Insights

The data series: Interview with the Director of Data Science at Kantar

Article written by Jacob Knight, Head of Data at X4 Technology.

As part of The Data Series, I had a great discussion with Tom Evans, Director of Data Science at Kantar to hear about his approach to hiring diversely, how we can encourage more people into data science positions by redefining the discipline and the advice he has for anyone looking to move their data career forward.

As data science is still a relatively new discipline, it was great hearing first-hand Tom’s experience of transitioning from working as a solo data scientist to managing a team. This fantastic interview is one not to be missed, full of valuable insights for both budding data scientists and businesses alike.

What traits do you look for when hiring a data scientist and what abilities make a data scientist successful?

No one thing in particular draws me in to one data scientist over another. A big problem with the data science industry is that it’s so ill-defined, people look for someone who can do everything, which is always a very difficult thing to do and if you do hire those people then they don’t stick around long because someone comes and offers them more money elsewhere.

I tend to look at my team as a whole and try to understand what’s missing. Does this person think in the same way that we currently think? If the answer is yes, then typically I’ll be thinking we need someone who thinks differently because the purpose of my team is to be flexible to clients’ needs and problems. If we’re in those situations and we all think the same way, then we’ll all come up with the same solution or no solutions at all. That’s how I tend to think about my hiring strategy because if I hire all the same people then I will very quickly find myself not being able to be agile and nimble with my team.

Some of my data scientists will challenge me because I’m not considering their way of thinking so it’s useful to have people who don’t look at things in the same way as you. And, the added benefit to that is that you hire very diversely because you’re not looking for a candidate based solely on their qualifications. You’re looking for someone with a very different background and then you get very different types of candidates across ethnicities, genders, etc. and it breeds a really positive team environment.

All of that typically comes from having a good understanding of what my team can set out to achieve, which is relatively difficult in data science because quite often, data science teams in businesses have just appeared. A lot of companies have historically set up data science teams without a full understanding of what they want them to achieve. So, for me the critical thing to do is to identify areas of weaknesses and hire for those weaknesses. For instance, it might be that you’re too top heavy and need someone to come in and do more of the groundwork, or it could be that you’re missing a particular skillset.

What do you think needs to change to help increase the number of data scientists moving into industry?

People need to think about data science in a very different way than how they do currently. I feel the issue that fundamentally exists around bringing more people in to this as an industry, is that it’s so immature because it hasn’t really been a job for a very long time, so the understandings of what that job involves and its purpose are very misrepresented by the people who are hiring and also by the people who are coming out of university who are looking at the job on the market.

I’m finding that a lot of data scientists I interview talk about wanting to build and tune models, and I think that attitude comes from the fact that that’s what they see online when they’re watching videos about data science. And, to be honest with you, once they get in, they’ll realise that that’s actually a really boring job and companies don’t necessarily want that.

A way to encourage more people in to data science is to redefine the discipline and that’s very hard to do, but I think there needs to be as I don’t think this catch-all term of data scientist can stick around for much longer. We’re already seeing a lot of newer job titles such as machine learning engineers and data engineers. You’ve also got to reclassify what a data scientist actually is to help people from different disciplines apply for these kinds of roles.

What is the biggest impact Covid-19 has had on your line of work?

In the industry of market research, it’s been going through a period of transition for a long time. That transition is more towards newer types of data and techniques, thinking about different ways of supporting clients and Covid-19 has accelerated that.

In terms of how much work we must do, it’s still lots but in terms of the type of work we do it hasn’t really changed. The way we work has fundamentally shifted – it’s obvious we’re all working from home and we’re having to be a lot more joined up. I have a daily stand up with my team, which I’ve found force those more personal moments by making space for people to say stupid stuff, talk about their weekends and discuss the fact that they’ve got ridiculous hair – all of those kinds of things are really important to make sure that you don’t get as disconnected.

I think in general Covid-19 hasn’t impacted data science as a function because so much of our work is remote and relies on technology that’s in the cloud, etc. so Covid-19 hasn’t had a huge impact on our industry specifically, outside of the fact that businesses are starting to struggle more. I think there’s just a repositioning of resource into differently impacted industries because of Covid-19.

Do you think there’s any positives that can be taken from Covid-19?

The answer is obviously yes, but I think more deeply than that there’s been much more of a focus around health & wellbeing in the workplace and flexibility, not necessarily just working from home but for childcare arrangements too. I think there’s been a transition to a much more two-way understanding environment where it’s much more personable and tailored to your particular needs.

Businesses that blanket wide say “This is now what’s happening” are the ones that find themselves struggling because the dynamic of the workplace is a lot more personalised. Personalisation is a massive thing generally, but I see personalisation in the workplace more. I have different conversations about how the needs of one person in my team differ to the needs of another, and how I want to be able to support them to ensure they’re continuing to deliver successfully. That’s where I think the shift is, less so than ok well everyone can work from home now.

Some of your recent achievements include winning the Computational Brain & Behavior Outstanding Paper Award and you’re also responsible for scaling STAN, a product developed by your team at Kantar. What’s been the most memorable moment in your data science career to date?

It’s hard to pin down, but something I keep going back to, a sort of transitional moment in my career is when I left dunnhumby working as a data scientist and took over the team at Kantar. It’s a shift that was something I wanted but not really one I understood entirely or really anticipated all the pitfalls and challenges that come with being solely responsible for the success or failure of a particular area within a business.

For me, that transition from data scientist to running a team was cathartic. In the data science space, what it means to lead a data science team is a little bit ambiguous and a little bit neglected. Data science is very much about model this, etc. and I think that actually because it’s such a new discipline, jumping in at the deep end and learning how to make more people successful than just yourself, is one I’ve found very interesting.

What advice would you give to anyone considering their next step in data science?

What I might advise other people if they’re looking to move their career forward is stepping outside your comfort zone is really important. Be comfortable with making mistakes and knowing that things will go wrong, and you won’t get every decision right. Sometimes actually just making a decision and going with it is really what’s needed in that moment, irrelevant of whether it ends up being a good or bad decision, that’s something I would encourage anyone to do. And, to make that step and not get bogged down in where you feel most comfortable in data science. It’s about really stepping into an area you don’t understand as it helps you understand things about yourself.

What excites you most about recent developments and the future of data science and AI field as a whole?

I really want to see how data science as a job function matures. No one’s really worked it out or has the answer to what data science is and does. At the same time, you’ve got new tech and new models, there’s loads of developments but I still think people haven’t worked out how to get hold of what already exists and start using that in a way that makes sense for one particular industry vs another.

So for me, what’s most exciting is seeing how that’s going to evolve as companies become much savvier with their requirements of data science teams. I think what will happen is it’ll really stretch the minds of a lot of people in data science about their own position and career, and about how they want to shape the company they work for or the industry they’re in. That’s where were going to really make gains in terms of tech advancements, its incremental gains which come from those ‘aha! moments’. That for me is what’s really exciting.

Stay connected: