Interview with Win Fuller, Ironside Senior Data Scientist

Predictive horizon concept

How Data Scientists Demystify Complex Questions

To help you understand Ironside’s philosophy on advanced analytics, data science, and predictive/prescriptive technologies, we thought it would be helpful to share insights directly from our data scientists.

Win FullerThis week I talked with Win Fuller, who has over three decades of experience with advanced analytics and business intelligence at a wide array of companies including Staples, VistaPrint, Upromise, Stax, and Bain & Company. Win holds a PhD in Econometrics from Tufts University, and specializes in predictive models addressing questions around churn, customer retention, and demand generation among many other analysis topics. He is also an expert in all aspects of data extraction, integration, and manipulation. His goal in any engagement is to use his experience to clearly understand and realize his clients’ business and analysis goals.

Keen Hahn: Great to have the chance to talk with you, Win. To kick things off, since you’re a long-time veteran in the advanced analytics and data science space, how has the discipline changed and evolved since you started with it?

Win Fuller: The major change that I have observed is the speed with which these analyses can be performed. Computer hardware can now store huge volumes of data and process them in the blink of an eye. Gone are the days when one had to endure long waits in order to get initial answers that relied on only kilobytes of data. The advent of big data and the practices around it has really been a game changer.

In addition, the software currently available makes it fairly easy to ask the relevant questions and to present the responses in visualizations that are simple to understand. In technical terms, this means that hypotheses can be easily tested and the answers can be synthesized quickly into organizational operating procedures.

KH: I can see how the unprecedented processing power everyone now has access to would make a huge difference, especially for data intensive activities like predictive modeling. So looking at the other side of the coin, are there any aspects of advanced analytics that have remained the same? How would someone new to the space get an understanding of these foundational points, if they exist?

WF: The fundamental questions that need to be answered remain the same: Who are my customers? What do they think of my products? How are my products viewed relative to those of my competitors? How can I convince my customers not to defect? How can I gain new customers? How can I convince my customers to buy more of my product? These and the many other questions like them that organizations are interested in are all just different ways of asking the same thing: How can I grow my business? For this reason, it would behoove someone new to the advanced analytics space to expend significant effort in understanding how businesses operated and what their most pressing needs are likely to be in order to be well-prepared to answer any manner of analysis question they may face.

KH: Seems like there are a lot of possible variations to consider, and as you well know advanced analytics can already seem daunting at first glance. Using the many client interactions you have under your belt as reference, what do you think are the best practical methods of demystifying these concepts for leadership and business stakeholders?

WF: My recommendation in dealing with clients and prospects is to focus on things with which they are already familiar. What are their most pressing needs? What questions would they most like to have answered? Only then should you move into the areas of data availability and specification of the most appropriate analytic techniques. You can also help to demystify the process by showing samples of the available visualizations that can be used to provide the final answers; however it is critical that the answers you provide to the client are actionable. Elegant analyses that do not provide insights into how the client can modify their behavior in ways that will advance their business goals are generally viewed as less than useful and thus a waste of scarce resources. Hence, providing some sort of ROI to the client is usually quite telling.

A final piece of advice is that it usually causes a client’s eyes to glaze over if you provide any more than a one-sentence explanation of such technical topics as which statistical methodology is the right one to use to answer their questions. Do not use terms that require additional explanations. Make the right choices regarding the appropriate tools to use and be prepared to defend them, but do not make such information the focus of your conversations with the client.

KH: That makes sense. People tend to relate better to contextual examples than they do to the pure theory, especially when it’s as potentially complex as it can be in this field. Since we’re on the topic of clients, are there any points in your career you can think of where you most clearly saw the difference that an advanced analytics implementation makes for a business?

WF:  The movement toward the adoption of advanced analytics techniques in the business world has been evolutionary. That being said, I was lucky enough to spend more than a decade operating in a company where the CEO was very data-driven at a time when this was not generally the case (the 1990s). As a result, this retail operation grew phenomenally from fewer than 75 brick-and-mortar outlets to having a worldwide presence with more than 2000 stores plus significant call center and internet operations.

Without the analytics we provided to senior management, primarily in the areas of marketing and operations (store location modeling), it was recognized that such growth would not have been possible without a strong advanced analytics team and strategy. The C-suite listened carefully to what we recommended and generally acted upon our recommendations. As a result of this experience, I became an even stronger advocate of the power of advanced analytics and was able to transition my academic training into the real world.

KH: That sounds like an ideal combination of executive leadership and technical expertise. What’s the biggest challenge you see being a roadblock to advanced analytics adoption on the scale you just mentioned?

WF: A potential roadblock to the more universal adoption of advanced analytics deals paradoxically with the ease with which this work can be done. I have seen cases where very different answers have been provided to the same questions using identical sets of data. This generally results from a lack of understanding about what tools are most appropriate to answer which types of questions. Hence, it is important that most advanced analytics work, when performed by relatively inexperienced analysts, be subject to some sort of peer review prior to publication. Otherwise the credibility of this type of work can be called into question, adversely affecting the ability to get senior decision makers to pay attention to what is said.

KH: Yes, I can definitely see it being very important to make sure people have the right resources and training to draw on before they publish something authoritative. To finish things up, let’s talk more about the future of advanced analytics. What’s your prediction about where this discipline will go in the future, and what part do you think you’ll play in its development?

WF: Advanced analytics and data science are here to stay. A very large number of companies are now claiming to be “data-driven.” In addition, more and more individuals in senior management positions have been trained in this arena and appreciate the power of gaining an increased understanding of their customers, competitors, and products through a sophisticated analysis of increasingly available data sources.

For my part, I continue to enjoy the process of delving into the fundamental needs of clients and devising creative ways to provide answers. I plan to continue to operate as a trusted advisor who can provide meaningful answers to any organization in need of them.

If you’d like to hear more from the Ironside Data Science and Analytics team, check out our previous interviews with Chi Shu and Pam Askar.