Knowing what customers want, and why, and when, used to be largely a matter of intuition. Experts could provide educated guesses, and focus groups could give some guidance. But consumer behavior was often maddeningly difficult to predict.
That's changed dramatically as technology has become more sophisticated. Companies are increasingly drawing on vast troves of customer data, which they use to create predictive models. The result is the emerging field of predictive analytics, which is dramatically changing the marketing industry.
If you're interested in the marketing field, or if you're already working in it but you haven't had much experience with predictive analytics, it might be wise to add it to your professional toolbox. Most major retailers rely on predictive analytics to some extent, and others are using it to highly profitable ends.
Companies that have had widely noted successes using predictive analytics include Best Buy, Olive Garden restaurants, Target and the United Kingdom's Royal Shakespeare Co., which boosted regular attendees by more than 70 percent and theater members by 40 percent, according to a report in Forbes.
In the last few years, predictive analytics has become a management specialty of its own, with teams of analysts, mathematicians, marketers and others working together to anticipate what will motivate customers to buy. Unlike the related science of forecasting, which predicts aggregate information like total sales or number of web visits, predictive analytics focuses on predicting the behavior of individual customers, according to information provided by Predictive Analytics World, a traveling conference.
So how does predictive analytics work, exactly? Eric Siegel, a former Columbia University computer science professor and currently the president of San Francisco-based Prediction Impact, provided a useful summary a few years ago in DM Review magazine (now called Information Management). Say a company is trying to reach customers who bought recently. Each customer would be assigned a value, called a predictor, based on how recently he or she made a purchase.
Seems simple enough: just contact the customers with the highest predictor value. But the math gets much more complex when you combine several predictors into a predictive model. The model will be most successful if it reflects the most important aspects of a customer's purchasing decisions.
This work is made by possible by companies' ability to track customer purchases over time, creating a detailed record of their spending habits. Some companies marry that data with additional information about their customers, such as their marital status, their Internet search habits and whether they own or rent their homes, for example.
The challenge for companies is to use this information effectively without seeming intrusive or creepy (a challenge highlighted in a lengthy Feb. 16 New York Times Magazine piece, "How Companies Learn Your Secrets." The company in question, Target, created a predictive model that figured out if women were pregnant, without the women divulging the information).
Along these lines, Forbes blogger Jerry Michalski worries about the implications of what he calls the "stalker economy," in which companies can collect massive amounts of personal data and use it for manipulative purposes. However, he also points out that large-scale data sharing has tremendous benefits and is empowering individuals, not just companies, through movements like participatory medicine and open government.
Permission must be obtained from CareerBuilder.com to reprint any of its articles. Please send a request to firstname.lastname@example.org.