Subscribe
Share
Share
Embed
Unlocking Leadership, previously Leadership 2020, is a podcast helping leadership lead in a world that is changing ever quickly. Join us as we interview even more inspiring people who provide information and skills on how to tackle the big questions affecting today’s leaders.
We blend real-life leadership experiences of our guests with the latest management theory to provide practical, relevant tips for anyone in a leadership position.
Unlocking Leadership asks the big questions about being a better leader in the modern workplace. Hosted by Clare Carpenter.
[00:00:00] Clare Carpenter: Welcome to Unlocking Leadership, a podcast about leading in a changing world, brought to you by Corndel, your strategic skills partner. I'm your host, Claire Carpenter. I'm joined today by Dan Kellett. Dan is the Chief Data Officer at Capital One UK, where he has worked for over 22 years, I think. Hello, Dan, thanks for joining us.
[00:00:31] Dan Kellet: Hi, Claire. Thank you for having me on your podcast.
[00:00:33] Clare Carpenter: Yeah, it's an absolute joy. We're going to get into the nuts and bolts of our conversation in a moment. But before we do any of that, let's find out a bit more about you, please. Who are you? How are you? How are you landing? Fill us in.
[00:00:46] Dan Kellet: Great. I'm great. Thank you. Yes, I'm doing well. I am a data scientist by training. I've always had a kind of deep interest in mathematics and statistics. I'd say my background is in mathematics. I joined Capital One straight from university, so I came out of university looking at what do I want to do in my maths background. Capital One was an emerging company at that point, and so I joined as a graduate, and I've been here ever since.
[00:01:17] Clare Carpenter: Wow. And so how have they kept you there? What's the sort of history of your time with them?
[00:01:24] Dan Kellet: It's always been a really exciting role, and I think part of the reason why I've been here so long, one is actually about the way that the role has evolved and my career has been able to evolve with that, and so I joined the graduates and as a graduate statistician, so my early roles were all around, how do we use all this amazing data that we have available to us to make better predictions to understand the way that we interact with our customers and to build models and to do that kind of analysis and so that was where I really cut my teeth as a graduate statistician. Progressed through various team leadership roles and probably about 15 years ago took on a role looking after the overall, what was statistician team at that time. About that point that you saw this real explosion of big data or kind of insert buzzword here.
But the challenge there was how do we build on some of those core skills that statisticians have as we expand out into, to what a data scientist could deliver, and so that's how do you build out maybe some of those machine learning capabilities or a better software engineering skill set, and actually that was one of the most exciting points of my career was to kind of build out this statistician team, add additional skills and just grow that into a real data science capability and then I guess the other big point from my career perspective was probably about three years ago where I moved into this chief data officer role for the UK business. The UK had never really had a chief data officer before, and so it was an opportunity to bring together the various different parts of the organisation who have lots of data skills, but provide this unified strategy and a unified center of excellence for those skills to, to flourish, and so I guess, kind of going back to your original question, but one reason why I've been there so long is just actually that the way that's evolved and there's always been a new exciting challenge that, that has kept me interested.
I think the other reason why I've stayed here is actually people, and so the opportunity to learn from so many people with different experiences, but also to build great, exciting, diverse, fun teams. Again that's a big part of what interests me and what has kept me here for so long.
[00:03:58] Clare Carpenter: But there's a fascinating connection and separation between those two things, isn't there? There's something around the idea that sometimes people have about data scientists, statisticians, that is, you're immersed in this sort of, sea of big data, and then you've just said you have this interest in people and building diverse and exciting teams.
How did they come together? How did you grow both those skills simultaneously?
[00:04:26] Dan Kellet: I think here it's a key part of that is, is finding things that interest you, to be honest. I love the kind of theoretical side. I'm interested in where things go. But actually, the thing that really interests me is how do you take some of that theoretical data science knowledge or experience and actually to apply it to a real problem all the way through to actually delivering something impactful, something in market and the only way you do that successfully is by combining those technical skills with the people side, because if you don't crack that, people, those interactions, you might have the best idea in the world. But it's not going to work, and so actually, I think those two aspects are probably pretty critical for a successful CDO to actually deliver real results in market.
market
[00:05:19] Clare Carpenter: So as you have moved forward in terms of seniority within the organisation, what's been important for you from your background to also add to your skill set to, I guess, make data and the work that you and your teams are doing more accessible to a broader range of leadership.
