-bank spending. Another is that they should figure out the objectives and key results for each project. A third is to have quarterly outcome-based reviews of those objectives and key results. Fourth, they should be able to communicate the value of their tech investments to investors.