The algorithm isn’t the advantage–the team is

Credit and financial-services organizations are using AI to reshape their offerings and interactions with consumers. But as Will Lansing, CEO of FICO, explains, mastering the technology is only one part of a broader transformation story for the company and the industry. “Leadership comes from followership,” he says. “It comes from people believing in the mission.” And that level of company-wide conviction is critical if executives want to solve some of AI’s hardest problems: explainability, fairness, and trust. On this episode of The McKinsey Podcast, Lansing speaks with McKinsey Senior Partner and North America Chair Eric Kutcher about what it will take to maintain FICO’s 70-year mission of enabling impartial lending, how the company can address emerging AI regulation, and how it can get the most from its data.

In this recurring series on The McKinsey Podcast, Kutcher speaks with top CEOs about the practice of leadership.

The McKinsey Podcast is regularly cohosted by Lucia Rahilly and Roberta Fusaro.

The following transcript has been edited for clarity and length.

To watch the full-length version of this interview, visit The McKinsey Podcast playlist on McKinsey’s YouTube channel.

Unbiased lending

Eric Kutcher: Tell us a little about where FICO started. What problems were its founders trying to solve, and how has the organization evolved into the FICO of today?

Will Lansing: Most people know FICO for FICO Scores. Back in the mid-1950s, some 70 years ago, a mathematician and an engineer got together and said, “Let’s use data, let’s apply analytics, let’s try to be more objective about this and come up with credit evaluation done in a more scientific way.”

For the first 30 or 40 years of FICO’s existence, we were essentially an analytics consulting firm. Banks would come to us and ask us to build proprietary scorecards so they could evaluate candidates for credit.

In 1987, we came up with the idea that instead of building scorecards for banks one by one, we’d build an industry-wide scorecard. One that anybody could use. So we launched the first industry-wide FICO Score, in partnership with Equifax, one of the credit bureaus.

I’d say that was transformational for financial services and the consumer credit industry in the United States. We were the infrastructure that let the consumer credit economy flourish. Before then, credit cards weren’t much of a factor. Revolving credit wasn’t really a thing.

Eric Kutcher: It’s amazing that it’s stood the test of time. It’s still enabling the economy to grow. Credit is so important to our consumer-led economy.

Will Lansing: The core question for banks or anybody dealing with consumer credit is, “If I give you money, are you going to pay me back?” If you could use only one data set to answer that question, you’d use the credit bureau’s credit payment history. You’d go into what’s essentially a ledger that says, “I pay my credit card bills on time. I don’t go beyond my limits.” Those are the kinds of things that inform the FICO Score.

Where AI can help—or harm

Eric Kutcher: In many ways, you’ve been an AI company long before people even started talking about artificial intelligence. How is AI affecting what you do?

Will Lansing: We’ve been an AI company for more than 20 years. What we’re really good at is predicting consumer behavior—that is, taking a lot of data and applying analytics to make a particular decision about how to interact with a consumer. We do a true one-to-one evaluation of the optimal way to interact with a consumer. Whether it’s someone coming into a bank branch, contacting a call center—inbound or outbound—or using email, chat, and so forth, that interaction should be informed by a 360-degree view of the consumer and how that relationship has evolved with the financial institution.

The big change about 15 years ago was moving from application software, where we answered specific questions about originations, collections, recovery, and fraud, to a much broader view of how to interact with consumers. That was driven in part by the internet and retailers’ move to one-to-one marketing. That trend really hit financial institutions during the COVID-19 pandemic, when people weren’t coming into branches anymore, and it suddenly became far more important to have a digital relationship with the consumer.

Our software filled that gap. It says, “Let’s understand everything you’ve ever done with this consumer and make sure that when we talk to them, we do things that are relevant and most likely to produce the outcomes we want.” AI has been a key part of this since the early 2000s. We’ve gotten fraud down to a relatively small number for our customers, and that’s all built on early AI neural nets.

We benefit, like other software companies, from the productivity that comes from not having to write code by hand. Our cost of goods sold, as a percentage of revenue, is decreasing. That’s a good thing, although the cost of the AI itself is not insignificant. The real power for us is figuring out how to do this economically for our customers.

There are other challenges with AI in financial services because we’re such a heavily regulated industry. We run into fair lending law challenges. You can’t just have AI make the credit decision; it’s not that simple. A basic example: In the United States, redlining is illegal. Redlining is the use of geographic location to inform a credit decision. In the old days, back in the ’40s, bankers would put a red line around a neighborhood and say, “We don’t like this neighborhood; we’re not lending to anybody in it.”

