The Joy of Mixing Data & Design
An interview with Jean-Baptiste Bouzige, founder and CEO of Ekimetrics
Data driven. Evidence based. Numbers focused. There are many corporate catchphrases to tout one’s analytical aplomb. On the civic side, we’ve seen how governments also aspire to incorporate data science and analysis to improve efficiencies and innovate. But we’ve also seen how a reliance on data and algorithms alone have wreaked havoc by wrongly disqualifying people from social services or unfairly targeting communities for policing. To dig into how data can be responsibly integrated in both the public and private sectors, we sat down to chat with Jean-Baptiste Bouzige, who leads the NYC expansion of Ekimetrics — a global data science consultancy that solves problems using a combination of data science and design thinking. Here, Jean-Baptiste shares his views on the founding ethos of his practice and the importance of balancing quantitative analysis with qualitative thinking.
Q: Tell us about how you got started — what did you set out to build?
JB: The way we think about our practice is data science with a human side. We created the company 13 years ago with the idea of solving strategic questions with data, but never data alone. We were five founders and it’s what we like to do — we’re not pure geeks! I don’t like to run complex models just for the sake of running them. What I like to do is to solve problems! And this way of working is now one of the most important convictions at Ekimetrics, and also one of our main differentiators. Now we’re around 250 consultants and everyone does both sides of the work: data science, as well as data storytelling and design thinking.
Q: Is it possible to find people who excel at both kinds of work?
JB: It’s a matter of recruitment and training. We invest a lot in training. A newcomer is doing something like 45 trainings in the first year and we do nearly everything internally. Our competitors use two or three different teams to deliver the same thing that one team at Ekimetrics can do. Our work is actually very straightforward. We start with a business question and then invent the methodology to answer this question. Most of our competitors look to productize what they do and create verticals for their work. But this can lead to bigger problems by creating silos around data, rather than using data to break silos. For CEOs or mayors, sometimes you ask a question and parts of the answer lie in different approaches or departments that are not fully comparable. Our job today is to create integrated approaches — to build bridges across business functions. That’s why everyone at Ekimetrics can act as a translator from the business problem to the data science methodology, and then, to the contextualized use case.
Q: On the human side, how do you act as a curator or educator to let people know the value of what data can deliver?
JB: We are a business advisor from the onset. As data scientists with a business focus, we have a point of view on what to do to give meaning to the data. We have seen so many projects that have failed to deliver when the team treats the challenge as purely technical and don’t care about the meaning of the data. The best teams I see on the client side are the teams where you have a mix of disciplines in the same team: it could be marketers, developers, IT, and data scientists. When the team is hybrid, it’s a key enabler to deliver something meaningful with data. It’s not a question of spending power to buy the best tool on the market. I often tell my clients, your first step when you want to create a data team is not to hire 40 data scientists. The best move is to hire someone who can pilot the data science role. It can be one person able to master this translation capability of applying data to problem solving.
Q: In this way, I think cities are setting a good example for businesses, as we often see local governments starting out this way: creating one role for a data and innovation expert and embedding that person in different operations to run experiments and see how it goes before scaling.
JB: Yes for sure, now I have more and more clients saying, I have all this data, but I don’t know what to do with it. I tell them it’s really important to start with the right sequence, you need to start with a purpose. My first advice is to link your data strategy to your brand or communications strategy, and this way, you’re able to build an editorial line. The second step is to see how you create a data roadmap that is in service of this editorial line. Once again, what is not important is the tech stack — that’s the last step, but that’s where many clients are starting. They say, I’m going to buy this tool. But since things are moving really, really fast, the most important thing you need is purpose and architecture. In this way, it’s not about creating a robust data lab or selecting a set of tools, it’s about creating a shared process. This is not an easy path. Clients may feel more protected if they pay a big software or consulting company to do the big changes for them. But little by little, we are showing the value of our incremental approach.
When you’re not addressing your data holistically, I call it the IKEA syndrome. For example, you have a big mess in your garage and go to IKEA and think, if I purchase this shelving system then I will be organized and solve my problem. A month later, your garage is messy and you realize that the problem was not your lack of furniture: the problem was you! We see this happen with data management: the sales promise of some products can be so beautiful. Companies can be excited to go for IBM Watson or Salesforce, but if you don’t do the prerequisite strategic work, you’re gonna spend a lot of money and make a lot of mistakes. And you’ll end up throwing a lot of things in the garbage.
Q: What is the most fun project you’ve worked on?
JB: One project I really like is for a global travel company. Our client is a Chief Marketing Officer who is also a Chief Data Officer and leads a team of data scientists, IT specialists, and marketers. She’s really pragmatic. When we started working together, she told me: “I don’t want to be an early adopter. I want to be a smart adopter or early follower, because my job is not to reinvent the data world.” When her executive committee asked for an AI roadmap, her first thought was “we’re going to be a data-driven company.” But she quickly realized that she wants to be a customer-driven company — that data is just an enabler. We built a story together on what is unique about her company and how we can use data in the service of these special qualities. She doesn’t want to build something new for the sake of novelty. She wants to be the best-in-class in what she does. It’s really inspiring to me and in particular, to my work in R&D using open-source software, reminding me I don’t have to create new algorithms every day. I have to be best at applying what exists in the community — that’s why it’s my favorite project.
Overall, it’s really inspiring to me and my work in R&D to use open-source knowledge as a reminder that I don’t have to create new algorithms every day. I aim to be the best at applying what already exists in the community and create the best translations of algorithms to the business world.