As part of our ongoing series to highlight specific presentations from Banking Transformation Week, we want to feature John Derrico, Vice President of Data Strategy at Mastercard. John presented on how to create a sustainable data culture, and he compares data strategies to cooking and ingredients.

You can watch the video or read the transcript below.


Hi, my name is John Derrico. I'm a Vice President of Data Strategy at Mastercard. And today I'm going to share a little bit with you about breaking down silos and creating a sustainable internal data culture.

Before we jump into it, I want to share how I've had the opportunity to work at some startups, some Fortune 500 companies, as well as government and public agencies. Everybody has a different play on their data culture, but the one thing that's common amongst all of them is that sometimes there are internal silos or ways that data is not thought about as consumable or reusable. And the whole thing is about what kind of data culture you need to unlock more value, to help you with that speed and innovation.

So first up, I want to talk a little about silos. If you think about a silo, things are just wrapped up, put away, and not able to be shared. But healthy cultures are built on trust. They require people to relate to one another, to help somebody effectively communicate. But even more importantly, being authentic really helps different teams and groups have trust.

And the quote I have here is about how we potentially use food to help build trust. Whether it be in a business meeting or other settings, food is always there and you can bond around it. I really think that food is at the core of just about everything.

For me, I'm Italian. I love food, grew up having the seven fishes for the holiday dinners and things like that. I remember my mom with those pastries, tri-colored cookies, and being that young kid and just enjoying the food.

I'm lucky enough to have one of my sons also enjoy cooking. A few months ago we had an interesting time with one of his baking recipes and we needed vanilla, and, go figure, just when you need something, it's not in the house. So, true story: I have to go out to the store and we have to buy vanilla. This was a really interesting experience because, if you didn't know, there is pure vanilla, vanilla extract, vanilla paste, and vanilla beans. It's like, okay, now I've got all these types. Then I've got brands. Is it ethically sourced and is a pure versus synthetic? And what I didn't know is that it's freaking expensive. Vanilla costs like 10 times more than it did a few years ago. Vanilla is sourced from Madagascar and like one kilogram of vanilla, like 600 bucks. So I was trying to deal with different stores, and we finally got what we needed.

If you think about it, food and ingredients all have qualities. Just as that example I have with vanilla, with my son, I'm also going to try to get you to understand that data also has qualities. Data can be just like the ingredients that we cook with. Not all ingredients are the same, nor is the data. Data could be at a different grade, or, from the food point of view, it could be wrong, it could be stale, or it could be expired.

If you think about it from a food analogy, I'm going to that tri-color cookie, those three pieces put together. Or vanilla. Vanilla can be used in frosting as well as the cake.

When it comes to accessibility, some ingredients are easy to find, and some are not. Same thing with data. But then even more importantly, the data that's fit for purpose. This is more about the ingredients that you need to follow. It's super important. So if you're going to deliver value upon all this, one of the ways we want to talk about this is how you start to break down those silos. And what I've shared here is for you to understand, if you think about the data at the hub and the silos are all the teams that consume that data. There's a lot of challenges with this model and the challenges get into the organization.

Things can be graphically disperse at times. Users have different needs. Each team may have their own sourcing departments. And then there’s how your data sourcing strategy fits in to unlock all of this. How do we connect the users in a way that benefits them to speed up innovation? You need these consumable data capabilities and you need to understand which utilities are super important for us to meet the needs of innovation.

Think about a supermarket of organically sourced data. It’s accessible and consumable in a sustainable way to different teams. With this, we've got vendors who provide data and are vetted. It makes things a lot easier. Not that you don't want to call a salesperson and deal with presales and find out everything you need to find, to get an NDA and do all this, all to get a sample to see if this data is going to fit your needs.

Being transparent, Amazon has a data marketplace and it's an external data marketplace. It's not an internal one. And there's nothing against any of these marketplaces. They just have a different focus and a different need. But if you think about the internal one is trying to provide the best data at the data that's ethically sourced where vendors are vetted against corporate data responsibilities and that the utility of that data is understood.

The next piece of this is how do we make this happen? The data champions, I feel, are key to making this happen. To empower all of those who need to access the data to understand what's fit for purpose. If you think about before we're talking about reading a product label, it's not really the best experience. How do we really understand and interpret the data that's for a specific use.

At Mastercard, we have a process whereby we can deliver insights, the data for our users. And some of this was born on answering our own, our own needs that we need to do for ourselves. And it grew into something really amazing. So I was working for the product owner for example, and the product owner was explaining to me why the data was not very good. I was like, “Okay, what's wrong with the data?” They said, “I need it to be 80%.” And I'm like, “80% of what field?”

After going back and forth for a while, we got a clear understanding of what he needed to launch this product in a given country. He needed to understand the data, its usefulness for his product, and the quality of that data along the way. So it’s not just stuff like fill rates, but it goes beyond the threshold of what can’t and can be used for the data and the utility gets into the common way for us to look at the data (for example, geolocation information for our merchant). If you're building something like an ATM locator, you need to know that precise location. If you're doing some insights, maybe you need the zip code. So the utility of the data — it's not just how it's filled, but how it's going to be used that’s super important.

Data champions are the ones who provide tremendous value to make all this happen. Data champions are folks on your teams who you may go to when you have a question on the data. Sometimes they’re embedded in other teams that have that “can-vs-cannot” attitude. And they’re super helpful because they federate these ideas. There's nothing against centers of excellence, but data champions really helped carry this forward.

This way it's a little bit closer to the users and the utility. And I think about this in a four step model, users can derive insights, you know, from quality and utility. But then also if they're a little bit more advanced on the business intelligence, maybe they can ask more questions of the data using some NLP tools. And there are different BI tools you could use along the way.

From there, there are other sets of users who maybe want to extend these dashboards and different concepts to explore the data. And then other ones may even want to build their own with these capabilities for loan insights. And the champions are the ones who we go to to pull all those other users in to really make this happen.

In summary, there are three main pieces that we see here about maintaining and growing that proper data culture, which I'm calling sustainable. And it's about building trust. Using the example of the food analogy, it's a clear dialogue that makes sense on what the data is and how to understand it. And the value is the second piece, the utility of that information, the accessibility of it delivers something in those data capabilities.

And then as you're doing this, who are your go-to people that you need to work with — those data champions who really help make this dream a reality for a lot of folks. And the quote is here is one from Julia Child, and it's something that I saw in a cooking book that my mom had. I think it was The Joy of Cooking and it was really neat because it talks about doing things that are not fancy or complicated — just good food from fresh ingredients. And that’s analogous to what we're doing. Some of the models we may build may not be as easily understood or may be overengineered.

And there's a quote we have with one of the data science labs I was managing a while back. We used to call it cheap learning versus deep learning. We don't always need to overcomplicate it. Taking that to heart is a really unique way to try to approach some of the challenges, the problems we have to just use fresh data and don't overcomplicate it.

In closing, I do want to talk just a little bit about Mastercard's data responsibility initiative. When we think about that, the data has the potential to fuel the next generation of innovation, but it's only if the organization's data practices are held to the high standards that we deserve. So the individual data rights of the folks to own it, control it, benefit from it, but we have to protect it. You could learn more about Mastercard's corporate data responsibilities. You find things online as well. But I wanted to just, again, thank you for your time. I’m John Derrico from Mastercard, and I hope we had a good time today.