Ed McLaughlin, President & Chief Technology Officer, Mastercard, joins CIO Leadership Live from Foundry’s CIO100 event

00:00 00:00:09:47 - 00:00:22:42
Welcome to CIO Leadership Live. I'm Lee Rennick, Executive Director of CIO communities for CIO.com. And I'm thrilled to be here at the CIO 100 and Symposium with Ed McLaughlin, President and CTO of Mastercard.
00:00:22:42 - 00:00:43:16
Ed, it is so great to meet you in person. We had a chance to have an interview last year. It was fantastic. It's great to see in 3D. Yeah, it's so great. Thank you so much for being here today. I appreciate it so much. So, we had a chat earlier in the year. and it was really great to talk about all the innovation you working on, but there's a couple of congratulations here.
00:00:43:16 - 00:01:08:48
First, Mastercard has won the CIO 100 award for some projects you've worked on, and you have been inducted into the CIO…I was going to say Rock and Roll Hall of Fame, because CIOs are like rock stars these days, and you are one of them. So congratulations on that. That was just amazing. So we just have to learn a little bit more about the award winning project and maybe, maybe just any reflections on being inducted into the Hall of Fame.
00:01:08:48 - 00:01:09:28
Yeah,
00:01:09:28 - 00:01:33:52
yeah. Let me actually let me start let me start with the Hall of Fame, because it really is in some ways an introspective moment. And thinking about it, we really do live in an age of miracles. When I think about the technology I started with the beginning of my career, what we've been able to do since I truly is amazing and it really is a global transformation led by and driven through technology.
00:01:33:57 - 00:01:51:10
just looking back at 20 years of Mastercard, we were a physical world square, a plastic organization doing amazing things. But as the world was changing around us, we got to help invent that future. So from the beginnings of e-commerce, where you could provide things like a payment, zero
00:01:51:10 - 00:02:00:41
liability to consumers, the right apps and services, and global acceptance, the Amazons, the Ubers, the Netflix were able to flourish because the services we could help provide,
00:02:00:41 - 00:02:15:43
with mobility coming in, the ability for Google and Apple and others to, to build their mobile payments apps using genuine Mastercard technology and standards that we help to put out there working with square to turn it in.
00:02:15:47 - 00:02:37:04
We had a theme of every device can be a commerce device, and really making that happen and going through the fintech revolution, thinking about what we're doing is AI, which I'll talk to in a bit about an award. It's just been an amazing, amazing time. But what I thought about the most is, you know, you're for Mastercard 20 years ago, there's about 3000 of us.
00:02:37:04 - 00:03:05:31
We have $4 billion market. We now $400 billion. Wow, 30,000 employees. But it wasn't the tech that did it. It was the people. And just this whole idea of who we are and how we use technology, that's what makes all the difference in the world. And it's it's always just so humbling to get an award like this and realize what you get to represent is all the amazing work of all the amazing people that you get to work with.
00:03:05:36 - 00:03:23:21
And that's why we love the CIO 100 award that we're here with today. It's great to win another one. Yes, that's why I love this example. I think this is something that applies to just about everybody here. So we get a tremendous amount of cases in that we have to work from our customers. Yes. As the business has grown, it's very, very complex.
00:03:23:21 - 00:03:43:58
We have merchants, we have bank issuers and consumers. We have all kinds of cases coming in as we work. We're in 210 countries and regulatory domiciles. Right? We've massively diversified our product set and the number of customers that we have. So we get a lot of unstructured inbound that we need to get to the right person as fast as possible.
00:03:43:58 - 00:04:02:58
Yeah. And from a customer standpoint. That's the moment of truth when they need your help. Now when things are going great, I need your help. How fast get the right person to the right place. So I think there's a lot we can do to take humans out of the loop of the simple stuff. Right. Getting the right expert to help the customer relies on the really complex stuff.
00:04:03:09 - 00:04:30:02
There's huge value there too. Yeah. So next year we'll do about a million cases that we have to hear. So it's pretty high volume. Also gave us about a $4 million training or 4 million case training set that we can work on. Right. So what we're able to do is take all the unstructured information, a lot of which had start with email, because that's where our customers are and be able to translate that into the actual product or service or country or region they were coming from, and use that for intelligent routing.
00:04:30:07 - 00:04:45:58
And you get all the folklore of, you know, a customer in India waiting for some of us to wake up so they could have the right. Exactly. Yeah. Or the worst thing for us is when something got mis routed, you be at the end of your SLA and you have to start over again, right? It wasn't the team that was supposed to handle it.
