Raj Seshadri on How Data is Helping Executives Rethink Everything

You'll have to carve out another hour for me to talk about the postdoctoral and PhD. Yeah, fascinating stuff. I worked on melting transitions. You know, a glass of water with ice going to water in three dimensions, it's a phase transition in two dimensions, like a Post-it note. That's a two-dimensional in-between state, like taking a rubber mat and stretching it with a crystal drawn on it. Anyway, we can spend a whole hour on that. So what I do, I'm very proud to work for Mastercard. Thank you, Josh, for your kind words. It's a company that believes in doing well and doing good. Both are important. What I do at Mastercard is I run our data and services business. That means we help customers, which could be issuers, banks, merchants, governments. We help them make smarter decisions for better outcomes. So we help them do well in their businesses using data and insights derived from it, as well as help them better engage with their consumers, whether that consumer is an individual like you or me, or a small business or a corporation. So that's what we do.

It is massive. Before I give you the exact number, let me tell you something about our data responsibility principles that govern how we think about our data. We adhere to it very strictly, including all the way through COVID. We collect transactional information when we process a purchase transaction, but it's very limited. We know the card number, that it was used at a particular merchant at a particular location, the date and time, and the total amount. There's a lot we don't know. For example, if I use my card, we don't know it was me as an individual, where I live, or what I bought for that $20. It could be a bottle of Tylenol and a pack of Kleenex. So it's very limited, and we're very focused on data responsibility. But it's a lot when you add up the transactions. We have 2.6 billion cards worldwide and roughly 87 billion transactions per year. With this limited data, we aggregate it, anonymize it, and use statistical techniques. We only look at statistically relevant groups of transactions, never an individual transaction. From that, we derive interesting insights. A good example is our Spending Pulse, which predicts the retail market. It's very important now with COVID-19. Spending Pulse takes Mastercard transactions, limits them to a retail sector, and uses third-party data and analytical techniques to extrapolate to total retail spend, including all card types, cash, and checks. That gives us interesting insights. For example, in April and May, the share of sales online doubled versus the previous year, not just in the US but globally. We also found that local matters. We released a report called Shifted Digital, which is a theme of recovery insights. What also matters is that you can go down locally to a country, state, county, city, or neighborhood. In the COVID crisis, for New York City, working with the Partnership for New York, we helped folks understand exactly what was happening down to the neighborhood level. So the data is powerful, but it's about the insights and what it tells you, how it informs you, helps you understand, assess, react, and think about what you do.

Our services business does focus on security as well. Data breaches and data theft are big issues, even bigger with digital and e-commerce growing. At Mastercard, we use an approach where we take our own experience and try to make it useful to customers. That's how we created Labs. This is another example where we innovated ourselves. We have a global network we need to keep secure, and we need to keep Mastercard transactions and the digital ecosystem secure. So all the innovation we did around that we now provide more broadly externally. It's not one solution; it's a combination of several solutions. We have a principle called security by design. We take leading-edge standards and technologies and combine them in the right way. We have a multi-layered approach. Think of your house: first, prevent, like putting a lock on your front door. Second, identify, so before I open the lock, I want to make sure it's you. Third, detect, like an alarm system that alerts you even when you're not there. So it's a multi-layered strategy. In addition to protecting transactions, we also think about protecting companies. We think of cyber risk, which is complicated. You have to look at your technology, workforce, and processes. We've done this by thinking about what we do internally and by buying companies or partnering. We bought a company called RiskRecon, a Mastercard company, which does an outside vulnerability scan for a company. We couple that with another partner we bought called Strategic, whose software we use internally, and then we can do an inside-out scan to identify vulnerabilities. It's under the same umbrella of understanding which risk is most impactful to mitigate, so when you spend a dollar, you get the most bang for the buck.

To answer that, you'd have to go back and work with the banks. Between our technologies, we have machine learning deployed into our network that detects fraudulent patterns and stops them, along with the banks' and merchants' own techniques. The 87 billion are transactions that go through, not those that get stopped. A lot of transactions get stopped. Machine learning and AI are very helpful in this space.

No. EMV, the cryptographic standards we have, applied to contactless too. Contactless EMV transactions are just as secure as physical EMV transactions. Contactless is interesting you mentioned that. That's just the payment, the tap-and-go. Let's talk about COVID right now. Contactless has gone from the payment to the entire experience. We now talk about touchless retail, which is how to do the entire purchase without touching anything in this COVID environment. It's ironic because for decades, high touch was the word. You wanted to touch your high-priority customers in a high-quality way with individual, person-to-person attention. In the COVID environment, we've done a flip-flop. Fewer people attending to you, less interaction, no touch is now about quality, care, and value. It's funny how it's flipped. Let me give you an interesting story. A CPG company came to one of our sessions at Mastercard, and they wanted to launch their first-ever at-home hair care product in 2019. They needed to design for virtual consultations and personalized recommendations remotely. Then COVID hit, and suddenly there were no hair salons. It became much more relevant. What looked like a niche market became the whole market. Our team helped drive engagement and adoption using data and approaches to understand the consumer and better serve them. They blew through their forecasts. So it's a different environment we're living in.

It's interesting. It is a buzzword and often misunderstood. It's actually a spectrum of analytical techniques. We're spending a lot of time thinking about it in many different contexts. One example is data responsibility. We're thinking about it in the context of our principles. You have to be deliberate about what data you use, how you use it, and for what purpose. As you do that, you minimize biases and unintended consequences. You don't want biases sneaking into your AI or machine learning, whether it's sampling bias or population bias. You don't want the AI to inherit all the biases built into what we do today.

It's not as simple as somebody sitting down and doing it. It's a question of thinking about it very systematically from when you design the analysis all the way through. You have to be aware of all the different types of biases as you think about the problem you're solving, the data you're picking, and the analytical techniques you're using. If you're automating something, it's not simple to do, but it's really important because you don't want to inherit those biases.

No, it's not purely a programming exercise. You need principles, the different categories, and a lot of analytical thinking. It's not simple, but it's super important to do. We're actually using AI across the board. One way, as I mentioned, is we put machine learning into our network. Another example is we're trying to speed up our analytical recommendations and analyses so that decision-making can be informed in the moment. The idea is to make us smarter and faster, saving time so we can push the envelope and start doing other things.

I read a lot of different things. But what I will say is it's not as important what you read, but to make sure you're informed and have timely information. Information in itself, like data, is not a competitive advantage for anyone or any company. It's necessary but not sufficient. What is much more important is how you think about it, how you use it, and how you deploy it. So I could tell you all the things I read, but anyone can read. The edge is not the information; the edge is how you think about it, how you use it, and how you apply it.