The perfect place for a data set. (Photo credit: Wikipedia)
IBM might not seem like the kind of workplace where metaphors are bandied back and forth, but it is. Because selling predictive analytics isn't exactly like hawking beer -- groups of friends won't lift their data sets after a long day at work -- metaphors are in high demand at Big Blue, and they take a little more to build than the average literary device.
For instance, IBM believes its five-year-old U.S. Open SlamTracker, which analyzed all Grand Slam tournaments this summer, is the kind of metaphor that can help business managers understand why predictive analytics apply to them. If you've watched U.S. Open coverage, you've seen the commercials already -- the promise it can help any business, from bakeries deciding what items to make each day to cops knowing where to go before the crime even happens.
Commercials help, but augmented TV is another level. Today at Arthur Ashe Stadium, when David Ferrer continues his quest for upset over Novak Djokovic or when Serena Williams mounts a home-country championship run, IBM data will be there to help fans analyze the game in ways John McEnroe can only intuit from years of experience.
"We are analyzing historical data over the last seven years from four different majors every year," says Deepak Advani, vice president Business Analytics, IBM Software Group, explaining the data points range from the basics (opponent vs. opponent, court surface) to the subtle (weather and a certain time of day). "We decided, let's capture every point. It ends up being almost 39 million data points that we're analyzing."
That would overwhelm a fan, so IBM has simplified the data down to critical elements in each match-up:
As data from the match is added, the analysis grows more refined. IBM has a long relationship with the U.S. Open, but understanding tennis is not the end game for SlamTracker.
"Say you're in auto insurance and there's a car accident and a claim's coming," says Advani. "Companies need to know what is the likelihood that this claim is fraudulent? And typically what companies will do is very similar to the predictive analytics used in SlamTracker. Applying the same algorithms will return, 'Here are the most important predictors for fraud.' And companies are saving lots of money, we have one client who got 400 percent ROI (from this technology), not to mention the positive equity for people like you and I who get approved right away."
Part of the power of a tool like SlamTracker is it helps you imagine how human interaction with data improves its accuracy. Good sports data would need to adjust to different strategies in real-time. Human developers have to recognize such variables to make the data work. And then there's the application issue.
"The real power of analytics gets unleashed when every decision is being made better in the enterprise," says Advani. "That means people with zero visibility into analytics -- say a call center rep or a bank branch teller -- should make better decisions based on analytics without know anything about analytics."
Some call centers, for instance, are using the technology to help employees provide the smoothest experience based on the customer's previous interactions and usage trends, but also be able to sell the exact right products based on those same data sets.
"One client asked why should call centers be cost centers?" says Advani. "We could also use this as a revenue-generating channel."
Speaking of revenue, it's not exactly as if sports is generating chump change. Could predictive analytics be the next big advantage for major sports organizations? You might predict the answer:
"You use analytics for customer acquisition, right?" he says, "But the same technology could be used for talent acquisition on football teams."
Is that bad for sports? Is there such thing as data doping?
"I think there could be rules to make sure the games stay pure," Advani says. "But we have one client in Europe who wants to use analytics to reduce injuries. That same idea could be applied to football here in the U.S. as well."
Advani says applying data to games at this level has not been received with as much suspicion as other technologies, such as instant reply. And that may be because there's another metaphor at play -- the predictions coming from the likes of Brent Musburger, John Madden and onward.
"You've always had experts," Advani says. "We're just using a little more science here."
