Welcome in to another episode of the Inside the Ice House podcast. Today's guest is Frank Cooper. He is the chief marketing officer at Visa. Frank, thanks so much for joining us inside the ice house. Happy to have you here.
Man, it's great to be here.
So I want to begin big picture for our listeners on the B2 AI study for the first time. This is a study put out by Visa, results are in. What was just the motivation first behind Visa launching it and the questions that you were hoping to answer with this study?
Yeah, I'm so excited about this study because we've been very familiar with the idea of marketing to consumers, we're very B2C, we're very familiar with the idea of marketing to businesses, B2B. But there's a new customer segment, and that's what B2AI means. It's the rise of AI agents as a customer segment. And you know, who's going to review your product? Who's going to recommend it and eventually buy it autonomously? It's agents. And so for us, we wanted to understand what are the motivations that people have, what are the fears, the blockers that would prevent them from adopting this, and what are their aspirations when they look at AI agents as part of their shopping and travel experiences. So that was the underlying motivation.
And so you talked about right, you have B2B studies, you have B2C studies, this B2AI, what just kind of makes this research different from the typical consumer or the business surveys that we've seen before around emerging tech? Because now you're focusing solely on this technological innovation.
Yeah. I mean look, money tech, you can go back to the ATMs and the trajectory is pretty similar. You know, people start off wary. You take an ATM in the 1960s and '70s, there's low trust and people were thinking, I'm not giving my cash to a machine. Banks installed the machines and there was little uptake at that time. But then you started to see distribution. You started to see the network effects, people participating in it. You cross that trust threshold. And so for us we wanted to know what would it take to get consumers to cross the trust threshold with AI agents, AI-driven commerce. And that's what the study really took a look at. And it was fascinating because you could see that there was already adoption of AI as part of the shopping experience, but there was a fear, and that fear was I want to have control, I want to have some autonomy. And so you saw about 53% of the people would say, yes, I use it in some part of my shopping experience and I enjoy it. But half of that, 27% would say, but I don't want that agent buying for me autonomously and I certainly want to have some control over it. You know, about 40% would roughly say, I want to be able to stop the purchase if possible. But I think we're at that early stage of that S-curve, you know, where people are just now starting to get an understanding of what does it mean to have an AI agent shop for you? And so I think we're just at the beginning of it.
And so there's a real intersection nowadays of technology, culture, and commerce, right? They're all sort of coming together with that convergence happening. Why did Visa feel like now was the time to study AI not just as a tool really, but sort of as a behavior, seeing how people are using it and how they feel it can be used moving forward.
You know, mainly because I think it's going to be so fundamentally different from what we used to think about when we wanted to attract and retain and create preference with people alone. So, with people, we thought about emotions first. It's like, okay, emotions are, behavioral economists have taught us all that emotions kind of drive the behavior and then you rationalize it. And so, we're sending emotional communications. We deal with limited memory and limited capacity for attention. And so like don't overload people with the messaging. All this becomes irrelevant when you're communicating with a machine. Machine can read hundreds of pages at a time. Doesn't matter. No emotion. And they don't see your billboards, they don't care about the tears. They don't put a baby in a commercial, they don't really care. The machines don't care, right? And so, but what the machine does need is that it has to be readable. And it has to be trusted, but most importantly, it has to trigger a response from the human at the other end that what I received was effective. And so, for us, it just changes the sales funnel in such a fundamental way. We wanted to understand how can we get ahead of that curve, as we start to communicate with both humans on the beginning of the prompt, but also the machines in the middle and then humans at the end again.
And I think whenever you go into a study like this, you sort of have your own expectations, right? Not necessarily a bias, but just expectations of what you think the results could bring. Whether it be about how consumers are using AI, the pace of its adoption, who's using AI, how did the results match what you and the Visa team may have thought prior to putting this study out?
You know, you're right because with a study, you're always looking for that moment where consumers say something that just parts the clouds and the sun comes through and you're like, well, I didn't expect that. For the most part, the study was consistent with what we thought might happen, which is there is an appetite for AI-driven commerce, but there's fear about AI-driven commerce. And what we hope the study would give us is a little more clarity about what are the barriers, what are the factors of fear that will prevent people from moving forward. And what I was pleased to see is that you can see the adoption on the business side. So there's a B2B opportunity clearly, and that's moving much faster than the pure consumer side. But I was also happy to hear that people have an appetite to adopt it if you can start to create the infrastructure, the network effect, the social proof, the kind of ubiquitous distribution that's always given them confidence about when money, technology, and trust kind of intersect.
And so I think a study is really only as valuable as what you do with it after the fact, right? You take these results and then you have to then turn these results into something. Whether that be something tangible, intangible. How is Visa planning on or maybe even now taking these insights and translating them into products, into tools, guidance, partnerships, so be it.