00:04:46:13 - 00:05:06:07
So we had 16 people just working on trying to route things to the right place. It would take an hour and sometimes half a day, to figure out what the customer was asking for. And one other complexity. Yeah, if I could. Yeah. I don't know if this happens in other organizations, but it seems like the products that we're generating the most cases, there's like the Witness Protection program.
00:05:06:07 - 00:05:24:42
Right. And changing it's to figure out what they are actually asking about versus what it was called today. Oh, and I think in a recent AI research, which we're we're really driving here, there's something about converting the unstructured into the canonical to drive real controls and processes around. Yeah. So fast forward to what we were able to do, right?
00:05:24:47 - 00:05:46:32
Biggest impact is we were able to eliminate a lot of that long tail and this routing really expensive stuff. Yeah. Biggest customer issue for that. We dramatically cut the time from 1 to 12 hours to near instant to get the cases to the expert who needs it the most. We're bringing it across other channels for that. Yeah. All the people that used to be working on just trying to get the to the right place.
00:05:46:32 - 00:06:07:08
Yeah, we now have working on customer Success to say how can we improve. Right. How we handle those cases to begin with, right after we work with product teams to eliminate those cases. Right. So the ROI on this one has been. Yeah. And it gives us a great foundational model of all this unstructured information that we can now tie back into real process.
00:06:07:13 - 00:06:30:54
So really, really excited and proud to have the recognition teams. Yeah that's amazing. And when we had our last interview we talked about your transit, all the work you did during Covid, right, as an essential service and how you had to really shift the way you work. So it sounds like to me, having talked to you once before, it's like that was kind of the beginning of looking at ways of building out your teams and building up best practices and then just adding on to this.
00:06:30:54 - 00:06:56:19
It's it sounds like it's been an amazing evolution and working back from hard problems to solve. Yeah. Saying can new technology allow us. Yes. We never could. Yeah. Yeah. So we talked about the project. Congratulations on that. But what about your own induction into the Hall of Fame? Like I said, it's a little humbling and it's really an honor to to represent, I think, all the amazing things that the folks at Mastercard early in their career.
00:06:56:24 - 00:07:15:06
Well, congratulations on that. You know, it's just phenomenal. And the work you're doing is amazing. And I, I really enjoy our discussions. And we're we'll be having a panel tomorrow talking about looking forward. So that'll be great too. All right. So I want to focus a bit on a little bit of technical technology stuff. So I speak with a lot of CEOs about the data in the cloud.
00:07:15:19 - 00:07:33:02
So some have moved like especially during Covid as we talked about off prem into the cloud. There's there's still sorting out what they should do. Some are saying they want to bring repatriate back on, you know, others are saying well now with Chennai we're going to clean our data up and work from the edge as best as we can.
00:07:33:07 - 00:07:48:24
So I would love to learn if you could share a little about your approach to cloud and data and any tips you could provide to anybody listening in. So, you know, so do we have it? I mean, we were joking. This is one of these things where I think there's a whole lot of is yet to be yet to be granted.
00:07:48:24 - 00:08:13:08
Yeah. This one. Right. Yeah. so maybe three things that I'd say are lessons that we're applying for us that, because the answer is yes. And it really is one of those situations. Yes. People blindly adhere to one or the other. I think we're missing all the nuance and counterfactuals. And so you start with what you already know the data set, the data needs to be in proximity to the systems that the consumer.
00:08:13:13 - 00:08:30:36
So we always start from an efficiency standpoint. I would also caveat we work at a scale that may be different for a lot of our data warehouse has pretty much a quarter of all internet transactions since there was an internet, you know, outside of China, and things like that. So we we deal with a lot of data.
00:08:30:41 - 00:08:52:34
So if you have relatively predictive load that you're using to run your business, are you running it on prem and continue to optimize, that is really important. I will also say I'll get back to this at the end. We need to work as a cost control part of the organization, right? And when someone presses a button, right, every light in the city, you're operating it and you can say, don't do that.
00:08:52:39 - 00:09:18:30
Yeah, right. If it's a runaway query on the cloud, you pay for. Yes. So I think the idea that there's certain things which is running our operations, it's maximizing the efficiency, it's bringing the data as close to where you need it to be. That will always have strong. Again, if you're operating like a Mastercard. Yeah I think the second thing that's important though is as you look at alternate environments for your data, it comes down to policy.
00:09:18:35 - 00:09:42:46
You know we've published a consumer data bill of rights. You have the right to know what data we have, right to know how we're using it. Right. It is regulators around the world, whether it's through things like GDPR or other requirements of knowing where your data yes. So the idea that data to be siphoned off into a dark corner of the cloud environment for people to do things that the organization might not have full oversight on.