[00:05:39] Dan Kellet: You may well have kind of hit the nail on the head there with exactly what it's all about, because I think so many organisations decide that they need data science without necessarily working out why, or working through what it means from a foundational level, or even how do you talk about this and align it to your business strategy.
So I definitely find that a large amount of my time is around connecting those dots. It's about taking the technical knowledge of the team and making sure that we can reflect that in a way that stakeholders understand. So you might talk about a huge amount of data. But actually, unless you're able to turn that into some insight or kind of a new strategy, it's just data and actually it's not really an asset at that point for the organisation. It only becomes an asset when you start to go, well, what's this telling me and how do I react to that to improve my strategy?
[00:06:43] Clare Carpenter: So just coming back to the fundamentals of data science, how would you describe that to somebody who really had no idea what that means?
[00:06:53] Dan Kellet: I think it's all around looking at the information that's all around us. So every time anybody does anything, whether it's kind of looking on a website or opening their phone or any kind of interaction, you create pieces of information. Now, a large amount of that information could just sit there, but actually I think a data scientist's skill is how do you pull together all those pieces of information and actually find patterns or signals that help you better understand the world.
If you better understand the world, then actually. You're able to adapt and evolve your approach so that you end up with a better outcome. So whether that's kind of growing your business or having greater customer engagement, kind of whatever your outcome is there, it's how do you use these pieces of information to better illuminate your world.
[00:07:49] Clare Carpenter: Yeah. By thinking about an organisation like Capital One, which must have an enormous amount of data available to it, over an extended period of time now, and not just in the UK, of course, but from a global point of view, how do you stop your team and yourself becoming swamped by that amount of data?
[00:08:16] Dan Kellet: Yes, definitely, and so if I just focus on our UK business, for example, so we have somewhere north of 4 million active customers who are using credit cards on a daily basis, lots of transactions, lots of payments, lots of interactions with us through their mobile app or through our web servicing platform, every time one of those things happens you're generating data and it is intensely a bit of a rabbit hole. You're right there's a huge amount of data there, and I kind of, I have a saying that I say quite often, which is, you have to be careful, you're not just adding more hay. You want to be adding more needles to that haystack as well, if that's what you're really trying to find, and I think the transactions is a really interesting area because we have information about where our customers are shopping and when and how much.
That's really fascinating. That's really interesting. But unless you're able to tie that to a so what. What's that telling me, you could spend weeks and weeks looking at that data, finding lots of interesting things, but are they useful things? And I think that's kind of a key skill for a data scientist.
[00:09:28] Clare Carpenter: I imagine that's changed quite considerably since you've been specialising in this field as well. So 20 years ago, I guess, the landscape of interaction with customers was very different, wasn't it?
[00:09:39] Dan Kellet: Yes. Yes, and in, I would say three distinct ways, one is speed. So the speed of which our customers interact with us has dramatically changed. If I go back to when I started as a graduate, the majority of communication was either written, so kind of letters or via the phone. More and more our customers are interacting with us directly through the mobile application or through our web service platform and with that comes an expectation of that speed of response in return as well. So there's definitely kind of a ramp up in the speed of turnaround that's required. The second area is the volume. The, I think with the speed of that interaction, just the amount of data that gets generated has just multiplied dramatically over that period of time and then I think the third area is perhaps more specific to financial services, which is perhaps more around the regulation side and how we think about protecting our data making sure that it's being used in an appropriate way, and also actually that we have good policies for deleting our data when we no longer need it so, I think that's a dramatic shift again over the last 20 years.
[00:11:01] Clare Carpenter: Yeah. I wonder what other shifts you've seen in terms of how data now plays more of a part in the strategy of your organisation and indeed in the decision making of the rest of your UK board and how you move forward.
