Getting Data Right From Day One | Chief Data Officer @ Dojo | Mike Seville

  Hello, and welcome to a new episode of The Start-Up Diaries, powered by Burns Sheehan, a leading technology recruitment business here in the UK. Welcome to the first episode of Season 7. In this episode, we have Mike Seville, who is the chief data officer of Dojo. This episode, Mike talks us through the impressive data platform that he's built at Dojo.

He then dives into the fundamentals and growing a data team, plus the growing pains to look out for. Following that, we talk through the changes in his CDO role and how that might reflect the future of data. And finally, he talks through building partnerships within an organization to benefit the data team.

We think it's a good one. Hopefully you enjoy it. Thank you very much. Good to meet you. Yeah, thanks for joining us. Do you want to start off by telling us a bit about yourself and your journey that's led you to Dojo? Absolutely, yeah. So hi everyone. I'm Mike. I'm as mentioned, currently Chief Data Officer at Dojo.

I've been in the role now for Getting on for three and a half years, which I think in CDO terms is, makes me a a red breed. I think the average I read recently was about two and a half years. So yeah, it's it's been a heck of a journey over the last over that time we've obviously went through the pandemic and grown a ton as an organization, so it's been an interesting journey, but kind of starting from the beginning, I.

Spent the first four or five years of my career working in e commerce largely in commercial roles. My background prior to that was graduated in physics. I already had a maths computers background really. But decided to move into a commercial role. Worked as an account manager, first at eBay.

And then worked over at a subsidiary of Tesco. It doesn't, isn't with us anymore. Then around about that time, kind of 2011, 2012, started here and get involved in this thing called machine learning, which seemed really interesting and kind of suited that that maths and computer science background that that I had through physics.

And it just became, Absolutely obsessed with it. And it was the, the, the defining mode or the thing that really convinced me that it was something to go after was seeing the very first Google Waymo video. So they put out a self driving car video. It's still available online. It's a converted Toyota Prius.

And there's a gentleman who's a blind been taken for a drive through pizza. And yeah. And it just blew my mind. It just the, the, the potential of the technology just, just really captured my inspired me. And I ended up applying was accepted to study machine learning at UCL back when it was only about, I think, a 25 person course.

So it's now I'm told hundreds of people every year it's a big course. But absolutely loved every minute was fortunate enough to be taught by people that went on to be major Contributors to the field David Silver DeepMind who did AlphaGo was one of our lecturers for instance. So it was a just a really amazing time, I think, to be getting into the the data and machine learning domains.

Just before I left Tesco, I was advised by the MD of the business function there, that if you ever hear of this company called Dunhumbie, they'll they'll be a really interesting place for you to go to because he'd heard about them as an auditor at Tesco and was working with some of their team.

And when I was Doing the project and sorry, doing the masters. I was thinking about what project I'd like to do. And a note wrote around the department saying, would anyone like to to do a project with Dunhumbie? And I thought, okay, this is a weird coincidence. I put my hand up had a chat with Pierce Stobbs, who has since moved on from there and Charles Pavey, who was also at Dunhumbie at the time, huge characters and amazing people.

And yeah, it got on super well with them. Ended up doing my project with them. They kindly offered me a job afterwards and spent two years or so. Working on ML problems at Dunhumbie, kind of figuring out who the customers were, how they were behaving, all the dynamics that were going on in that business.

And then after two years decided to make a move, went to a startup called Farfetch spent Four and a half years there working on first building up the data science team and the part of the analytics group. And then eventually took on the responsibility for all of the marketing data functions.

So including some of the BI, the the analytics, as well as the data science teams that we assembled there. And just had an amazing time. It was a wonderful business. And then, yeah, back in the tail end of 2019 you know, Farfetch had gone through an IPO. Business has started to mature. So, it's a good opportunity to outside and join what was then Paymentsense, now Dojo.

