Wine, Movies and Love: Data Can Predict Winners

[View:https://www.youtube.com/watch?v=iJZawg2ZPdY]

Executives and managers use "gut feel" to make business decisions most of the time, according to a study by Bloomberg Businessweek Research Services (BBRS). The problem is, "gut feel" often leads to poor outcomes-and keeps companies from exploring profitable areas that intuition often will not lead them to.

"Math matters more than human judgment," declares Ian Ayres, a professor at Yale University Law School and author of Super Crunchers: Why Thinking-by-Numbers Is the New Way to Be Smart.He says any substantial organization that is not using predictive analytics-such as regression analysis, which are mathematical models to predict the result when different variables interact-"is presumptively screwing up."

Analytics can uncover many surprising avenues to profit. Take Epagogix, a company that can accurately predict the success of a movie before it is filmed by analyzing elements of the script using an algorithm that compares 200 variables from thousands of past movies. Ayres notes in his book that the formula predicted the film The Interpreter, starring Nicole Kidman, would gross $45 million in the domestic box office-a mere $2 million off its eventual gross.

Perhaps even more interesting, the model suggested tweaks that would have raised the film's box office. "If the film had given a sidekick to Nicole Kidman and focused more on New York, it would have added $12 million to the box office," Ayres says. It turns out that location is really important in real estate and movies-audiences get confused if a film switches between too many different places.

He points to numerous other examples where analytics leads to counter-intuitive success. eHarmony, the matchmaking service, uses regression analysis to determine the compatibility of potential daters based on thousands of personality surveys of successful couples. eHarmony matches "you with people you might not have thought you'd be interested in," Ayres says.

The BBRS study found that 61 percent of executives make decisions by "gut feel" at least half the time (see chart, "Gut Feel Dominates Business Decisions"). However, algorithms can only improve on human judgment when fueled by sufficient data. For example, Orley Ashenfelter, an economist at Princeton University, was able to predict the quality of Bordeaux wine as it aged in barrels by comparing information from many vineyards over many years. His algorithm used four variables, including the amount of rainfall and the average growing temperature for the grapes, to predict the quality of the wine before anyone tasted it.

Analytics do not help with one-off decisions, like whether it makes sense to construct a 10,000-server building in western Kansas if no one has ever done such a thing before. "You need comparables," to predict success, Ayres says.

He adds that when companies first begin using analytics they should expect "the iron law of resistance" from people who do not want to trust numbers-based decision-making over intuition.

"Robert Parker, one the great handicappers of Bordeaux, was not amused when Orley started doing first predictions and said some intemperate things" for attempting to do so, Ayres says. Yet, "Orley's predictions year after year were more accurate than some of Parker's."

And so over time, people throughout organizations see the benefits of analytics-based decision-making, especially as the data guides them in unexpected directions that pay off.

Andrew McAfee, a research scientist at the MIT Sloan School of Management, wrote in Bloomberg Businessweek that counter-intuitive measures succeed because people often do not understand what they really want. "It turns out that we often like having fewer choices instead of more, even though when asked we'll always express a preference for more, more, more," he notes. "Choice can be confusing, paralyzing, and worse than unsatisfying-it can be dissatisfying."

McAfee adds that people may get a lot of value from having their autonomy taken away and from technologies that put constraints on their behavior. He says one example is web-based commitment contracts. McAfee specifically cites a website Ayres cofounded called stickK, which allows people to set up such contracts to meet personal goals, like quitting smoking or exercising regularly.  

Ayres says commitment contracts work best when people put money at risk, have a referee monitor if they are meeting their goals, and make the outcome public so friends and coworkers know if they succeed or fail. Ayres, for example, set up a $500 commitment contract for himself that he would lose a pound a week for 20 weeks-and if he failed his money would automatically be sent to a charity of his choice. In his case, he tweeted whether he was losing weight. And he met his goal.

One interesting discovery is that people tend to stick with their goals more often if they choose "anti-charities"-that is, failure would mean their money would go to a charity whose views they disagree with.

"Our English soccer fans often choose a football club they can't stand," Ayres says. "It's another way we care about other people; we care about who will be helped by our success or failures."