Follow Datanami:
July 23, 2012

Are Predictive Analytics Overrated?

Ian Armas Foster

In many enterprise data analytics circles a key topic of discussion topic has centered on the value of getting predictive analytics right.

However, in a recent talk, Gary Jackson, who directs the business analytics group at CSC, said that the important elements of predictive analytics are overlooked and the emphasis is being placed on the wrong concepts..

“I’ve been in predictive analytics for 14 years,” says Jackson, “and we’ve been able to predict the future for ten of those years and we’ve done nothing. We’ve predicted, but we haven’t really made any business value.”

For Jackson, the emphasis that has recently been placed on storing and processing big data has been misplaced. He implied that predictive analytics could turn into a self-fulfilling prophecy of sorts, noting a friend who went into a doctor’s cancer-free until it was discovered that she had an 80% of developing cancer. Despite ensuing preventative measures, his friend developed and still has cancer three years later.

Instead, Jackson prefers the predictive analytics that let customers know where they are in relation to the populace. “It’s very important,” he says, “for customers to know where they are…We do all this demographic segmentation and we keep it in house, in our marketing departments. We need to let our customers where they are. People tend to buy a product when they know they’re just like someone else.” He likens this approach to that of his father, a successful insurance agent who connects with his customers. Jackson appears to be wary of pushing humans out of the process, a development that would not sit well with customers.

Jackson brought up an interesting note on categorizing unstructured social media such as Facebook and Twitter status updates. “It’s not about social media,” he says, “it’s about you as a human being. When you update Facebook or Twitter, you’re doing two things: you’re putting up life events that are either lifestage or lifestyle events.”

 Examples of lifestage events include getting married, having a child, going to college whereas lifestyle events include skiing, going to a concert, watching a television show according to Jackson. Getting a grasp on these events and how they shape a person may help a company make decisions on how to market to individuals and put together packages that cater to certain lifestyles and lifestages.

Jackson also suggested an approach to billing clients that is similar to a lawyer’s promise that they will only charge you if they win the case. “You will not pay,” he says, “for the ETL, the analytical modelers doing the data mining…You pay us for every customer you prevent from quitting.” Jackson’s presentation is focused on getting results from big data processing, a point he feels is getting lost in the race to analyze all the data.

Related Stories

Dispelling Predictive Analytics Myths

How 8 Small Companies are Retooling Big Data

Wharton Professor Pokes Hole in Big Data Balloon