And again, it was a rapidly growing business and it seemed like a great opportunity. So, let's jump onto that. Well, I think when we were preparing for this episode you mentioned that basically the foundations of the products are actually largely driven through data and data is a massive part of what Dojo do.

And it's the future of the business as well, which I think was really interesting comment to make. So can you talk us through this and the data platforms at Dojo? Absolutely. Yeah. So, so I've, one of the reasons I joined Dojo and what inspired me when I sat down and spoke to Jan and George, the two founders was in terms of the opportunities that Payments as an industry and financial services more broadly has very rarely been very successful at capturing the value of the data that gets developed.

And going back to my time at Dunhumby, a large part of my, certainly my master's project when I was there was taking shopping basket data and turning that into value, business value. It was looking at what people put in their shopping baskets and figuring out something about that individual And then using that to then help them discover or help the the manufacturers and retailers to, to, to, to harness and, and, and keep hold of those consumers coming back time and again, nobody's ever done that in payments, really and certainly in the the market we're, we're in where the large part of the the merchant base that, that our dojo and payments as customers is in the hospitality sector and restaurants, bars, pubs that, that, that access to that data is, is unheard of.

I mean, the, the average, your average small, medium sized businesses country has no idea who the consumers are that walking through their business every day. And for me, that, that data asset represents an enormous opportunity. And so when we when we first started sitting down and talking about what the potential of the organization was and and where data fits into the strategy, it seemed a great opportunity to all of us that Figuring out how to pivot the products at some point as the Dojo platform matures and is developed across the country.

Figuring out how we can build the tools and services that help those merchants to understand who their consumers are understand how the business is evolving and changing with respect to the consumers coming in and most importantly, figuring out how to help them attract and retain those consumers was was an amazing opportunity for us, but also for those for those merchants that do business with us.

And so with that in mind, as we were developing the dojo product, we started to look at how we could build out these products and services. And with that, we acquired a walk up, which has now been fully integrated into the dojo portfolio. Back in 2021 now and with a view to starting to build a B to B to C business, providing a platform for consumers to access Dojo merchants, find places that they're going to really enjoy backed by the payments data that we hold recommending products and services you'll like based on what we've seen about your behavior, and then helping those merchants to find consumers who will like their business and, and keep hold of them again in the future.

And this is an evolving kind of product area for us, but it's a significant one. I mean, we're putting a lot of effort into it now, you know, over a hundred thousand people a month for finding places through the platform and it's continued to evolve. So, so it became a very important point for us to, to invest into.

And as I say, those are now he's. I think significant, to a significant extent, driven by the data we're, the data asset we have. Now, you asked me what the, the platform looks like in a technical level. At the moment, it's, it's still very much a I think relatively nascent in terms of the platform we've developed is good enough to serve the use cases that we've developed, but I think it's a long way for us to go as well.

So in the back end, we work on a Google cloud store data lake feeding into big query. We used DBT, we use Airflow, we orchestrate all of our data pipelines through those services, and now increasingly starting to introduce Vertex to help support our machine learning. And as we go forward, we're looking at introducing graph databases and recommendation systems on Tufla to further enhance the consumer experience.

But Yeah, early days still even on the consumer side. So what you're saying is if you're someone in data, this is the place to work, basically, with all the fun tools. I think so. Maybe I'm biased, but I think so. I think one of the things that's happened for you personally as well, your role as CDO has pivoted slightly more recently.

So could you talk us through about the sort of, yeah, what's happened there and why you decided to make the change? Sure, absolutely. So the data function within Dojo at the end of Q1 this year had grown from when I joined from a handful of people up to an organization of 60 to 70 people. So it's a big team, I think, by any standards, certainly relative to the overall size of the organization.

So Dojo is about a thousand people altogether. So it's a significant component. And The organization, the data organization was broken out between engineering data engineering, analytics, engineering BI, the guys responsible for modeling and understanding the data analytics and data science. And it was distributed across the organization and embedded in different ways, depending on supporting different components.

