Live From NCDM: Delta of Analytics

Author and analytics maven Thomas Davenport was once invited to speak with a supermarket chain about information management. The chain’s managers proudly showed Davenport how they captured data at the point of sale via scanners, how they transferred the data to tape and disks, and how they sold the data.

“We make more from the sale of data than we do from meat!” they crowed.

“But what do you, yourselves, do with the data?” Davenport asked.

“We store it underground so it’s very safe,” they replied.

Davenport pressed them. “No, I mean how do you use the data for your business’s purposes?”

And the chain’s managers looked at each other. “We haven’t quite figured that out yet.”

Davenport is a man with answers. Information’s role, he said, can change with the times. When business conditions are flush, it can be used to create growth. During rough periods it can be used to cut costs. “The good news is that it can be used for both,” he said.

The key is using data for higher-level functions such as trending models; determining why certain results are occurring; predicting what will happen next; and attempting to steer results toward the best possible outcomes.

Unfortunately, what is currently called the state of the art in analytics is what Davenport calls “reporting” – generating reports that give snapshots of the very recent past. These are good things to do, he acknowledges, but they aren’t enough to bring an organization the competitive analytics realm.

The trick is making analytics and fact-based decisions a key element of strategy and competition, instead of blindly relying on models of past behavior, Davenport said.

By way of example, he cited the case of a Houston Marriott hotel that saw an unseasonable spike in bookings in September 2005. According to computer models, this would have been a perfect time to raise rates. But a marketing manager took a look at a newspaper, and surmised the additional traffic was coming from Hurricane Katrina refugees – and overruled the increase.

Davenport uses the acronym “delta” as a mnemonic for the elements an organization needs to become competitive on analytics.

D stands for data. According to Davenport, an analytics-focused firm has to have something distinctive about the data going into its systems if it is going to get something distinctive coming out of it.

Davenport cited the Boston Red Sox using Bill Jayme, a baseball-obsessed night watchman at a baked beans manufacturer, who generated new ways of evaluating baseball players’ likely productivity. Since hiring Jayme, the Red Sox have won two World Series championships after an 86-year drought.

For E, Davenport stresses the value of an enterprise-wide approach to the data integration. Yes, he rallies against siloing, -- unconnected repositories of either data or technology, and uncoordinated analytic talent. “Highly analytical companies say ‘let’s put this together,’” Davenport said. “You can’t really do this with a bunch of spreadsheets. Spreadsheets are useful tool – I call them the duct tape of business intelligence. But 20% of them have errors. Do you really want to build your strategy around something that’s that likely to be wrong?

“And if you want a multichannel strategy, you have to think about a multichannel organization. That’s particularly critical if you’re going to move ahead at the same rate in all these different channel domains.”

The L, leadership, is “probably the single most important thing.” According to Davenport, Harrah’s Entertainment Gary Loveman requires analytic backup for marketing assertions, and is fond of saying “Do we think, or do we know?” There are, according to Davenport, three ways to get fired at Harrah’s Entertainment: Steal; harass women; and fail to use a control group.

A company that wants to be competitive on analytics also needs Targets. “You can’t be analytical about everything,” Davenport said. “You need a focus for your efforts.” For Harrah’s the emphasis was placed on loyalty and customer service. The New England Patriots measure not only how to choose the right player, but the total fan experience as well. The team measures 25 elements of every home game, including the wait times for the women’s restrooms “In addition to winning the Super Bowl three times in the last six years, they also have the shortest women’s restroom lines of any team in professional football.”

Finally, a company that wants to be good at analytics has to have Analysts. Davenport estimates that 70%-80% of firms are analytics amateurs: Their number crunchers can use spreadsheets and reports, but that’s pretty much it. Another 15%-20% employ analytical semi-professionals, who can generate reports and simple models and use visual tools. But the analytics professionals are the 5%-10% of firms employing “hard-core propeller head types” who can create algorithms.

Candy maker Mars likes to hire what it calls “PhDs with personalities,” according to Davenport.

“That does tend to limit the hiring pool a bit.” But the personalities are essential: The PhDs have to do more than come up with the right answer – they have to explain that right answer to someone else.

Davenport spoke during a keynote presentation at the National Center for Database Marketing Conference on Tuesday. The conference runs through Wednesday.