[00:11:19] Dan Kellet: I think for Capital One in banking terms is not a particularly old organisation. Capital One has in itself been around for 30 years. At the core of the foundation of Capital One was this focus on data and analytics, test and learn and really how do you embed those skills into the key decisions that you make.
Now, one of the upshots of that culture is actually our board are very numerous, our board have a really good understanding of how do you use data to make decisions, and so that is very helpful as a chief data officer, that's definitely something that helps. But I think one of the key requirements on me is how do I stay abreast of developments in the world of data and in the world of data science, so that I can better advise the board on where do we go next?
Well, what should our investment plan be over the next five years, next 10 years? And so I think data is probably more at the forefront now than it ever has been, and it's probably better aligned with our business strategy than it ever has been before.
[00:12:30] Clare Carpenter: Yeah, I'm thinking about how unpredictable the last few years must have been in your space, in particular, as you saw habits changing, perhaps through the pandemic coming out of it. This sense of, you know, really dramatic events in the UK economy and worldwide that, you know, five years ago, I guess none of us were really in a place to predict, were we? How does data help an organisation like Capital One respond to those really unexpected changes?
[00:13:04] Dan Kellet: I think it's a good example of investing in your infrastructure in processes in the good times so that you're able to react at times of emergency or at times of change. I think it's definitely fair that, that kind of, if I could throw back three years ago, the majority of my team were spending a lot of their time actually just combing the horizon. So where do we see the economy? What are the impacts of some of the governmental initiatives on the way that consumers were lending and paying back? And we saw all kinds of patterns that, that you're right, we'd never seen before. A big question for the team at that point was, okay, if the data is changing, what does that mean for the models that are built on our data? And then the decisions that rely on those models, and so a large amount of our time was spent on, do we still trust these models? Do we still trust these predictions? Now we actually ended up in a good spot there for a couple of reasons. One, I think actually our models are pretty robust to these things. We try and test our models through different economic situations, which means we feel generally pretty confident in their ability to weather change.
But the other is actually. More of a kind of process side, which is how quickly can you get to that information, process it and come back with a recommendation that says, yes, we feel okay or not. I think if you're in a situation where perhaps you haven't invested in your data infrastructure, when you need to react to those things, it's a lot harder, you have to jump through more hoops or kind of stitch different data sources together to get to that answer. Fortunately, we were in a position where we invested in our skills, capability infrastructure and we're able to get some of those insights reasonably quickly and react to those.
[00:14:58] Clare Carpenter: I'm thinking about something you said earlier about your joint interest within the world of data science and statistics and in people and people development and growth. How do you identify then talent of the future who would have a keen interest as you do in that space of statistics and data science. How do you know at an early stage when you meet somebody? Yeah, that's somebody who's gonna have a future in this particular area.
[00:15:29] Dan Kellet: It's a question I often get asked when I'm talking to, maybe I'm talking to undergraduate groups and one of the questions often comes up is like how do we get a role? What are you looking for from a great data professional? And I think it's easy to get drawn into a list of technical requirements.
So we're looking for someone who knows Python, or we're looking for someone who has a background in statistical test design, and all those things are definitely useful and important. But actually I would say kind of the biggest indicator is curiosity, I think the curiosity to ask questions and bring together something you've seen over here and go, Oh, I can apply it over there and that kind of mindset actually is really valuable and actually kind of sounds out from a mile away. We can work on providing some of those technical skillsets. I think its really hard to grow that curiosity, and so actually someone who's got that deep need to understand why I think that's a, the makings of a really great data professional.
[00:16:37] Clare Carpenter: I think that's really fascinating I love that sense of curiosity and that for me is bringing the science part of data science alive. It's that exploratory approach to what the statistics are telling you, rather than an assumption, isn't it?
[00:16:53] Dan Kellet: It is because I think if you don't change anything, nothing ever improves, and so that is why that curiosity is so important. I mean, obviously you need to make sure that you kind of bound that and it has some good scientific theory behind it, but actually kind of say that idea generation, that ability to go, oh, what if. Super important.
