InformationWeek
Defining The Business Value Of Technology
March 15, 1999
 Print this story |
continued...page 2 of 3
| Related links from our sister publications: |
|
|
The Times stores the information in an Oracle data warehouse. Using SurfAid tools, the site's research team can determine, for example, how often 20-year-old males return on a regular basis, what pages they hit, and--most important to advertisers--what ads they see and click on. The company also matches that data with demographic data that helps advertisers understand, say, the predisposition of a 20-year-old male in New York's Westchester County to buy a sports car. Advertisers use that data in making decisions about ad placement and frequency.
But the Times goes further: It hands the data over to IBM, which performs an even more detailed analysis to understand patterns of behavior in how people respond to advertisements. Using data-segmentation models, the approach can reveal patterns in the behavior of Web-site visitors that might otherwise go unnoticed. The system might find, for example, that readers who buy lots of personal electronics also spend time reading the travel section. "It gives a different behavioral look at how people respond to ads," says Peter Lenz, research director for New York Times Electronic Media.The Times' own analysis can predict Web-site visitor behavior with 50% to 70% accuracy, Lenz says. Using IBM's models, that behavior prediction accuracy can be increased to 70% to 90%. "It enables us to support the CPM [cost per thousand] for advertisers who buy impressions on our site," says Lenz.
In addition to the SurfAid Analytics tools and services, IBM intends to offer commercial products for collecting and analyzing Web data, such as WebSphere Site Analyzer, an extension of IBM's WebSphere application server line, expected in the second quarter.
John Payne, solutions executive at IBM, says that in the last several months Web-site managers have come to realize how much they stand to gain by collecting and analyzing Web data, and incorporating it with information from other sources. "They're really starting to understand that they need to go beyond hits and page views," Payne says. "They need to look at specific user behavior or the behavior of groups of users. You can really make some decisions about how to run your business based on data rather than just intuition."
There are few better examples of that than Dell Computer, whose much-heralded Web site now handles $14 million in orders per day, or about 25% of the PC maker's sales. Richard Owen, worldwide VP of Dell Online, calls the Web "manna from heaven" as a sales channel for Dell's direct, build-to-order strategy, but also emphasizes the immediacy of customer feedback. "Being direct, we've always had the benefit of good customer information," he says. "But the Web gives us real-time information on buying patterns. Being able to see the immediate effect of pricing is invaluable."
Dell's Web data collection efforts go beyond measuring pricing effects. "What I'm really trying to do is identify my best customers," says Owen. "And there's no off-the-shelf database for that. You need SQL and structural skills to parse data and draw conclusions, but the real differentiator is people. It's really a creative process."
At Dell, "technology is certainly an enabler," says Andersen Consulting's Johnson, "but the real focus is on execution. They take a small amount of data and execute in response." For example, online data helped Dell determine that its business customers had little interest in the original network computer concept. Dell combined traditional telephone feedback with E-mail messages and analysis of Web user behavior, such as configuring and pricing models online. That analysis identified the tendency of online browsers and buyers to focus mostly on powerful high-end configurations. As a result, Dell has been conservative in its approach to the NC--a smart move, at least so far.
Retailers selling on the Web see a world of difference in online data collection as compared with their physical-world stores. Barnesandnoble.com, for example, added software products such as TurboTax to its Tax Answer Center book area after noting the tendency of tax-book customers to navigate their way to software. "In a store, people might say, 'Don't you have any software?' but you'd have no record of it," says Brenda Marsh, director of merchandising at Barnesandnoble. com. "In the online world, you can watch your traffic patterns."
Barnesandnoble.com uses Accrue's traffic-analysis software in connection with Red Brick Systems Inc.'s data warehouse tools. The company is also considering GroupLens from Net Perceptions Inc., software that recommends products based on purchase histories of users with similar tastes. But "all these tools are in various states of suitability," says Marsh. "You also need to follow your expertise and your gut."
Rapid Pace Of Change
The immediacy of online customer feedback, plus the ability to make pricing or merchandising changes quickly, brings an unprecedented pace of change to retailing. But it also puts even more importance on restraint--knowing when not to react to the data. "I don't have to wait for someone to run a batch report; I can click on my sales report all day long and know what's moving there," says Marsh. "But you have to be careful with that and not make wholesale changes based on what happens in one hour. It's so tempting to see a spike and rearrange the whole offering."
continued...page 3
return to page 1
Read sidebar story, "Bear Stearns: An Online ".edu"-cation."
Photo of Lenz by Catrina Genovese
Back to This Week's Issue
Send Us Your Feedback
Top of the Page
Featured Microsite