That’s one kind of problem. Another is that regulators want to understand how you made the credit decision you did. They want to see that it is objective and scientific, that you’re applying the same criteria to everybody, and that you’re not favoring some groups over others. That’s very hard to do with AI. People describe it as a black box: You put in a question, and you get an answer.

We’ve done a lot of work in this area. We have what we call explainable AI—we have patents on it—where we use blockchain to produce an audit trail so we can go back and understand how the AI made its decision. It doesn’t solve the redlining problem, but it does solve the black box problem.

The effects of agentic commerce

Eric Kutcher: You talked about understanding how a consumer pays back debt. What happens going forward, when the consumer may not be operating on their own behalf—when I’m effectively letting an agent make purchase decisions against a set of criteria for me? How does that change the behaviors you look at?

Will Lansing: We’ve spent a lot of time thinking about this in the context of credit and financial services. For the past 30 years or so, financial institutions have wanted to be the agent for the consumer on all things financial. That’s been the holy grail of the industry for a long time. Now suddenly, that possibility is quite real. I anticipate that consumers will have financial agents that go out and shop for loans and credit cards and figure out the best deal.

Eric Kutcher: We’re already seeing it. From the perspective of credit card shopping, it’s been fascinating. The agent may not fill out the application yet, but we’re pretty close. I did this with a client not long ago. We asked them to type in, “I live in this area, I’ve got this profile, I need a credit card, who would you recommend?” It was fascinating to see who got recommended and why, and how it differed by market. This particular company hadn’t thought about that at all.

Will Lansing: I think most financial institutions—certainly the most technologically advanced ones—are already exploring how their agents are going to talk to consumers’ agents to make a deal. That is clearly coming, and it opens up all kinds of new possibilities.

Eric Kutcher: We’re spending a lot of time working with financial institutions to reimagine the underwriting process—within the context of the laws, but with a very agent-first approach. I’m curious what you’re seeing from the other side, since agents must interact with your models and software to pull in the information needed to do the underwriting. What are the big changes you’re seeing?

Will Lansing: The opportunity with AI in underwriting, within whatever parameters regulators allow, is that you’ll be able to extract signals from data sets beyond the usual suspects, and with those signals, you can underwrite for a broader population, potentially giving a bigger loan or better pricing to people who deserve it.

How does that work? It has to do with access to data. Some of it’s still proprietary, but increasingly you have consumer-permissioned data. FICO’s credit score is built on credit bureau data—here’s who paid on time, here’s who was late. But if we talk to the consumer and say, “Can we look at your checking account? Would you be OK with that?”—that opens up all kinds of possibilities. We can look into a demand deposit account. We can see the frequency of overdrafts. We can see regular income coming in and regular rental payments going out. There are a lot of signals in those checking accounts. The trick to applying AI in this environment is figuring out what data you can access and what you can’t and how to capture the signals from all the data you’ve historically not really had access to. There’s still plenty that won’t just be out in the ether, but I think the possibilities for better underwriting are real.

Eric Kutcher: It’s super interesting; for once, consumers might actually get paid for all the data they give away, presumably through better pricing and terms.

Will Lansing: That’s exactly where it comes in: You get a loan you wouldn’t have otherwise gotten, or you get a better price on your mortgage than you otherwise would have.

Optionality allows for backup plans

Eric Kutcher: How do you think about maintaining a connection to your legacy—who the organization has always been—while innovating, especially at a moment when the pace of innovation is what it is and some of the competitive moats you’ve historically had might not be the moats that carry you forward?

Will Lansing: Every business is different, but some things are universal. It’s always useful to figure out what you’re really great at—what you do better than everybody else—and how to take that to the next level. That’s what we’ve always focused on.

Having the best people matters. It’s a cliché, but it’s so true. If you have great people, they figure out the answers to problems and the path forward.

I’d say a key factor for success—depending on how fast-moving the industry is—is optionality. You do your forward-looking planning, scenario planning, and strategy. I don’t like singular bets because if they go wrong, you’re really in a lot of trouble. I like to preserve optionality. As we’ve evolved our strategy—whether it’s dealing with AI or competitors or anything else—we think through the path we believe is right, but we also consider what we’d fall back on if that path didn’t work out.

Eric Kutcher: I’ve heard the expression “one-way versus two-way doors,” and I’ve really taken to it. You don’t have to be overly thoughtful about a two-way door, because you can go through it, figure out if it works, and come back and do something different. However, you have to understand the cost of that, and I don’t just mean dollars and cents, but the organizational impact. A one-way door, I’d better be certain about, because I can’t come back through it.

What have you learned as CEO that you probably didn’t anticipate when you first started?

Will Lansing: So much of what I’ve learned has been about how to get the best value out of the people I have. Leadership comes from followership; it comes from people believing in the mission, agreeing with it, and wanting to contribute to it. I spend a lot of my time trying to make sure everyone is on board with the strategy and believes in what we’re trying to do.