[00:17:15] Clare Carpenter: Yeah, I can really see that. Where do you think the future of your field of expertise is? Where's it going next, data science?
[00:17:26] Dan Kellet: That's a big one. That's definitely a big question. I think there will be peaks and troughs in the world of data, depending on where people sit on the hype cycle. Now, you know, you can't pick up a newspaper or look at a website without some discussion at the moment of generative AI, and that's exciting. That's definitely exciting as a data scientist, as a data that we're seeing these advances. But advances and evolutions are very rarely linear, and so you have these kind of big spikes of excitement followed by perhaps a bit of disillusionment or some kind of adaption. I think the data scientist of the future needs to be able to kind of ride some of those waves in order to really sell the benefits.
But also have a real focus on the actual pragmatic delivery. I think what worries me as a data scientist I love data science. I would want to continue doing data science. My worry sometimes is that hype overplays the impact, and so a big part of my role is actually going, well, no, what is it that we're really trying to solve here? And how do I use these exciting advances, but in a pragmatic way that really kind of pushes our business forward or delivers a better customer outcome or better helps us meet our regulatory requirements. So it's tying, kind of all these exciting things that happen over here with the needs of the business or the needs of the customer.
[00:19:04] Clare Carpenter: Yeah, and I guess you must be experiencing as well, more interest in your area from leaders in other parts of your business. I'm thinking in particular about how HR directors now are demanding more people analytics in terms of onboarding, attrition levels, background, all of those kinds of things.
Obviously there's a strong link between your team and the financial performance of the organisation from a customer and revenue point of view. Where do you see other growth areas of interest in using data and data analytics, I guess, in particular in your organisation?
[00:19:47] Dan Kellet: I think they're two really good areas and that's definitely some of the focuses of the team. I think the people analytics bit in particular is something that we looked at five or six years ago. We worked with our head of HR to really build out the case for what could be possible here, and actually, you know, we have a data scientist who is fully resourced towards people analytics and helping us understand the dynamics of our associates and our employees.
I think a particular area of interest is potentially around the automation side. So again, we have a small automation team that looks at how do we potentially replicate some of those really manual changes that happen on a regular basis. In a way that actually is kind of maybe more well controlled or more efficient. So that's a whole area. But then I think the other area of interest for me is we have all these different interactions with customers. How do we bring our lab together to help make our operational department really effective or really insightful? So how do we bring more of those kind of day to day interactions with customers so that they bubble up in a way that's insightful and impactful.
[00:21:00] Clare Carpenter: I guess there's an associated need to present that finding and data output in a way that is really compelling to your internal and external customer base.
[00:21:11] Dan Kellet: Yes, and I think they, with that, you're starting to get into the realms of how do you think about data literacy and data democratisation there as well? Because definitely I find as you look to try and expand the remit for what a data department can provide. There's so many options. There's so many potential things that you could work on that, to be honest, we just don't have the people to be able to do all the things that I would love to do and so I think the answer there is how do you use data or platforms or tools to actually just make that, those insights more accessible to everybody. So that you don't need to know large amounts of Python to be able to get to particular pieces of information, and so again that's a big part of my role is thinking about how do we make those things more available. Make it so that you don't need to go through seven levels of access or 16 different training courses to have the skills, to be able to get to some of those kind of core metrics or core pieces of information that would benefit everybody in the organisation.
[00:22:20] Clare Carpenter: Yeah, it's fascinating. I think data literacy is a really interesting area to explore perhaps in a little bit more detail because I've spoken to a guest of this podcast previously who talked really compellingly about data literacy and actually shockingly about the levels of data literacy generally in the UK population being really low in comparison to what's required by organisations like yours.
[00:22:47] Dan Kellet: Yes and we're no different there, you know, there is a certain amount of data literacy and broader numeracy required in pretty much every role at Capital One, right? everybody works with numbers or data in some form or another, and I think the challenge is how do we give people the support to be able to up some of those skill levels in a way that is nonjudgmental, but also in a way that it's really effective and actually focused on kind of the pragmatic use of that data and those numbers.
