You recently wrote that "employees hired for their expertise with numbers ... make the best decisions: big and small, every day, over and over." When do you think strictly following the numbers works best and when might it cause trouble?
People don't need encouragement to make intuitive decisions; it's easy, fun and kind of romantic. Blink, for example, is a romantic book suggesting that intuition is a marvelous guide to decision making, but it's misleading. Many of the situations described have been preceded by years of analytical work. Wherever possible, explore the analytical option. It's not always possible--you may not have time for analytics. If you can't get any definitive data, or human motivations are involved, you have to use intuition.
So a start-up company that doesn't have five years' worth of data to draw on should perform small tests and analyze those smaller sets of data?
I'm enamored of the test-and-learn culture. Most organizations could use more rigor and discipline in their experimentation. All companies experiment, but most don't learn anything from their experiments. Gary Loveman, CEO of Harrah's--a company that's achieved a huge turnaround on the strength of its analytics--says not using a control group is one of the ways you could get fired there.
Some describe business intelligence as a rear-view mirror, and they stress the need for real-time analysis. Do you share that view?
Real-time data has value, but from an analytical standpoint the more interesting thing is the rise of predictive analytics. Advertising agency DDB has an analytics company that took two years to gather enough data to do predictive analytics on the potential implications of a particular ad campaign in a defined market.
Is there an increasing need to make decisions quickly--to react to competitors, for instance?
In general, the pace of decision making has accelerated. There's more automated decision making going on. For decisions that are made frequently and that need to be made quickly, such as whether to issue a mortgage or an insurance policy, automated decisions are something to think about.
Rules are applied in banking and credit-related fields, too, but what other industries are making more automated decisions?
The airlines pioneered this a few years ago with yield management, but now you see it in department stores--should I cut the price on merchandise? You see it in hotels; even some apartment complexes are performing yield management to set rents. Operational decision making is also happening in energy. Some companies that didn't lose power in the 2003 outage had good automated decision-making systems that shut down their power grids in time.
What's key to using analytics to your advantage?
Analytics are more effective when they're narrowly directed. Companies that do it well focus on specific goals--in Harrah's case, it's customer service and customer loyalty.
So you ought to know your business well enough to know where to apply analytics?
You need to know what your distinctive capability is or what could be your distinctive capability if you were better at it. Some companies don't have a clue. It's the same problem companies have found with data warehouses; they become data landfills in many cases because people throw so much stuff in there not knowing what they might need.