00:09:42:46 - 00:10:08:26
Yeah, that's a fundamental flaw. So making sure that your data policies are consistent across all operations. We have we have things like our encryption. We need to control the keys for that is the standard. Right. But if you look at the recent snowflake it's yeah. So you always have to have multi-factor authentication regardless of. Yes. So making sure you have consistent policy across environment I actually think is the most important element.
00:10:08:30 - 00:10:28:28
Which leads me to the third thing I'll say. And this is with great enthusiasm, there's amazing capabilities and amazing things you can do by intelligently leveraging public or third party cloud resources for the ability to meet our customers. Where they are. Yeah, the ability to use common battery data sets in the ways that the the race that's going on.
00:10:28:28 - 00:10:53:31
So people looking to serve us by investing in incredible capabilities, tools and workbenches. So with that second bucket will control. Yeah. To move appropriate data into the environment to do things you would never want to do. It right becomes very important. Yeah. But the one caveat to having the feedback, if you have to have if the first bucket is how you run your business, the stuff you're doing are traditional cost containment, ROI type.
00:10:53:33 - 00:11:12:17
Yeah, it's even more important in the cloud. We can say, what are you doing it for? So you're moving everything into a really, really expensive analytical environment will make your data science is really happy. But if you used to, if you're still going to the grocery store and you used to ride a bike and now you have an SUV, you're still just going to the grocery store, right?
00:11:12:17 - 00:11:29:18
Yeah, yeah, yeah, yeah. So making sure that that feedback loop is there. So a lot of things and we'll talk about this a little bit tomorrow within ops and the ability to do attribute. Why are we in this environment. What's the expense in this environment directed back to the benefit to the business. It's kind of hard to do if it's to run your business stuff.
00:11:29:33 - 00:11:59:42
I think that's the critical success factor for effectively all the third party. I think that would be very helpful for anyone listening. And that second piece, that second piece around security and how you're encrypting and, you know, interacting with the with the customers, I would assume that's ever evolving. Like, how are you? I just out of curiosity, we didn't have this one scripted, but I know for my in my instance, my bank in Canada, well, I just transferred some money and now I had to do two different types of verifications in order for that to happen.
00:11:59:47 - 00:12:19:48
Right. And I was like, oh, this is new. So actually, as a customer, I felt extremely happy about that and protected. So yeah, I'm just wondering how you balance that. Like how do you create those policy and ensure that as you know, everything changes and there's more people hacking and all of this stuff in the security aspect, sounds like you've really got a process or strategy in place to make sure you're constantly looking at that.
00:12:20:01 - 00:12:39:03
That is something which constantly evolves. Yeah. I'll be the first to say, I think it was Andy Tannenbaum saying, the great thing about standards is there's so many to choose from, right? It's really important for companies to set the appropriate standards for what they're doing. Yeah. And then apply the technology against that and not compromise. Yeah. Yeah.
00:12:39:03 - 00:13:09:55
So for us we constantly look at one all the regulatory requirements that are around in the payments industry with the payment Council. PCI yeah, just data security standards. So these are tier two for years now. Personally identifiable information is even more fraught in how you can do it using. So again we really try to distill all the regulatory obligations, all of the standards that we want to adhere to, that we get from the bodies and then end up and saying what are Mastercard good practices?
00:13:09:55 - 00:13:28:39
Yeah, that we want to have and make sure that we always enforce inspired. Yeah, I love that. And like you're the first in the business doing this to ensure that your customers are secure. I really like that. And I do think I love your example of the bank you work with, because trust is so fun. I think some of the elements of the tech were deserved.
00:13:28:50 - 00:13:54:02
Yeah, people are comfortable. Yeah. They're being data is being used and who they're doing. Yeah I think that's led to regulatory overreach even at times. So our ability to be clear and comfortable with how we're handling. I think the critical success factors. Yeah. Yeah I appreciate that feedback okay. So tomorrow we're going to be doing a panel at the CIO Symposium 100 on Sustainable City.
00:13:54:09 - 00:14:12:44
Yes. you mentioned something to me that I'd actually never heard of. So you were talking about a discussion around the double materiality, and you chatted about how Mastercard reviews, environmental usage and looks at it in a global sense. Or you said the triple A materiality, and how you innovate in the sustainability area. So I did look it up.
00:14:12:44 - 00:14:29:33
I got myself educated. But I love this idea of the triple materiality. So I would love you to talk a little bit about this. So let me start with and I think it's a really important analysis for any company. You start with double materiality. What are the things which do truly have societal benefit. Yeah. And are great for your business.