If you crack that, it has a whole load of benefits for your organisation. But also it has a whole load of benefits for your employees as well. It opens up much more mobility when it comes to different career choices, different options, and also kind of, it helps with your retention and general engagement as well. So yeah it's definitely an area I'm really passionate about.
[00:23:42] Clare Carpenter: Yeah, I know that you're committed to ongoing professional development, as I may or may not have stalked your LinkedIn profile. However, I wonder what you would recommend to people who are new to this area, perhaps who are emerging leaders in other parts of organisations, not specialist in data science or data analytics. What's important for them to do to, I guess, grow their experience and potential in this as a leadership skill?
[00:24:12] Dan Kellet: I think there's a few things. I think my first piece of advice is don't be scared. I think can be seen as quite an impenetrable area. There's lots of terminology. People are more than willing to kind of throw their PhDs into the ring and say, Hey, I know this. I'm about this. The fundamentals of good data management, data science, data analysis are pretty straightforward. I think it's about making sure your data is in a good state. You understand where it's come from. You understand what you're trying to achieve, and then you apply some techniques to it. More and more, I'm finding actually the application of the techniques. There's loads of great stuff out there.
There's lots of kind of training courses, online communities. So that's no longer the barrier that I think it might have been historically. Lots of good things that kind of work you through case studies, most of which are open source, and so actually there is some really easy ways in to that. If you look for those things.
I think the second thing that I'd recommend is finding some good mentors, kind of building out your network. I find LinkedIn an incredibly useful resource there. I think the data community more broadly is very generous of their time, and actually kind of we're all data geeks and it's our... what we're trying to do is kind of share that love really, and so, so, you know, I found that my network is very generous when I'm a bit stuck and asking for different opinions, and I think anybody who looks to try and step into the world of data will probably find the same.
[00:25:47] Clare Carpenter: Yeah. We haven't talked about the relationship of, I guess, this notion of data democratisation. I love that as a phrase and its impact on diversity in your recruitment space. Are there parts of our employee community who you think benefit enormously from that sense of it being more open and available now?
[00:26:13] Dan Kellet: Definitely, and we've seen some real examples of that over the last few years. If I skip back kind of 10 years or so, I think our approach was typically graduate focused and probably very numerate degree focused as well. So we brought statisticians in who maybe had statistical backgrounds, master's, PhD level. More and more we're finding actually you can teach those skills and going back to my curiosity point, if you're finding the right person, you can kind of help them grow. One of the benefits that we found recently is actually opening up roles within the data organization to people who might work elsewhere in our operation. So people who might be on the phone with our customers, we opened up a series of roles, allowed people to apply and move into those data roles where we said, there's no expectation that you know how to code here. There's no expectation you have statistical knowledge. We will teach you that. What we do know that you have is a really good understanding of Capital One's customers and our processes, and also kind of the right attitude that says, Hey, I want to try something different. I want to learn something. That's been an incredibly successful program for us. It's brought in a group of people into the team who bring different opinions, different perspectives, and they're just kind of really fresh in depth the overall department. So my advice to any data leader is actually don't overlook some of the amazing people who you have elsewhere in your organisation, you might just be looking for that opportunity to hop into a data career.
[00:27:50] Clare Carpenter: Yeah, I'm thinking that we all read these articles and blogs about the roles of the future don't even exist today, and I guess 22 years ago when you joined Capital One, perhaps you didn't know that the chief data officer role was going to have your name attached to it a little bit later. What do you think is the future of how data is going to move forward in terms of its, both its maturity and also the potential that it has to open up new directions for people?
[00:28:17] Dan Kellet: I think you may find that the data roles, Chief Data Officer roles, for example, start to split out a bit. Depending on specialism and depending on the need of the organisation, I think you already find that there are perhaps some chief data officers who are more focused on getting the foundations in place, a data quality, data management, whereas other data officers might work in their more monetisation angle, so okay, we've got this data. How do we best deploy it? I think those two areas will still be really important. I think there's also a growing importance and need, I think, for. some of that, whether it's data ethics or whether it's kind of how you combine artificial intelligence with human decisioning. I think that's a data and people challenge to be solved and what you may find is more and more organisations start to put people in thin roles to actually kind of address some of those things.
