 from the Corinium Chief Analytics Officer Conference, Spring, San Francisco. It's theCUBE. Hey, welcome back, everybody. Jeff Frick here with theCUBE. We're at Downtown San Francisco at the Corinium Chief Analytics Officer Spring Event 2018. About 100 people, really intimate, a lot of practitioners sharing best practices about how they got started, how are they really leveraging data and becoming digitally transformed, analytic driven, data driven. We're excited to have Vishal Morday. He's the VP of Data Science at Barclays. Welcome. Glad to be here. Absolutely. So we were just talking about Philly, you're back in Delaware. And you actually had a session yesterday talking about Barclays journey. So I wonder if you could share some of the highlights of that story with us. Absolutely. So I had a talk, I opened the conference with Data Science journey at Barclays. And we have been on this journey for five years now where we transform our data and analytics practices and really harness the power of big data, machine learning, and advanced analytics. And the whole idea was to use this power of newly found power that we have to make the customer journey better. Better through predictive models, better through deeper and richer consumer insights, and better through more personalized customer experience. So that was the story. Now it's interesting because we think of financial services as being a data driven organization already. You guys are way ahead. Obviously Wall Street's trading on microseconds. So what was different about this digital transformation than what you've been doing for the past? I think the key was we do have all the data in the world. If you think about it, banks know everything about you. We have our demographic data, behaviors data, from very granular credit card transactions data. We have your attitudinal data. But what we quickly found out that we did not have a strategy to use that data well, to improve our own productivity, profitability of a business, and make the customer experience better. So what we did was step one was developing a comprehensive data strategy, and that was all about organizing, democratizing, and monetizing our data assets, right? And step two was then we went about the monetization part in very disciplined way. We built a data science lab where we can quickly do a lot of rapid prototyping, look at any idea in machine learning data science, incubate it, validate it, and finally it was ready for production. So I'm curious on that first stage, right? So you've got all this data, you've been collecting it forever. Suddenly now you're going to take an organized approach to it. What did you find in that first step when you actually tried to put a little synthesis and process around what you already had? Well the biggest challenge was the data came from different sources, right? So we do have a lot of internal data assets, but we are in the business with, you know, we do have get a lot of internal data. Think about the credit bureaus, right? Also we have a co-brand business where we work with partners like Uber. Imagine the kind of data we get from them. We have data from American Airlines. So idea was to create a data governance structure. We formed a chief data office, officer kind of forum. We got all the people across the organization to get understand the value of data. We are data driven company as you said, but it took us a while to, you know, take that approach and importance of data. And you know, then data and analytics need to be embedded in the organizational DNA. And that's what we kind of focus on first. Get the awareness of importance of data, importance of governance as well. And then we could think about democratizing and monetizing, organization is the key for us. Right, right, so how did you organize? How was the chief data officer? What did he or she, who did he or she report to? How did you organize? Right, so it was directly reporting to our, you know, CEO. Into the CEO, not into the CEO. Not into the CEO. We had a technology office. We do kind of have a line of sight or dotted line with technology, but we made sure that that office has a lot of high level organization buy-in. They were given budgets to make sure that the data governance was in place. Key was to get data ownership going, right? We were using a lot of data, but there are no data ownership. And that was the key. Once we know that who actually owned this data, then you can establish a governance framework. Then you can establish how do we use this data and then how do we monetize it. So who owned it before you went through this exercise? It just kind of, it was just kind of there. Yeah, there wasn't a clear ownership, right? And that's the key for us. Once you establish ownership, then it becomes asset. You are not treating data as an asset. So that was a change in kind of mindset that we had to go through. That data is an asset asset. And it was used as a means to an end rather than an asset. Right. What about the conflict with the governance people? Because I'm sure there was a lot of, wait, wait, wait, we just can't open this up to just anybody. Sure, pretty interesting discussion because you have to open it up to more people, but you still have to obviously follow the regs. Right, and that's where there are a lot of interesting advancement in data science where in the area of data governance, right? There are new tools out there which lets you track who's actually accessing your data, right? So once you have that infrastructure, then you can start figuring out, okay, how do we allow access? How do we actually proliferate that data across different levels of the organization? Because data needs to be in the hands of decision makers, no matter who they are. Like, could be our CEO, or somebody who's taking our phone calls. So that democratization piece became so important and then we can think about how do you, you can't directly jump into monetization phase before you get all the ducks in order. So what was the hardest part, the biggest challenge of that first phase in organizing the data? Creating that 360 degree view on our customers. We had a lot of interesting internal data assets, but we were missing big pieces of the puzzles where we're looking at, you're trying to create a 360 degree view on a customer, it does take a while to get that right. And that's where the setting of the data governance piece, setting of the CDO office, some of those are the more painful, more difficult challenges, but they lay the foundation for all the work that we wanted to do. And it allowed us to kind of think through more methodically about our problems and it established the foundation that now we can take any idea and use it and monetize it for them. So it's interesting, you said you've been on this journey for five years. So from zero to a hundred, where are you on your journey do you think? Right, I think we're just barely scratching the surface. Just barely scratching. I really barely. I think you're good at this. Yeah. Because I do feel that the data science feel itself is evolving. I look at data science as like ever evolving, ever mutating kind of beast, right? And we just started our journey. I think we are off to a good start. We have really good use cases. We started using the data well. We have established importance of data and now we are operationalized. Some of the machine learning data science projects as well. So that's been great, but I do feel there's a lot of untapped potential in this year. Right. And I think it'll only get better. Then what about on the democratization? We just in the keynote today, there was a very large retailer. I think you said he had 50 PhDs on staff and 150 data science. This is a multi-billion dollar retailer. Right. How do you guys deal with the resource constraints of your own data science team and PhDs versus trying to democratize the decision making out to a much broader set of people? So I think the way we thought about this is start, release, think big, but start small, right? And what we did was created a data science lab. So what it allowed us to kind of, and it was a cross-functional team of data scientists, data engineers, software developers kind of working together. And that was a primary group. And there was ably supported by InfoSec guys or data governance folks, so they're a good support group as well. And with that cross-functional team, now we were able to move from generating an idea to incubating it, making sure that it has a true commercial value. And once we established that, then only we move forward with operationalization. So it was more surgical approach rather than spending millions and millions of dollars on something that we are not really sure about. So that did help us to manage a resource constraint. Now, only the successful concepts were actually taken through operationalization. And before operationalization, we truly knew the bottom line impact. We could know that here's what it means for us and for our consumers. So that's the approach that we took. So we're going to leave it there, but I want to give you the last word. What advice would you give for a peer not in the financial services industry? They're not watching this. But you know, in terms of doing this journey, because obviously it's a big investment. You've been at it for five years and saying you barely are getting started. You're in financial services, which is at its base basically in information, technology, industry. What advice do you give your peers? How do they get started? What do they do in the dark days? What's the biggest challenge? Yeah, I feel like my strong belief is data science is a team sport, right? So a lot of people come and ask me, how do we find this unicorn data scientist? And my answer always being that they don't exist, they're figments of imagination. So it's much better to take a cross-functional team with complimentary kind of skill set and get them worked together. How do you fit different pieces of the puzzles together? Will determine the success of the program rather than trying to go really big on something. So that's, the team sport is the key concept here. And if I can get that word out across, that'll be really valid. All right, well thanks for sharing that very useful piece of insight. Absolutely. All right, he's Vishal, I'm Jeff Rick. You are watching theCUBE from the Carinium Chief Analytics Officer Summit San Francisco 2018 at the Park 55. Thanks for watching.