So our sales teams, our marketing teams, operations, consumer services, as I've just described in product. And a large part of our effort within the data team was increasing, had always been supporting a lot of the day to day functions of the organization. And even, even things I think that traditionally a lot of places wouldn't consider data responsibilities fell, sat within our our remit around things like providing reporting directly to consumers, doing reconciliation safeguarding working with our risk teams, all these kind of things became key components of data.

Over the time, I think a large part of my personal focus and a large part of the management team within the data functions focus was towards those kind of tasks. And I think last year, both personally and within the leadership team, It was apparent that we weren't paying enough attention to really the things that are going to develop the value of the organization over that 3, 5, 10 year time horizon that data really has to look to.

We all saw, for instance, what was happening with large language models, GPT 3 and chat GPT what had happened to Google and the kind of surprise that they were almost, the back foot that they were caught on in terms of the disruptive technologies that are starting to emerge. And I think it's apparent to anyone paying attention to this, this, this space that the velocity of change is, is enormous and requires a significant amount of focus.

And, and really what the the, the, the motivation, therefore, was, How can we get more time on those things that really can be both disruptive in a good sense for us But also in a potentially negative sense if we if we're caught on the back foot like like Google has been versus that kind of day to day BIU responsibility that I had.

And so my the focus we've taken therefore is that I'm now focusing entirely on the forward looking things. And we've dispersed the data team across the organization to different reporting lines. So that the executives that they're supporting are now directly managing those people.

And I'm now focusing now. Well, that means for me personally, it means I get to focus on all the fun stuff, which is amazing. And what it means for the team is that they're able to work and be close to the stakeholders that risk that they're accountable to essentially in health of that account, which I think, again, the lines, a lot of the the incentives and the provides them with the support they need, I think, to, What do you think this means for the future of the Chief Data Officer position?

I gave a lot of thought to this question actually, because this is, this is a tricky one. I think I'm, I've gone backwards and forwards in my thinking on this. So I think on the on the one hand, data is only going to become more important. Looking ahead and in the context of AI that we were just talking about, you look at what Reddit announced yesterday, for instance, where they're now going to start charging for access to their data sets.

They're explicitly out or or restricting in their terms and conditions. The use of Reddit data to train a eyes unless you pay them for that access. I think it goes to show that companies very quickly are realizing if you hold a bespoke proprietary data sets, you have an enormous amount of value. As, as I said, we, we, we realized a couple of years ago.

So therefore it is logical to think that the responsibility of a CDO type role is gonna become all the, that much more important especially when coupled with the innovations with of, of, of ai. On the flip side of things, I think there's also a, a, blurring of responsibilities more so than perhaps previously with CTO type roles, CIO type roles, where data infrastructure, data engineering, those kind of things.

It's always been muddy anyway, where the lines of responsibility for those things sit. I think that's only going to continue as things like software engineering are disrupted through the kind of technology we're talking about. So I'm not sure. This is completely honest. The honest answer. All I can say is, I think the people with the kind of skills that we're talking about that a typical studio possesses around understanding of the commercial value of the data, understanding the technology of, of understanding how to drive innovation and monetization through the internal data asset.

I think it can only become more valuable. So I'm bullish on that side of things. I'm just not sure how it's going to pan out in the future. I need to ask you, obviously, as you alluded to earlier, you joined with a handful of. People in the team, you got up to, I think you said around 60 earlier. So what are the fundamentals for someone who's listening?

If they're building a data team and how do you get them up and running? Good question. If I figure it out, so it's probably a little bit of a cliche, but there's three components to this always. So there's the people, there's the tools and there's the process. So on the people side Obviously your talent acquisition your pipeline, how you go about attracting talent into the organization becomes critical importance.