[00:29:17] Clare Carpenter: One of the skills that leaders have, I think, in their portfolios is capacity to storytell in a compelling way around their area of passion. Now you are a self confessed data geek. I wonder as we come towards the end of our conversation, if you would like to share one of your best stories with me?
[00:29:37] Dan Kellet: Yes, of course. Definitely. So I talked a bit about transactions before and the way that customers might spend in different shops for different amounts, that is my own personal favorite data source, and I maybe spent way longer looking at some of that data than I should have done. But actually one of the things that's really interesting is how do different shops or these customers shopping in different shops correlate to each other, and it helps you build out a really good understanding of different types of customers and maybe how they're using their Capital One card. So one of the things that I've really liked that we've done over the past is some kind of clustering of these merchants, these shops.
So let's go out and look at all our customer transactions and try and find the links between different stores. So, okay, people who shop at store A are much more likely also to shop at store B. That was really fascinating to us because I'm not sure what I expected in terms of how strong some of those links were, but actually we found some really distinct types of customers.
We found customers who will use their cards on everything I spend, so they'll, you know, it's their weekly shop. It's it's kind of everything that they're going into town at the weekend and buying. It's their Amazon spend, and this is kind of their primary card. They use it all the time. We have a group of customers who go, well, actually this is my card for my holidays.
So I'll book my flights on it and my hotels, and then actually what's great about this card is it allows me then to pay that off over, over the course of the next three months let's say, and then you might also have a second group, a third group of customers who. This is their card for online purchases. So I kind of have this stored on my computer or my phone or whatever, and that's the thing that I trust to make online purchases. Now you might communicate to all three of those groups of customers in completely different ways, or they're interested in different things, and actually that's a way in which the data has shown you that actually you've got these different pockets of customers and so it's allowed you to create a bit more structure, a bit more understanding about all those customers you have and how you might talk to them. So that's one of my favorite things. Like it's a, you use data source and you know, you have to do lots of cleaning to get to that point, but actually once you've done that, it's really insightful.
[00:32:06] Clare Carpenter: I can imagine how that is. I'm sort of slightly worried about potential of someone tracking me in my shops and where they go, where I go, but you know, that's okay. I'm all right with that as long as you communicate with me appropriately afterwards.
[00:32:19] Dan Kellet: Exactly, and I think that's really important is whatever we do, we always need to bear in mind the angle of how is this in the customer's best interest. You definitely want to avoid getting into some of those awkward situations of using data, and that's why I think it's really important to be clear, why are you doing this, to what end, and how does it link to the customer's best interest?
[00:32:43] Clare Carpenter: Oh, how interesting now we're into data ethics which is an entirely different podcast probably. As we reach the end of our conversation, then I wonder if you might have some words of wisdom to offer somebody about to embark in the world of data science and data analytics as they step out of university or into an apprenticeship, perhaps from a practical learning point of view, what would you say to them as they start their career in leadership and in this area?
[00:33:15] Dan Kellet: So, two things. I think firstly, build a network and use it. Ask lots of questions, don't be afraid of putting your hand up and going, I don't get this, can you explain it again? Because the likelihood is... Quite a large number of people will also be thinking the same, so kind of ask as many questions as you can, build that network and then I think the second thing is data can be really fun. It can be like, you know, it can open up lots of opportunities. It can present you with all kinds of chances to do new things or go to different places. Just enjoy it.
[00:33:51] Clare Carpenter: I love that. Enjoy it. What a wonderful way to end our conversation. Thank you so much for joining me today. It's been an absolute pleasure. Thank you.
[00:33:59] Dan Kellet: Thank you. It's been great.
[00:34:02] Clare Carpenter: Thanks for listening. If you've enjoyed this episode of Unlocking Leadership, you can subscribe through all the regular podcast channels, and please do leave us a rating and review there. We'd also love you to share any episodes you've found interesting so that others can join the conversation and share their experiences.
This podcast was made in association with Corndel. It was produced and edited by Story Ninety-Four.