"The key for us isn't necessarily eyeballs," says Mike Regan, E-marketing director at the Morgan Hill, Calif., company. The goal is to identify and retain repeat customers--especially those who buy high-margin items such as helmets and other cycling accessories. Regan says attracting and retaining those customers will require E-intelligence: information that helps companies use online sales and marketing initiatives to understand who their customers are and what entices them to buy. And while the ability to generate that type of data by integrating business-intelligence tools and database and transaction systems is still relatively difficult, the competitive landscape is such that the most aggressive companies are pushing ahead. "Everybody is going for the best customers right now," Regan says.

Until recently, the focus for many dot-com startups and click-and-mortar businesses had been on getting E-commerce systems up and running. Efforts to collect detailed data about online customers have been relatively limited. "Most people are just trying to get their arms around traffic analysis and page hits," says Wayne Eckerson, an analyst at the Patricia Seybold Group. Such basic clickstream data is used to report Web-site traffic levels, judge the effectiveness of Web-site designs and the popularity of site content, and calculate the ratio of new visitors to repeat surfers.

Ecampus.com, an online seller of textbooks and college paraphernalia, went live less than nine months ago. The company's priority has been to make sure that its E-commerce infrastructure--based on an Oracle8 database and Oracle Application Server running on 10 Sun Microsystems Solaris servers--can handle its rapidly growing Web traffic, which exceeded 1.3 million visitors in January. Its E-intelligence efforts have been limited to gathering and analyzing clickstream data using WebTrends Corp.'s Log Analyzer software.

By scrutinizing clickstream data, Ecampus.com managers can see how many of its Web-site visitors simply peruse the book or merchandise sections, and how many actually make a purchase. "What we're trying to find out is how many people are just browsing," says CIO Jack Garvin. Clickstream data analysis also reveals where Web-site visitors have come from, which helps Ecampus.com determine whether banner ads on other Web sites such as CollegeClub.com are generating traffic.

Garvin acknowledges that clickstream-analysis tools alone don't provide much behavioral information about Web-site users. Because students often visit Ecampus.com from computers other than their own (from college machines or friends' PCs, for instance), "it's very difficult to identify who our users are on a repetitive basis," he says. The bookseller is installing Net Perception Inc.'s Recommendation Engine, which suggests products to online customers based on the Web content they view, says operations director Brent Tuttle.

Online marketers increasingly are looking for answers to more-complex questions: Who are the most profitable customers and what marketing and promotional efforts attract them? What Web-site content most frequently leads people to buy online? Why do some visitors abandon their shopping carts before completing a transaction? And for companies with brick-and-mortar stores or catalog businesses, how do online sales correlate to sales from other channels?

Understanding what makes online customers tick is particularly critical for dot-com companies whose competition is always just a click away. Could a company such as Outpost.com, an online retailer of PCs and consumer electronics, survive without such efforts? "Sure, we could function as a business. But it would be easy for someone to come in and overtake us," says Dan Bachman, Outpost.com's business-intelligence director (see sidebar story, " Aggressive Analysis Puts Outpost.com Out Front").

Knowing more about online customers, from their ages and where they live to their interests and product preferences, is a goal for many E-commerce companies as they take steps to customize Web-site responses for individual visitors. "There's a growing recognition that personalization and one-to-one marketing won't work unless you already know something about your customer," says Forrester Research analyst Eric Schmitt. E-businesses are paying close attention to the growing debate over consumer privacy--and, in some cases, re-evaluating their privacy policies ("Mind Your Business ," InformationWeek, 3/6/00)--but that's not sidetracking their E-intelligence strategies.

One problem is that understanding customer behavior requires more than simply analyzing clickstream data, and taking it to the next step isn't easy. Just matching information about which Web pages an online shopper viewed and data about the shopper's purchase is beyond the capability of most clickstream-analysis tools. That's because clickstream data is generated by Web-page servers, while sales-transaction data (generally encrypted) is generated by commerce servers. So determining what Web-site content increases sales, and what doesn't, can be tricky.

An even greater challenge awaits retailers that want to compare online sales data with sales information from brick-and-mortar stores. That data is stored in different systems, often in different formats. And pity the poor E-marketer who tries to build a complete view of customers by adding call-center data and other operational systems to the mix. "The guys who developed Web servers never dreamed that anyone would want to analyze the log data, let alone integrate it with data from other sources," says Dan Drucker, general manager of Hyperion Solutions Corp.'s E-business division.

A growing roster of software tools from new and established business-intelligence software developers attempts to pull E-commerce data from various sources and combine it for analysis. "In the last six months, just about every vendor in the business-intelligence market has come out with an E-commerce tool," analyst Eckerson says.

The demand for combined online marketing technology and analysis capabilities has also sparked a wave of acquisitions and mergers. For example, Broadbase Software Inc., a business-intelligence software vendor, recently completed an acquisition of E-marketing application supplier Rubric Inc. And in January, DataSage Inc., a vendor of data-analysis and personalization software, agreed to be acquired by Vignette Corp., which sells content-management software for online marketing. In addition, E-commerce software suppliers are also adding business-intelligence capabilities to their products. Blue Martini Software Inc., which already builds data-mining tools into its E-marketing system, last month launched an E-commerce data-analysis service.

Specialized Bicycle is adopting some of the new E-intelligence technology. When the company began selling its products online, its Web site offered mere brochureware on its products. Though the company used Log Analyzer to gather clickstream data, Regan says, not much was done with the data beyond using the page-hit count to justify the cost of maintaining the site.

continued...page 2, 3

Illustration by John S. Dykes
Photo of Regan by Gary Parker
Photo of Garvin by Jim Callaway