And for me, that's often come from, Being able to articulate a clear vision for data within the organization and how it drives value and why we're investing into these things and making that really exciting and compelling for people. If you can genuinely tell a convincing story around why what you're doing matters, I've always found that to be a huge part of how we're able to attract talent into the business.

Then there is also a couple to that. The learning and development opportunities you're providing, the scope and the breadth and the the freedom, really, I think, to innovate and be creative about how you're using data to drive value in the business. And I've always looked for people that are motivated by the value rather than the technology.

But nevertheless, the coupling of the two becomes really important. I'm personally excited by both of those things. Yeah. Then in terms of the process, I think being able to create an environment in which people have a clear sense of accountability for what they're trying to do and the support and frameworks to do that and the organization, how that slots into what the organization is trying to do, that becomes extremely important as well.

So it's, it's, Incumbent and I think incredibly important to be as a CDO or as a senior data leader with the CDO otherwise to be able to articulate how you're driving value and how you're Working towards the goals of the organization and that comes through some of those process type things in addition to The core process domain of, of data, which is, you know, building out your retail pipelines, your quality management, your governance, and then the value delivery processes that you go through, how you actually do analytics, how you actually do data science, how you release ML, like defining those things and articulating clearly becomes really important.

And and, or at least identifying with your organization, who's going to be accountable for, for doing those things. And then finally is the tooling. And I think. Making sure the team obviously not only has access to the tools they need to do the job, but you're constantly thinking about what else you can do better through the tools you've got and how you solve pain points.

I mean, in our, it's one of the, the interesting things about the development of data over the last 10 years is just the proliferation of data. The modern data stack is often gets referred to and the constant kind of innovation that's going on across, across the ecosystem in terms of how different things fit in.

And invariably all those things are doing is. Automating and reducing pain for your team. And I think if you're able to explain to the people you're trying to develop and bring into the organization, how you're looking at those services and tools and how that's going to help them to deliver and deliver more.

Again, so much better. The amount of times I've been asked, What tech are you using? What tools are you using? It becomes really important. And you never want to be on the back foot and saying, Oh, actually, yeah, we're using stuff that was cool 10 years ago. Were there any growing pains when it came to growing this team?

Was there anything that was a particular challenge? I think so, yeah. There were a fair few things. We do a diverse array of things in the data team in Dojo. So, yeah. It's ranged from, as I said, kind of very low error low error or low, low, excuse me, low fault tolerance processes and products. Things like I was talking like reconciliation and treasury and reporting the amount of money that gets paid to our customers, right?

That is stuff you can't get wrong. That's the stuff that makes me lose my hair. Those are the kind of hyper sensitive things that you really want to put a lot of rigor and robustness around. But at the same time, we also have. We did a hackathon just a couple of weeks ago on how do we apply language models and how do we develop all these AI techniques internally.

Those different things require a vast array of personality types and thinking around what you do in each of those areas and I think, Building up an organization at pace that can serve both of those types of domains and potentially go from supporting a fast paced commercial organization through to a products organization that's trying to make the exact right decisions on, on what we're, on what we're developing for our customers.

And the sunk cost that comes with those things, the significant cost that comes in investing in products. Again, there's a vast arrays of different skill sets and time types that we need. So yeah, I think the growing pains have always been. How do you keep all those things in balance without over indexing one area or another?

And with an eye to the future, but also dealing with the now. That's the biggest challenge, and it's spinning a lot of plates at the same time. So, not a specific thing, but it's just the general challenge that comes with the role. I think this is quite relevant to something you said earlier as well. But how do you get data to have a good partnership within function of the business and win them, I guess, win the hearts and minds that you've alluded to some of it, but also I think it'd be quite relevant because of the structural changes that you've alluded to as well.

Now they're sat underneath certain, you know, possible C sweep functions across the business. How, yeah. What did, what does the team need to do around sort of that? Yeah. By getting that buy in. Well, now I get to say, Hey, this is your problem. Yeah. So I think one of the most important things that you need is in any leadership role regard data.

