Bringing analysis back to data analytics
Joel Shapiro, JD, PhD, is the Executive Director of the Program on Data Analytics (PDAK) at Kellogg. Since joining the school in spring 2015, he has been engaging with students via the Big Data and Analytics Club and developing a new course for the PDAK curriculum. Shapiro helps businesses understand how to better take advantage of data and to improve decision-making across the enterprise. He has served on faculty at Northwestern for 11 years, and in 2010 built the first online degree program in predictive analytics.
In this Q&A, Shapiro talks about big data, his vision for data analytics at Kellogg, and more.
What do you think is the biggest misconception about data analytics?
There are so many – choosing the biggest is a challenge! One critical mistake is thinking that a company’s data analytics initiative should exist primarily to generate dashboards or similar types of reports. Sure, it’s good to have snapshots of what’s going on at any given point in time, but these tools often say nothing about what actions we should take or decisions we should make.
Plus, any time data are used to generate a dashboard or other summary of data, someone on the back-end of that tool is making decisions about what data are important and how they should be aggregated or parsed. That can’t be a cavalier process – it needs to be highly strategic with buy-in at the highest levels. Very frequently, there’s way too little thought given to how data should be synthesized to provide a useful overview of what’s happening.
Another misconception is that analytics is primarily an IT function. No one can argue that good IT systems are critical to data analytics – they absolutely are. We must have easy and efficient ways to collect, store and access a tremendous amount of diverse data. But analytics doesn’t help us make good decisions unless someone with key business strategic responsibility purposefully uses it to answer a well–defined question.
It seems to me that the hype around analytics often leads – quite unfortunately – to the removal of “analysis” from “analytics.” At the end of the day, you can’t get insight from analytics without knowing the business context and having great analysis and critical thinking skills.
In your bio, it says you have “a strong focus on how to apply analytic solutions to solve real-life problems.” Would you provide one or two examples of a problem most people face that you think could benefit from analytic solutions?
Most people use data analytics to describe what’s happening or predict what is likely to happen, but then don’t know how to move to action. Analytics can help us define what actions we should take and improve our decision-making, but only if we use it carefully and strategically. Too often, people rely on data that describe or predict behavior and then make unfounded guesses and assumptions about what they should do to bring about a desired outcome.
For instance, I used to do some work with a firm that was trying a new strategy to retain customers. They had learned that the percentage of customers that reached out to their customer support team was much higher for repeat customers than for non-repeat customers. The firm did a great job of using data to describe and even to predict what would happen – if someone engaged with customer service, they were much more likely to turn into a repeat buyer. But the firm used this evidence to justify a new strategy of having customer support reach out to all customers, hoping to turn even more of them into repeat buyers. That action is not at all supported by the data.
What is Kellogg’s approach to teaching data analytics?
We believe that data analytics is primarily a management and leadership problem, not an IT or data science problem. That is, business leaders need a strong working knowledge of data science to derive the benefits of data analytics, which fundamentally change the way we make decisions. Analytics isn’t useful if it doesn’t help solve real and important business problems. Thus, we teach a strong foundation of analytics methods, while being relentlessly problem–driven.
We want our students to not only understand the methods, but to practice using them in diverse contexts. After graduation, we want them to have the facility to see any problem in any context and understand how data analytics can be used effectively.
How would you describe the opportunities available for students?
Within Kellogg, students will have the chance to learn critical data analytics tools and to practice, practice, practice. No two real-world problems look exactly alike, and we want our students and alumni to be able to recognize and solve problems across widely-varying business contexts. We want our students to engage deeply with their professors, who will help them recognize patterns and problem-solving strategies.
The professional opportunities for students are immense and tremendously exciting. Our goal is to train the best bosses that the best data scientists have ever had. It is hard to overstate the value of someone who can merge business strategy with a rigorous data analytics decision-making mindset and capability. Those people are in short supply, and they will succeed. Quickly.
What is your vision for what students will learn about data analytics moving forward?
There’s so much hype about big data and analytics right now, and it’s important to keep a sense of perspective. Big data can not and will not be a panacea to all business problems. Unfortunately, it’s so compelling for firms or departments or employees to say that they are data-driven, that we’re finding a lot of imposters. Those will shake out over time.
That’s why students and professionals need to learn the fundamentals of data science. Becoming a truly rigorous, data-driven decision maker will not go out of fashion. Those who possess a strong working knowledge of data science and use it strategically will make better decisions, and ultimately see better outcomes.