00:14:29:33 - 00:14:58:26
Yeah. So one example is for us transit is a fantastic growth market. For us it's an anchor activity. The work we did going back to I think was 2006, when we did a pilot in Lexington and the line in New York. So you can tap, right? Yeah, get onto the subway, which is just normal now, right? Yeah. But so that is something which leads to more sustainable, platforms, transit and commuting.
00:14:58:37 - 00:15:20:19
Yeah. And it's great for Mastercard as a business. Yeah. So again, we emphasize that for things like financial inclusion, we had a pledge to bring half a billion people into the formal financial system. Because even if we can connect to you technically, if you can't participate economically, right. So cut off right. So again, sustainable economic development is wonderful for our business.
00:15:20:19 - 00:15:41:21
Many of our groceries are going to come from unaddressed markets. So you get that double materiality. Yeah it's good for society. Great for your business. Yeah. The triple materiality I think is kind of unique to tech, which is resilience is often intentional in efficiency. Okay. So let's say that one more time resilience is often intentional in efficiency okay.
00:15:41:21 - 00:16:06:08
Need to have backup systems okay. Interesting. Yeah. So part of our materiality is we run critical national infrastructure right. Which again is our social responsibility. It's great for our business to do that. Yeah. But it also means we need to make investments. We might look in economics to make sure we're delivering what we need to do. So the role we're playing in society.
00:16:06:10 - 00:16:33:53
Right. So within that you can then say, okay, green technologies like running compiled applications versus interpreting applications. Right. And network fabric where you can move loads around so you can take advantage of things like green power or environmental computing. Yeah. You can both meet your obligations for resilience, for the critical national infrastructure, while also optimizing on other things which are beneficial to your business.
00:16:33:53 - 00:16:59:51
Right. And our CFO, Sachin here is pretty hard nosed, loves the idea that it's to win is anything which one is more efficient? Right? Right. It's usually less expensive. That's incredible, I love that, I love hearing about that. And I love the way you're thinking about it. One of the other things that we did, and this may be for, our submission for next year's CIO Awards, we've done a lot of work to try to tie our carbon usage.
00:16:59:56 - 00:17:21:18
Yeah. Back to our compute. Right. Of course. Allows you to do is if you can know exactly where you compute and cost it. Yeah. You can associate also carbon cost. Yep. And we're now seeing our customers have reporting requirements and obligations where they're looking to us. Is there scope three suppliers to be able to say here's what we consume and carbon in order to do that.
00:17:21:24 - 00:17:38:00
Yeah. But it's a double win. Yeah. Because in order to calculate something like the carbon cost, you have to know exactly what you are, exactly where you are. And again those optimization engines can kick in. Yeah. And I think you're setting an example for other vendors and companies out there, like maybe some of your own tech vendors. I know a lot of them have to report on that.
00:17:38:00 - 00:18:00:47
Right. So you're kind of setting up this opportunity where your company can report that back to some of your customers. You have your suppliers reporting that back to you. It just probably creates a better environmental efficiency, for sometimes that virtuous cycle is more so using much more efficient compute. Yeah. By definition you're getting that benefit. The end the other consumption have.
00:18:00:52 - 00:18:28:41
And maybe one of the stars starkest examples I love to give of technology choices. So just using electricity consumption as a proxy. Yeah. Mastercard transactional is somewhere around 700,000 times more efficient than a Bitcoin transaction. Wow. So if you think about, you know. Yeah, yeah. Burden that comes from that, how you design your systems as a profound impact in the footprint on their systems.
00:18:28:46 - 00:18:50:15
Amazing. Thank you for sharing that. All right. This segue as well into the next question. So you know, you just talked about your project and the AI automation that you're working on, really to increase productivity for the teams and customer service and ML. So, I would love to know what trends like you're seeing in AI right now.
00:18:50:15 - 00:19:06:09
I, you know, here's what I'm hearing. Okay. Maybe it's maybe it's right or maybe it's but what I'm hearing from a lot of the CEOs I speak to is a lot of organizations are using it right now for internal productivity. We we have been talking about that ways in which you can increase your productivity with your teams, with your company and some.
00:19:06:10 - 00:19:23:24
I was just talking to a colleague, you know, we have we have it on our computer so we can use it just regularly at work, right, to increase productivity. But then there are more sophisticated ways of doing it. many companies are sort of still harboring, okay. If we take it to our customers, how are we going to do that?
00:19:23:24 - 00:19:43:20
What what how do we what support do we need? Right, right. What how do we need the help to get that there? What expertise do we need. Because we don't have that yet. So that's what I'm hearing. There's just we're just I feel like at that precipice with some companies of like, we're just about there and ready to launch something with our clients, but we're maybe not a customer service based organization.
