When Kevin Lyons says he tries to learn continually on the job, he means it. The first person in his family to earn a college degree, the senior VP of analytics at digital data firm Exelate took a decidedly roundabout way to get into ad tech. Indeed, with educational and professional experience as diverse as medieval history, insurance sales and data science, it's no wonder Mr. Lyons is contacted by headhunters just about every day.
Evaluating why history is told the way it is – historiography – became integral to the way he studied early medieval monasticism at Ohio State University and later Rutgers. Studying the obscure subject matter "resonated with me and helped me understand things that we as westerners all believe and don't even think about," he said.
Mr. Lyons eventually parlayed an insurance sales gig into work for direct-marketing agencies where he learned data modeling and web analytics. He aims to ensure that the colleagues he supplies with data -- often including Exelate's head of product -- get what they need and know why they need it.
As an insurance salesman and agency account exec, he explained, "I was on the receiving end long enough so I know what it can feel like to be given a recommendation that you can't grasp and almost be told, 'Trust me.'"
Among data scientists, he said, "There's a lot of 'trust me.'"
Exelate brought Mr. Lyons on board in April 2011 to introduce analytics into the mix. "Analytics is core to the business model. We're moving from analytics being this incremental 'nice to have' add-on to something that is fundamental to the way companies do business," he said.
Recently, Mr. Lyons led the company's initiative to refine analytics infrastructure and reporting systems for age and gender data to improve accuracy of the information flowing through its system for targeting ad messages.
Ad Age: Why do you think analytics has become so commonplace for marketers? Are there things you wish more marketers would consider when they come to you for numbers?
Mr. Lyons: To the extent that analytics has been adopted, it's been adopted because it delivers results for marketers, often in a quantifiable way. Interestingly, there's a flip side to the adoption of analytics. Data science is now believed to have near-miraculous powers in certain quarters, including the ability to solve the most intractable business problems 'by sometime next week.' Some problems just can't be solved, or at least not with the currently available data and analytical techniques.
There are four main drivers of analytical success: business acumen, data, algorithms and operations. Unfortunately, data and algorithms all too often get most of the glory. More often than not, projects fail because a clear, achievable and practical business goal was never fully defined or because no one considered how best to usefully implement the findings. So, you should leave the data and algorithms to the data scientists, but you should thoroughly think through what you want to accomplish and how you'll put the numbers into action.
Ad Age: How did the study of history and historiography help inform your approach to data analysis?
Mr. Lyons: There are so many facts available to historians that it's impossible to incorporate them all. Rather, the available facts must be sifted through and a historian has to choose which are 'important' and such decisions shape how their story is told. It's like two individuals standing on two different sides of the same mountain; both can objectively explain the mountain as they see it. Yet, their descriptions might be wildly different. As we incorporate ever more data, with its immense amount of insights into consumer behavior, we need to keep this is in mind … while you attempt to look at the data objectively, you should recognize that there will be many equally valid interpretations, grounding ourselves not so much in theory but striving instead for practical importance.
Ad Age: You say you are contacted by headhunters representing other companies, sometimes multiple times each day. What's the most bizarre example of an inquiry from a headhunter that you've gotten?
Mr. Lyons: There is often a large disconnect between what a company needs and what a recruiter thinks it needs. This problem is compounded by the fact that there are a number of different types of analysts, each with their own set of goals and tools which they use to reach those goals -- and pay scales. And, especially with data scientists, many of the skills (such as Java, JavaScript and Hadoop) overlap with IT positions. So, while not exactly bizarre but at least typical I was offered a job for a junior-ish-level, data-management position, to support a transactional system in the middle of nowhere for less money than we'd pay an entry-level employee.
Perhaps a bit more odd, my past does rise up every once and again and I get contacted with opportunities for insurance sales agent positions.