Otherwise, again, is you need empathy. You need to be if you're in a function that at least in part is responsible for supporting other domains, you have to have empathy with all those people are trying to achieve. And so it starts there. It starts with spending time with your colleagues in those areas, understanding the frustrations and challenges they're dealing with, and finding constructive ways to try and support mutually towards a given goal.

I think the the areas where I've probably had, had the most difficulties is where it's been less clear what the outcomes are that those domains are trying to achieve. And, and sometimes that requires us to take a step back, get a whiteboard, and actually stand back and figure it out. Or it requires us to understand that certain domains move much quicker than others.

And therefore, they can't, if they can't, they can't necessarily say what's going to be of top of most importance to them in, A month's time, then that's fine. You just have to be able to adapt it. And what's, what's worked for me, I think here at dojo is just recognizing that and understanding that we're dealing with different domains.

Like I said, that, that vast range from core infrastructure through to fast moving commercial fast paced commercial teams and identifying personality types that suit those different domains and can respond and, and, and, and compliment those different areas. I think where, if you try and.

Take too rigid an approach and you can't adapt. I think that's where it's been. That's where it's been most tricky and helping helping people I think to come along the journey of Why we have to make certain trade offs while we have to make difficult decisions about what we can prioritize what we can't that's that has always been of high importance for us as well being able to communicate and articulate these these kind of These kind of considerations we have and and bringing people along that journey.

So I think It's, it's, it's been, you know, truth be told, it's been fits and starts. There's areas where we've progressed much better than others, I think, because the, we've just made greater progress and built trust in those different domains. I'm not sure I'm answering your question. I mean, that's, I think that that trust, I think it's something you said earlier is having that story and providing the value to the business.

So they buy into it. I guess you have to win hearts and minds like people, data is still new, right? As a function, like it's still a growing thing and people have to be educated around it. It is, it's true. And one of the things I've I've been extremely grateful for her at dojos. I've never I've never had people resistant to data.

It's always, it's always the opposite problem. People want more data, want more, more support than, than we can always realistically provide. So it's about giving people as much as they can do to serve themselves. Well, it's also then providing the expert support and guidance where, where appropriate. So and then as you say, and you use the word yourself is, is trust.

It's, you start small, you prove value, you deliver. In a smaller area and then that eventually compounds into something much larger and much more encompassing. I think if you try and change the whole organization at the same time, I think that's where it becomes very very tricky to, to deliver.

I think that's sort of what we just said there might lead quite nicely into this next question. It's something we were talking about when we're putting this podcast together is how data is probably one of the fastest evolving areas within, you know, within tech. Firstly, do you think that's a good thing?

With, you know, accessibility of data, making it maybe easier to deploy and what does that mean really for the future? I think, is it a good thing? Let's answer that question first. I think data is still immature. I often look back to how technology has developed, how the internet functions developed.

You know, I remember, I'm just about old enough to remember where you had job titles like webmaster, and it was the guy from IT who was responsible for the whole of the website of a particular company. And and you look at now at companies say, like, who are digital native, like Microsoft, Previous company far fetch, for instance, where you had a huge number of different array of job types that serve what was previously done by a very small number of people.

I think data is still a relatively immature function in that regard. I still think it's hard sometimes for people to understand the difference, say, between a product analyst and a data scientist and the different roles and responsibilities those roles have. I mean, even within data, it's hard to articulate what data scientist is, right?

So we get confused. So I think you know, for our colleagues in, in other areas and outside of the space, I think it is, it's still relatively difficult to understand. I think as a result of that, do I think the fast paced changes is a problem? No, I don't think so. I think it's, as I said, I think it's conduced that that maturity and an understanding and recognition of the value that data is bringing into companies.

And, and as we go back to the conversation we're having around actually the necessity now for for the use of data, not just, it's not a luxury. It's, it's, it's the absolute center of a lot of companies. But I think as an, as a, an industry, I think we could probably do a better job of explaining how we deliver value at all levels of the organization.

