The Risky Business of Predictive Analytics
As the football season comes to a close, it seems appropriate to take a lesson from the NFL.
In 2011, a dramatic game took place between rival teams the New England Patriots and the Indianapolis Colts. With two minutes and eight seconds remaining, the Patriots led 34 to 28. It was fourth down, two yards to go for a first down, and the Patriots had the ball on their own 28-yard line. If they had punted, they would be giving the Colts and future Hall of Fame quarterback Peyton Manning a chance to score and win the game. But if they converted on fourth down, they would maintain possession and almost certainly go on to win the game. Patriots coach Bill Belichick decided to go for it. However, the Colts defense stopped the Patriots and scored a touchdown with 13 seconds remaining to win the game 35 to 34.
Predictive Analytics in the NFL
Did Belichick make the wrong call? The media seemed to think so. "Ludicrous and ridiculous," stated one paper. "Inexplicably arrogant and football suicide," stated another.
However, let's apply predictive analytics to give Belichick some credit. Statistical analysis since 1971 has shown that NFL teams should try converting fourth downs more often. In fact, it is projected that in a single season, teams should have attempted a fourth-down conversion 160 percent more often than they did.
So why don't coaches go for it more often? It's risky!
As Nate Silver stated in ESPN the Magazine, "NFL coaches aren't irrational or necessarily ignorant of the statistics as much as they are poorly incentivized to get these decisions right."
If Belichick had decided to punt, few would have criticized that move, even if the Colts ended up scoring anyway. But Belichick saw the opportunity to secure the game. It didn't work out, but statistically it was the right move.
Understanding the Risk with Predictive Analytics
Consider another scenario. Suppose you are a salesperson. It's coming to the end of the year and your company is pressuring sales to hit its goals and giving salespeople the opportunity to discount the company's services. Using predictive analytics, your manager has recommended that certain accounts be offered a discount (because that is what it will take to make the sale) and certain accounts not be (because they will likely buy anyway). Aren't you tempted to give everyone the discount? After all, if a customer does not buy, at least you want to say, "I did everything I could to make the sale." Predictive analytics can be an excellent tool as long as you provide the right incentive for people to put it to use.
EMC Embraces Predictive Analytics
In Walker's "Customers 2020" study, it was forecasted that customer service will become more personalized and proactive. EMC, a provider of data storage solutions, provides a great example of this.
Most customer support centers have a process to gather feedback from their customers after they have addressed issues. EMC issues short surveys to make sure issues were resolved and customers are happy. But what if the customer does not provide feedback? Most companies would just assume everything is okay. EMC took it further. By incorporating additional customer intelligence with whatever feedback it received, it was able to project which customers are likely having problems. With this information, EMC proactively contacted customers to see if they had problems and would work with them to get them resolved.
There was risk involved—what if they contact a customer who doesn't have any problems? Wouldn't that customer be annoyed? Even in these situations, customers were generally impressed that EMC reached out. And consider those who were having issues. The company was showing it was eager to take care of them.
A Growing Need for Predictive Analytics
Customer expectations are increasing rapidly, and companies need to apply advanced analytics to better anticipate customer needs, deliver the right solutions, and provide service in a proactive manner. Analytics will help optimize the insights they receive.
Will predictive analytics always be right? No. But if companies develop practical uses for analytics and provide the right support and incentives, it can be an excellent way to develop lasting customer relationships.
Patrick Gibbons is a principal at Walker, a leading customer experience consulting firm. You can read his blog at http://blog.walkerinfo.com/blog/engaging-the-enterprise. He can be reached at pgibbons@walkerinfo.com.
Related Articles
Applying Analytics to Build a Sales All-Star Team
22 Aug 2014
Focus on real-time visibility, not past performance, for sales success.