BOSTON – Analytics team leaders from a variety of industries gathered at the INFORMS Conference on Business Analytics and Operations Research this week for a panel discussion on how to build a well-rounded data-analytics team and manage data professionals: where to position them; how to keep them challenged, engaged and motivated; and what it takes to lead them.
The panel was moderated by Julia Kirby, editor at large at the Harvard Business Review, and included team leaders from major corporations, including Brian Eck, quantitative analyst at Google; Kerem Tomak, vice president of marketing analytics at Macy’s; Erica Klampfl, global future mobility manager at the Ford Motor Company; Dayana Cope, manager of operations research engineering at Disney; and Jeanne Harris, managing director of IT research at Accenture.
Kirby led off the discussion by asking what she referred to as “the perennial question”: Where do data scientists sit in an organization? Do you keep them together in their own group so they can keep their skills sharp with constant interaction with each other, or do you distribute them throughout the company’s business units because that is where they must have an impact?
“I think where the analytics professionals sit in an organization depends on the size of the organization,” said Klampfl. “It’s not one extreme or the other. I think it needs to be a combination of that. I do think that analytics professionals working together provides them the ability to learn from each other and have their own community, but it’s nice to have rotational opportunities to bring in lessons learned and work closer to the business. So I think the sweet spot is really a combination of those.”
“I think there’s advantages and disadvantages [to keeping them in a group],” said Cope. “The advantages are that you have the opportunity to share learnings, a centralized approach and you have more resources the ability to hire specialized knowledge. The disadvantages of a centralized approach come with when you put the word ‘business’ in front of analytics because that requires a concerted effort to connect it to the business and that is harder to do when you are centralized. At Disney, we have more of a hybrid approach. We have centralized analytics functions that have smaller subgroups that are connected to the business, that have business alignment. We find that is very useful and works well for us. One thing to keep in mind is that, when you have a centralized approach, relationships become very important, and at Disney we are very proud of our analytics culture because we spend so much time and effort developing those relationships and maintaining those relationships.”
“About four years ago we did a survey of how companies are organizing their analytics talent,” said Harris. “And back then there was more of this debate of, ‘Should we centralize them or should we decentrailize?’ And what we found was the right organizational model is largely dictated by the level of maturity of the organization itself. And if you think about it, that makes a lot of sense. An organization that’s just getting started needs to keep those analysts very close to the decision makers. But today I think the thing that’s really changed, and I think we are hearing it reflected by the companies on this panel, is that there’s much more of a movement toward going back to first principles, trying to define, ‘What’s the right operating model for us?’ with regard to analytics and defining certain principles which then dictate whether an individual or a group is going to remain local in the business unit or in a business function, or get centralized.”
Kirby raised the question of how companies keep data professionals, with their in-demand, specialized skills, happy and engaged in their work.
“I think it has a lot to do with what you are working on and whom you are working with,” said Eck. “The ability to network with peers with a similar level of expertise that may be in different areas is very enriching. And also the amount of autonomy you have in your work plays a big role.”
“To keep these types of people happy,” said Cope, “I think we have to, first and foremost, acknowledge that they are different in the organization. I think the analytic professional should be led by analytic professionals because they understand those differences and those analytics leaders should communicate those differences throughout the organization constantly. They should acknowledge different processes for them, they should have different technical ladders, different job expectations. They should also acknowledge that they have different motivations. I think you keep them challenged, that’s what motivates them. And if you can’t keep them challenged all the time, you can provide that framework to share learnings. The last thing I want to point out is that it’s important to communicate and always celebrate wins.”
“[I am asked] what’s keeping them with you, especially in Silicon Valley, where they can walk across the street and go somewhere else,” said Tomak. “And I was thinking about it and I came up with several ideas. One of them is I basically get out of their way. I empower them, I give them the tools. I give them the exposure because we are deeply embedded in the C-suite. The senior executives at Macy’s want to make decisions based on data and they trust us tremendously because we have proven ourselves to them. So I essentially get them in front of the C-suite and give them the platform to talk to them directly as to what they are doing. I think that’s what’s keeping them there, because they see they are driving the bus.”
Kirby then asked the panel what it takes to be a great leader of this kind of function.
“I think it’s important to get people with complementary skill sets,” said Klampfl. “You don’t want to have all people who do stats or all people who do optimizations, so you really have to build a team recognizing individual strengths. Analytics is a broad space and there are a lot of specialties, so it’s really understanding that difference and I think that’s why if you are an analytics professional it makes it easier to be an analytics leader because you understand the space, you understand specialties. … So to be a good leader of a team you really have to focus on the individuals on your team and help them succeed.”
“How well the leader can identify opportunities for the team and position the team to solve those opportunities is a key issue,” said Eck. “Reaching out to the organization, the person needs good leadership skills, not just strong technical skills. But they do need the technical literacy to be able to position the problems correctly.”
“You cannot be too disconnected from the people you are managing. You have to keep yourself up to speed as well,” said Tomak. “You have to understand what they are talking to you about and you have to talk back to them in a language they understand.”
“Someone who leads an analytics group, whether it’s a single analytics group in a single department or a Chief Analytics Officer for an entire company, they really need to be a tireless advocate for analytics and an agent of change,” said Harris. “The truth of the matter is, being able to communicate, being able to persuade people, helping them understand the potential of analytics to transform their business and being able to work with them to take those ideas, put them into practice and actually get the results at the other end, that’s maybe the most important thing a leader can do.”