How to sell the value of AI inside your financial institution

Artificial intelligence (AI) and machine learning (ML) can offer consumers better access to fairly priced loans, and better margins for banks through more accurate lending. When deployed properly, ML underwriting has clear external value. However, selling the value of AI internally is still a heavy lift.

At last month’s BankAI conference, Mike de Vere, Zest AI’s Chief Operating Officer, and Linda Duncombe, Executive Vice President at City National Bank, discussed these challenges — and how to overcome them. Here’s a recap of what they discussed.

Mike acknowledged that AI myths abound, such as that it’s too hard, too risky, and too expensive. Busting those myths requires heaps of education and patience.

The biggest obstacle, both agreed, is fear—of change, of data vulnerabilities, of lost jobs. De Vere told the story of one chief risk officer at a top bank who declared “there’s no world” where he’d “give over” decision-making to a computer because he felt it reminded him too much of the evil Skynet computer network from The Terminator films.

Duncombe emphasized the fact that banks need to prioritize protecting their client’s money and data. “It’s not negotiable,” she said. However, she conceded that too much trepidation about AI will come at a tremendous price: “The big firms we all work with—Accenture, Deloitte—are saying if you don’t take this on, you’re missing the opportunity to save trillions of dollars.”

De Vere lamented that IT departments can be especially intractable. “During one client visit, we had showed savings of north of $100 million,” he recalled. “The CEO was behind it. The business lines were behind it. Risk management was behind it. And we got compliance on board. Then we hit IT.” A big AI project wasn’t exactly on the strategic roadmap, so it got pushed out 18 months. “They were willing to forego that much shareholder value because they couldn’t act fast enough as an organization. Total disconnect.”

How, then, to rally all your troops around AI?

Start with a broadly compelling mission statement. ML has to be seen in terms of business outcomes, and more than just fancy algorithms. “The math is just math, and it’s been around a while,” said de Vere. AI must be sold internally based on its ability to reduce risk, accelerate loan growth, improve customer experience, and as a tool to make consumer credit markets fairer and more transparent. “We can lift the [loan] approval rate on average by 15% and decrease losses [charge-offs] by 30 percent, all while making sure the process is fair. We give organizations the ability to do good.”

Second, take cues from more innovative industries. Banks tend to focus on their immediate competitors—a self-limiting prophecy, warned Duncombe. “When you think about your target audience, they’re talking about experiences they’re having elsewhere, and those are enabled by technology like AI,” she said. “A lot of that tech isn’t new to our ecosystems, it’s just new to our industry. Being able to show where it’s worked previously can be quite powerful.”

De Vere expanded on that thought, highlighting AI’s potential to build stronger relationships with clients and even create new revenue streams. “Each of us has this mosaic of experiences. The beauty of machine learning is that you’re able to pull all of those data sources together to get a complete picture and deliver something unique.”

Third, expand new services slowly. If you suddenly try selling a slew of new products that customers aren’t used to getting from a bank, you might scare them off, warned Duncombe. Instead, she said, “pick the one or two biggest frustration points for clients, solve for that and build on it.” De Vere concurred: “If you build trust with customers, they’ll give you permission to sell them something else.”

Fourth, build consensus by bringing in all key players from the start. “Have the lawyers and compliance in the room” with technology providers, advised Duncombe. “It’s a much more productive conversation if they know the challenges you’re having from day one.” (Oh, and if you’re pretty sure a vendor’s project won’t happen, “tell them quickly,” she added. “Because it gets expensive to keep up with a big company’s procurement process.”)

Finally, urged de Vere, always look to raise your company’s overall “tech IQ.” Take a hard look at your hiring practices, too, and seriously consider candidates from other tech-savvy industries.

Selling the value of AI within large financial institutions may be slow-going today, but is highly likely to accelerate soon. Big banks tend to think their credit-scoring models work just fine, said de Vere, and competing priorities — or “big company muck” — will thwart progress. But smaller companies are getting it. “Watch out for the credit unions and the mid-sized banks,” said de Vere. “They’re coming.”

The industry has no choice but to embrace AI, said Duncombe. “I don’t see innovation slowing down,” she said. “Anyone who thinks they have time should really think about the longevity of their job. There is no time.”