Yeah, I mean, these studies are basically facts and observations and data points. We try to translate that into insights. Again, what are the blockers for adoption. And so the Visa Intelligent Commerce Platform is essentially a response to that. And what we get clarity on is that people want more confidence about identity, is the person behind the agent a real person. And so you get Trusted Agent Protocol from Visa. They want to get greater comfort around authorization, does this agent have the authority to actually make this purchase at a certain spend level. So Visa Acceptance Platform kind of fits into that category. They want to know about liability, what happens if something goes wrong, if there's fraud, how do I know this transaction is secure? And so we're creating a set of platforms and protocols and APIs that allow for a more seamless delivery of AI-driven commerce, but also to create that trust layer so that people feel more secure in the transaction and understand that their identity, authorization, and liability is covered.
And you mentioned that word trust again and you spoke a couple of questions ago about that sort of trust threshold that needs to be crossed for people to really have the confidence in these AI agents to be able to one day maybe make purchases on their behalf. How do you get to that point of crossing that trust threshold? Is it commercialization? Is it proof of concept? How do you determine or how do you help people determine that now is the time that we can finally trust this?
You know, sometimes it's unpredictable. Contactless, for example, really didn't hit the inflection point until the pandemic. Lots of talk about tap to pay and contactless pay. But when the pandemic happened, it became the standard and you saw the spike. But I think it's a combination of things. I think number one, people need social proof. They want to see others use it safely and then comment about it. You need the infrastructure. You have to believe that the infrastructure is there that actually gives you some confidence that this can work. You need ubiquitous distribution. And again whether it's the ATMs, whether it's trusting strangers online with payments, even crypto which is kind of in the midstream right now. It's all going through that same phase. Do I see the network effects? Do I see ubiquitous distribution? Do I see an infrastructure that gives me confidence? When you start to see that then the behavior follows, then you get the social proof and then I think you start to see it scale. AI-driven commerce though, I think it's going to move very, very fast because people are already using it and are not completely aware of it. So they're tapping into Gemini when they now are searching in Google. They're not completely aware, but it's getting a direct answer. People are scrolling down less. People are using it for cross-border travel already. And so it's helping people along the way and it's helping them do things that they otherwise could not do without it. And so I think that to me is where you start to hit the tipping point where the behavior change is better than the status quo.
And can you just dive a little bit further into trust specifically when it comes to Visa because I feel like trust and Visa, right, those are two words that are kind of synonymous with one another. But how is Visa, with a technology like AI that is still relatively new to a certain extent, how is Visa really maintaining that trust with anything AI related? Obviously, we're focusing on the B2AI study, but just with trust and Visa, how is that coming together?
Well, look, Visa has been synonymous with trust because we've been synonymous with broad acceptance. So, anywhere you go, it will be accepted. You don't have to worry about that. So, carry your Visa credential with you and you're going to be taken care of. We've been synonymous with security. So we will protect your identity. We'll protect your money with some of the best fraud protection software in the history of fraud protection software. We've been using AI for over 30 years in that effort. And we've been trusted because of zero liability. We make sure that the user has zero liability. I think those same things will apply in AI-driven commerce. But you know when you think about trust in general, all the studies of trust, whether it's Rachel Botsman who can you trust or you go back in time into the thousands and thousands of definitions of trust, they all come down to three things. Does the person or company understand me? Are they capable? Do they have the capacity to actually solve the problem that I have? And then three, do they have integrity? Do they kind of walk the talk? And so for us, we try to do all three things. Understand the consumers that we're serving and the customers that we're serving. Continue to innovate and build capability and remain consistent. And I think that consistency is really important. You know, one of the things that we've done recently, we've done a bunch of ads from everywhere you want to be and anthemic ads and highly emotional ads and those are great. I love them and we're going to continue to do it because I think it makes a difference. But we really have been emphasizing the functional advantages of Visa, most secure credential in the world, most accepted, zero liability. And I think reinforcing that particularly in this time of AI-driven commerce is going to be really important.
And so the B2AI study to me really paints this sort of picture of a changing consumer landscape kind of in real time. And obviously we're seeing AI and its implementation and integration in businesses and we're seeing business landscape change through it and business landscape of all sectors be disrupted. But to me this sort of paints a picture like I said of the changing consumer landscape as Visa's CMO. How do you guide the organization through this kind of rapid evolution, things are popping up left and right, things are evolving and you are one of the leaders that's in charge of making sure the organization continues and adapts along with them.