We have three corporate values at FICO: Act like an owner, delight your customers, and earn the respect of others. They sound like motherhood and apple pie, but in practice, when I sit down with a new employee to explain “act like an owner,” I say, “This is your business. You have stock in the business. I want you to do your job as though this were a family business. You control resources, budget dollars, the time of the people who work for you, and your own time. Any time you feel like you’re not doing the right thing, you’re supposed to stop and say, ‘Hey, this is not what we should be doing. This is a bad idea.’” It’s taken 15 years for our culture to really get there, but I’d say most people at FICO act like owners. They believe that’s what we should be doing. It’s taken a long time to inculcate these values, and we’ve pushed stock ownership pretty far into the organization to make it happen. But it works.

Eric Kutcher: Not every young hire has that same orientation today. How do you bring them along?

Will Lansing: There’s some extra pressure on young colleagues these days. FICO has always had this view that we want the best people we can get, wherever they live, however they want to live—the smartest mathematicians, the smartest scientists, the best data scientists. We were a virtual company long before COVID-19. We had people scattered all over the world: People who live in Chamonix because they like to ski, but they’re brilliant, so we like it, too; and people who live in the Outer Hebrides. Fortunately, we have internet access—that was always our thing: Live wherever you want.

During COVID-19, people scattered even further. That’s fine for people who are already involved in the culture and understand FICO well. But it can be a challenge if you join a company while working virtually. We go out of our way to find time for in-person get-togethers three or four times a year for different groups, so people actually spend time together. It’s harder to forge a connection when people aren’t physically together.

The hard part is balance

Eric Kutcher: You’re entrusted with this incredible institution—its legacy and its path forward. You have to balance innovation with the responsibility that comes with it. How do you think about that trade-off?

Will Lansing: I think the idea of the one-way versus two-way doors is clearly part of it. We invest a huge amount in R&D—as a company, we pour money into it. We have hundreds of patents, and we’re constantly getting more. It’s in our nature to try to be cutting-edge on analytics, so innovation isn’t the hard part. The hard part is striking the right balance between R&D and running a business: You have to produce profits for shareholders, compensate employees fairly, and make your customers happy. And it can’t all be theoretical. We’ve got to get it out to market in a reasonable time frame. A lot of energy goes into that balance, especially with software releases: What’s the balance between feature functionality and getting it out the door? Every software company wrestles with that, and we do, too. We’re moving to a much more continuous release approach because the technology now allows for it.

Eric Kutcher: The pricing models are changing, too. You’re not doing perpetual licenses anymore. People used to pay just once; now it’s more of an annual or monthly relationship.

Will Lansing: We went from perpetual licenses to term licenses and then combined various kinds of intellectual property licenses with a usage-based component. As we moved to the cloud, we had costs associated with running it, so we moved further into usage-based pricing. Historically, pricing has chewed up way more time than it should; it’s a shame. We’re moving to consumption-based pricing, where we provide the infrastructure and don’t renegotiate every time you want to expand your use. Use what you want; you know what it costs. For a software company with a land-and-expand strategy like ours, it’s perfect: You get installed, and then the customer decides how much or how little they want to use. I love that we’re spending less time negotiating the next deal or the next chunk of capability, and it’s much more up to the customer to pull down whatever they want as they need it.

Eric Kutcher: What do you hope for in FICO’s next chapter?

Will Lansing: Our core skills and competencies in credit aren’t going away—that’s still going to be the center of the company. It’s a gigantic market. We’re the cornerstone of the consumer credit economy in the United States, and that’s not going to change. But what’s critical is making sure we leverage analytics, technological advances, and AI to be the best at what we do. I hope that looking back ten, 15, or 20 years from now, my successor and the people working at FICO then will think we laid the groundwork to continue doing that. The basics are the same: Find more data with more signals in it, figure out how to extract data at a reasonable cost, and make the very best, most optimized decisions we possibly can. That’s been true for us for decades, and it will remain so.

Eric Kutcher: The final question: If you were to give one piece of advice to someone aspiring to be a CEO, or just starting their journey, what would it be?

Will Lansing: You’ve got to do a good job at whatever job you’re in, but once you’ve mastered it, you’ve got to move as fast as you can to get broader—develop more skills, more breadth. A good path to a generalist leadership position like CEO is to develop broad skills. Don’t get stuck. Figure out what you need to do to get to the next level and go do it. There’s some politics involved, too; you can’t get anything done in a company by yourself. You need the help of everyone. You want them to be rooting for you and helping you succeed. You’ve got to help them—then they will want to follow you.