How to make Big Data a business-world blockbuster
In the recent blockbuster Gravity, astronaut Matt Kowalski (George Clooney) often intones, “Houston, I have a bad feeling about this mission.” He might very well have been talking about reactions to McKinsey Global Institute’s (MGI) article on Big Data.
The article’s essential claim is that there is a coming dearth of “big data” analysts—deep analytical positions—within organizations. Researchers forecast a shortage of 140,000–190,000 analysts by the year 2018 in the US alone. Even more alarming is the need for an additional 1.5 million managers in the US “who can ask the right questions and consume the analysis of big data effectively.” The issue facing both universities and corporate America is whether this gap can be filled—and if so, how. This was the topic of a recent panel discussion at the Mu-Sigma Decision Sciences Summit held in Chicago October 24–25.
Appropriately titled “Re-imagining education for the learning worker,” the session cast the issue more broadly as an imperative to create a culture of learning (rather than knowing) among employees—with both academic institutions as well as corporate America contributing to this endeavor. From the perspective of a business school, one of our primary responsibilities is to create good consumers of big data analyses.
What does this entail? First, it is important to learn about the scope of big data and the types of questions that it can help managers answer. Research centers within business schools (for example, the Data Center at the Kilts Center for Marketing) that house, facilitate and disseminate research using large data sets (in the case of the Kilts Center the data relate to household purchase behavior across a wide variety of product categories over time; how products in these categories perform at the store level over time along with factors that drive this performance; and the advertising expenditures of the products over time) can play an important role by interfacing with MBA students and communicating Center activities. A key aspect of such learning is changing the one-off market research mindset to bring such information to managers’ desks on a continuous, real-time basis.
Second, it is important to learn something about the process by which insights are generated—not so much the specific algorithms one would use but rather the range of approaches to obtaining insights. Insights can be obtained by looking at raw patterns in the data. For example, Google Flu Trends looks at search activity to say something about the prevalence and severity of a flu season. Given the sheer volume of data being collected, the rate at which the data are being generated, and the wide variety in data structure (text, numbers, pictures, videos, etc.), understanding data patterns is an important first step to getting insights from big data.
Insights often also require experimentation on the part of the manager. If I always spend $5 million on advertising each year, it is not possible to understand what might happen if I increase or decrease that number. Understanding the importance of experimentation is necessary for managers to make more informed decisions in the future.
A third way of generating insights is by applying econometric methods to previously collected data—predictive or response modeling. Again, managers do not need to be methodologically skilled; but they need to understand the assumptions being made when such models are being taken to the data as well as how to reasonably interpret the outcomes from such analyses.
How can business schools help in this learning process? Curriculum development and enhancement is certainly one way to accomplish this. Chicago Booth offers classes in data driven marketing (Günter Hitsch), data mining (Matt Taddy), Statistical Insight into Marketing, Consulting, and Entrepreneurship (Zvi Gilula) among others that advance knowledge along all three means of generating insights.
An important aspect of these classes is their interdisciplinary nature—drawing insights from marketing, statistics, and in some cases, computer science in order to provide a more comprehensive picture of what can and cannot be done with big data. (Other schools and universities have started offering dedicated Master’s level degrees on big data and analytics. INFORMS (the operations research and management sciences society) offers a certification program for analytics professionals.)
Third, educational institutions and universities can work with corporations to provide a more complete picture of big data and analytics. Companies such as Google and Microsoft have research labs that produce a large volume of research that can enhance managers’ skills with respect to being consumers of big data. Additionally, companies like Mu-Sigma have internal “universities” that not only collate and communicate knowledge and insights but also have formal programs for employees and clients to help them become better “learning” workers—i.e., not just focused on acquiring a knowledge base but keeping it constantly updated and refreshed in order to meet new challenges and opportunities that companies may face. Formal and informal relationships across these sets of organizations and institutions can help MBAs appreciate the possibilities with big data and also become better consumers of the resulting analyses.
To paraphrase the words of the famous Autonomous Robotic Organism (Optimus Prime in the movie Transformers), “There is more than meets the eye to how educational institutions are responding to the big data challenge.” With such efforts hopefully the picture in 2018 will be quite different from that predicted by the MGI.