You know, I love this period of time because it's uncharted territory. No one's been here before. And so I feel like we have two choices. We can sit and wait and react to what comes at us or we can try to shape the future. And so for me, shaping the future takes you back to almost first principles. And so if an AI agent is a new customer segment, what would you do if for any customer segment anywhere in the world, the first thing you want to do is understand how that segment operates. And so you start to reduce the sense of risk and certainly reduce the sense of uncertainty simply by understanding how does it work. And so we spend a lot of time thinking through how do you develop loyalty and preference from a machine? It just takes it down as a consumer segment. But the thing I want to make sure people don't lose sight of is that there's still a human connection to all of this. It's kind of the human-AI sandwich in a way. And in the beginning someone wants to do something, a person, and they're going to put in a prompt. They're going to put in some kind of request. They can put in a generic request if they want to, but if they have a preference or if they're loyal or they're advocates for a certain brand or a certain product, that prompt changes. And so we can affect the prompt, we can affect the person upfront, but then at the end of it, AI is going to deliver something. It's going to do its thing, computational statistical predictions, and it will deliver a result. At the end of that result, there's a human being receiving that and responding to it. Again, I think we have an opportunity to affect that. So we have to win twice now. We have to win both with the humans and we have to win with the machines. And if we do that I think we again we won't have all the answers but if we kind of look at it through that lens at least we have a greater sense of control but more importantly we start to get data and deeper understanding of how this all operates. And because no one knows we can start to shape some of that.
And how do you do that with an LLM? Right. Like that's such a challenging thing to do, right? Like you said, you have to win with the human, you have to win with the machine. I understand the sort of science behind winning with the human, Visa has done that for decades. But winning with the machine, I mean, that's a whole new ball game. Like you said, uncharted territory. No one's really done this before. So what are the steps that need to be taken?
You know, look, it's really understanding how the machine is programmed. And so the pre-training piece of it, I think, is really important. And what's fascinating to me is I love my engineering brethren but I think they have a very narrow view of marketing. And so if you looked at the earlier versions of the training and the algorithm, it actually preferred reviews over company websites and company information. And that preference was really clear. You just plug it in. You could see. And so at that time you were driving to reviews. Exactly. But then it became clear that actually the best source of information about a company and a product is often its website. Suddenly .com becomes important again. And so just having that basic understanding of where it's sourcing its information, what it prioritizes, what it trusts, I think is really, really helpful. Secondly, is having the appropriate APIs so that it's easy to access the information. But then I think the third layer is this, and this is where it gets a little more speculative. As you start to have agents speaking to agents, and kind of shaping the way they think, I think marketers in particular have to make a fundamental shift and think less about trying to persuade and move and think more about how to provide fundamental information that can be trusted. And for me, again going back to the first principles, it's not about when people think of marketing, it's not about the next Super Bowl commercial, the next anthemic film, the next big experiential activation. All those things are great. I love those things. But for me, the fundamental quest of marketing within any firm is to drive profitable growth. And you do that by changing people's perceptions, emotions, and behaviors. And now that you have this fourth customer, structuring it in a way that AI agents can read it and digest it and produce it in a way that's favorable to your firm.
So would you say that the goals of marketing have really stayed the same at the end of the day but now how you market is changing?
100%. 100%. Goals have not changed. Now does the C-suite across every company and every enterprise see it in that way? I don't think so. I think what often happens is that people get locked into a formula or to muscle memory. So you'll get some companies which will say we're all about performance marketing which is a terrible term because all marketing should be performance-based, but they're talking about bottom of the funnel customer acquisition, minimize CAC, customer acquisition cost, and maximize lifetime value. Some of them lock into that and say that's what marketing is. Others lock into I need the big emotional communication that people talk about. I think we all should lock into do whatever it takes to drive profitable growth by changing people's perceptions, emotions, and behaviors.
And so Frank before we pivot the conversation to some recent happenings with Visa specifically around sports marketing, one last question on the study when you just look at it holistically and you could take one or two or maybe even three, the biggest takeaways coming from it, what would you sort of describe as sort of like the kind of aha moments?
The aha moment is that in order for AI-driven commerce to scale, in order for it to cross the chasm from these early adopters to the mainstream, we have to cross the trust threshold. That's the biggest takeaway. And the data kept coming back to us with this kind of dynamic tension between an aspiration to dive into AI-driven commerce and the fear that these agents, that they could be living in the matrix and the agent suddenly take over and you realize that you have no control over your agent and they're spending your money in ways that you don't want to. And so crossing that trust threshold for us is the most important takeaway. And what I feel great about is that we have a playbook for that kind of money tech growth adoption curve that we've seen before. And that playbook is build the infrastructure, give people the confidence that the distribution will be ubiquitous, build the security layer and the trust layer in a way that people can rely upon it and then establish social proof. And on the marketing side, I think a lot of what we're going to be doing is really demonstrating the social proof. It's not so much marketing features and benefits. It's showing that people believe in it. It's working for them. And they've had great outcomes in using it.