And I think and I think also there's still with, in parallel to that is a long way to go around the maturity of the data stack itself. I mean, we were talking a moment ago about just the absolute proliferation of tools. I find it hard to keep up with all the different things that are coming out onto the market, let alone you know, again, my colleagues.

So I mean, it was, I'm sure at one point my CFO thought I was just making up these tools that we were assigned to look at and evaluate. But I think So I think, I think there's still a long way to go there. And I think what we'll probably see going to the question of the future, I think almost certainly there has to be an aggregation of the tool set.

It looks like Google with I think it was a data, data form that they've released recently is a, is a step towards that direction where, where things like say a DBT and an Airflow and some of those kind of ETL tools Fivetran, those kinds of things that we start to see more of a, an aggregation.

And it, I, I would be shocked if we don't see over the next couple of years, Google and Amazon start to make meaningful headway against those things because it's, it's, it's It's just a it's extremely expensive. It's extremely complicated. And and it's extremely brittle as well, right? Any one of those things can fall over and all of a sudden you, you know, stakeholders could don't have the right numbers in front of them.

And again, it's extremely difficult and embarrassing to deal with. So so I'd be surprised if that if those things don't start to come together. So in terms of the future of data, I think there's gonna be an aggregation of the different tool sets. And probably a move towards oh, sorry, away from the the specialisms that we, that we see probably into more value focused and more less technically intensive roles.

Let's say, I think, I think it's, it's fairly apparent that The certainly on the analytics side, for instance, the accessibility of the tool sets is becoming easier and easier for people to deal with and therefore the need for a full stack data scientist to be able to do everything is probably likely to diminish, I suspect, over the next few years.

Yeah, perfect. I'd love to dive into the last two questions that we always ask on all our podcasts is I guess, what's the biggest challenge that you've had in your career to date? I was thinking about this question a little bit. I think the. I've had challenges at every stage, right? And if you think about it broadly in kind of three, three stages, first stage of being an IC, being a data scientist is you know, getting to, getting scripts, the tools, the tech, all the, the jargon, and actually figuring out how do you develop, deliver value as an individual.

Then it jumped into, Being a manager, how do you start to lead and how do you start to influence within your specialism, other people bringing together and building the fundamentals of building a team that we were talking about. And then as a senior leader, it's how do you change an organization?

I mean, almost certainly that last one is the most difficult because you're you're dealing with the realities and the trade offs that the companies have to make. But I think every single one of those, at the time I was there, was feeling like it was an enormous challenge. So anyone who's dealing with any one of those three, I sympathize with.

Yeah, again, if you figure those out, let me know. And finally, we're going to put a bit of a twist on this one, but what piece of advice would you give to someone as a data leader who's thinking about joining a scaling business? Thanks. Understand how your function and data delivers value into the business that you're joining.

I think whether it's a data centric business, if you're building ML systems that, that deliver value as part of the core product, or whether you are a supporting function that's helping others and empowering others to to grow quicker, minimize cost. It's imperative that you are able to understand that and develop a narrative that is attractive to the people that you need to bring into the organization to help you succeed, but also to your colleagues and to the wider organization about the the remit and the role that you play.

Otherwise, it's so easy to get pulled from pillar to post in terms of what you should be focusing on and how you should be going about doing things. And I think being very clear about that, in addition to being adaptable to change. But I think being very clear about what you're there to do is the most important thing.

The truth, the truth around fast growing companies is There's always way more things to do than there are people to do them and therefore data being the adaptable Skill sets that we have we can end up getting pulled into a million different things I think it's been very very clear about what you have to do and remaining fixated on that it's by far the most important thing to do.

Perfect. Well, Mike, that'll wrap it up there. Thanks for being part of this. Have you enjoyed it? Thank you. Yeah, I did. Thank you very much for having me on.