TORONTO – The most important element of successful business intelligence may be knowing when to ignore it, analytic theorist Tom Davenport told a Canadian IT industry audience Thursday.
Speaking at a breakfast symposium sponsored by SAS Canada, Tom Davenport, professor and director of research, Babson Executive Education,Babson College, discussed some of the main points from his book Competing on Analytics: The New Science of Winning. Despite the inroads software companies have made in deriving insight from customer and transactional data, he said intuition plays a bigger role in successful analytics than most people recognize.
“It’s not the data, the quality of the equation or the fit of the regression line,” he said. “People have sometimes wondered if, because of analytics, intuition will disappear . . . but if you think about it, even the process of forming a hypothesis is an intuitive act.”
Davenport went through a number of the Fortune 500 companies he profiles in his book which use analytical software to guide their business processes. These include Marriott, which uses its business intelligence reports to determine how much to charge for rooms. In some cases, however, unexpected events like Hurricane Katrina could inflate demand without providing the right context, so Marriot employees may override the system. That’s not the case at casino and hotel operator Harrah’s in Las Vegas, he said. There, employees are forced to dogmatically stick with what the system tells them.
“It’s a funny thing, this question of overriding the system,” Davenport said. “Some companies expect you do it. Other companies will fire you for it.”
In partnership with Accenture, Davenport said he has conducted two surveys on the adoption of analytics as a business strategy among a sample of 600 firms. The first took place in 2002, and the second took place last year. According to his research, 47 per cent of firms surveyed last year demonstrate above average analytic capability, compared with 19 per cent in 2002. Another 37 per cent have at least some capability, whereas only 27 per cent did four years ago. Davenport suggested this might be in part due to a maturation of analytic software tools. Reporting tools, for example, are moving away from merely telling users what happened but are offering cause and effect information.
At its SAS Global Forum user conference earlier this week, SAS chief executive Jim Goodnight said embedding analytics in the enterprise is a large undertaking, though the benefits usually justify the effort.
”These are enormous computational problems involving a million rows and millions and millions of columns of linear and non-linear programming optimization,” he said, using the example of applying business intelligence in retail organizations. “It can take quite a few hours to perform the analysis, but once you’re finished then you understand how to do markdowns, and you understand how to do promotional pricing.”
Davenport told the Toronto audience successful analytics takes time, and that many Canadian organizations were only at the early stage of this wave. “It’s not a black box thing by any means,” he said.
–with files from Joaquim Menezies