00:19:43:20 - 00:20:07:40
So we're holding back a bit, especially around security and all sorts of stuff like that. So tell me, what are you thinking? So, so much around that. Okay. And one of the things I have to do, it's a little editorial. It is amazing for me how right now it seems like every question now is, but is it a, you know, maybe a decade ago everyone seemed to have to answer, you know, and everything was on the blockchain.
00:20:07:49 - 00:20:25:32
Right. Exactly. The database. Yeah. So yeah, you start with what we talked about earlier. You really have to focus on what's the problem you're trying to solve. Right. And does this new technology allow to solve it in a way you never could before. Yeah I'm really going to give you that. All right. Now too many times and too many times I see people saying, can we use it here?
00:20:25:35 - 00:20:50:49
Yeah. Rather than what are the hardest problems that we could never solve, right. That this will let us do that. And, you know, I just have this vision of people like pounding screws in the very expensive socket wrenches. Right. You have to see what's appropriate for this tool. Yeah. So for Mastercard, last dozen years, we've had massive, massive benefits in using AI and a lot of experience running out of production, which isn't easy to do.
00:20:50:49 - 00:21:11:34
And can you punish the expense. Yeah. Yeah. If you think about things like model drift in the applications, but most of the models we use to date we have about 14 different major techniques we use or based on the structured information we've always gotten through our transaction stream and big benefits there. Right. What I see with is generally the generative AI capabilities.
00:21:11:34 - 00:21:31:28
We have some great ways of working with unstructured, having much more of a human and natural language interface. So what are the hard problems to solve that? So I usually use three categories okay. Which is just toil. You do it. You got to do it again. You got to do it again. How can you eliminate toil. Because you can offload that to a system that can be reliable.
00:21:31:33 - 00:21:52:56
Okay. Lots of techniques can apply that. Some of this is coming in the award we won. So routing inbound right. Agent better has taking toil. Yeah. Much of what you were talking about and where I think most of the advantages are day is the human computer interaction. How does it aid people in doing what they do and not replacing it, but augmenting.
00:21:53:01 - 00:22:17:09
And what we've done is broken everything into what we're calling workbenches. So we have a data science workbench. In the Software Engineering Workbench you have a knowledge worker workbench. And quite often the software you buy for the information systems you want to have surrounds that are benefiting greatly from generally. Right. But I would say rather than trying to save 10% of coding time, if you gave you developers an hour or more a day, they could actually code.
00:22:17:14 - 00:22:34:08
There's even more benefit there. Okay, I can fully agree for that. And one of the thing I do worry about to the knowledge worker side. Yeah, you know, I'm afraid people are going to use these generative tools to generate massive amount of new text, right? No one will be able to handle. Right? So then use the tools to summarize it.
00:22:34:15 - 00:22:56:17
Yeah. So we don't have any creation of new information. Is is big inflation deflation pretty much the same stuff. Yeah. There's always unintended consequences. Yeah you have to look for it. But for the human computer it's pretty easy to to start quantifying what's the ROI of doing it. What's the benefit of doing that. We see that the third area though is production, not just the productivity for the production side.
00:22:56:22 - 00:23:15:56
And for that one, what are the things you can never do before? That's where I really think the biggest payback for a business is going to be, but that's where you really have to learn how to use things in production, how to make sure they're reliable. And you can have, like the other systems. And one mistake I see people doing is thinking of an AI system with code, right?
00:23:16:02 - 00:23:35:08
You build it, you deploy it. Yeah, it's always the same. Visualize in dynamic environments, the inputs will change outputs. Yeah. To monitor in ways you didn't have before. You need to understand what they're doing in ways you've never had before. So getting those policies right are really important. But I'll give you one example, which I thought was amazing.
00:23:35:13 - 00:23:59:24
Talk a lot about how we use sorry. We also the new tool we put out there called our Authorization Optimizer, a AI based system using some general techniques, but a lot of machine learning work. It looks at why transactions decline and recommends when to retrain. Right? In the last year, 8 billion transactions, $27 billion in sales for our merchants went through the network.
00:23:59:24 - 00:24:24:49
That wouldn't have before because of how we planned. You know, so that's an example of optimizing humans can never so, yeah, at an appropriate scale with pay that's historically harmful. Yes. So I always say you got to separate the productivity stuff from the production environments. And what do you see out of production. And I like this model of just the 12 nation human computer workbenches and the new values.
00:24:24:49 - 00:24:35:55
You can create that you never had before. Fantastic! Ed, it's always a pleasure speaking with you. I look forward to our panel tomorrow, and I thank you so much for joining me today. Thanks for seeing you. Thanks.