While Web design so far has been largely improvisational, Internet analysts predict the most successful next-generation Web sites will offer each visitor a highly personalized interactive experience.
"To date, Web site design has been typified by experimentation, massive start-from-scratch overhauls and a hodgepodge of features and information," says Paul Hagen, an Internet analyst with Cambridge, Mass.-based Forrester Research. "However, this catch-as-catch-can design is no longer valid."
Mr. Hagen, lead author of the Forrester report "Interactive Technology Strategies," says Web sites should be designed around an "action-driven" interactive paradigm, in which every element of the site blends together to enable a visitor to quickly accomplish a task.
One site that's leading the way in this area is eGenie, a personalized arts and entertainment guide designed by Cambridge, Mass.-based Bowne Internet Solutions. This site keeps up with users' entertainment preferences over time without being intrusive.
The eGenie site was originally designed as a marketing tool to introduce hordes of Net users to the OpenSesame software behind the site. AT&T WorldNet now uses eGenie as part of its arts and entertainment services. And other commerce sites are looking into adopting the software.
Analyzing without fuss
The core of the eGenie site is its OpenSesame learning agent, which analyzes users' interests without requiring them to fill out long questionnaires. Instead, the agent tracks the path each user makes through the eGenie site and uses artificial intelligence and machine learning techniques to discover the visitor's preferencesÅ"and to make product suggestions based on that ever-growing knowledge.
"The thing people liked most about the site was how eGenie learned without telling it to do something," says Alper Caglayan, Bowne's VP-solutions strategy. "In particular, users enjoyed learning when a band released new CDs and finding out about local concert tickets matching their individual preferences."
In addition, eGenie's learning agent adapts to users' changing interests. A user regularly visiting the site who develops a yen for science fiction six months later, for example, will find that over time the Web site automatically begins recommending more sci-fi entertainment.
The smart eGenie
eGenie also has the ability to learn about the temporal habits of its users. A daily eGenie user who is business-oriented during the week and leisure-oriented during the weekend, for example, will find his or her personalized daily eGenie Web page will start reflecting those differing orientations.
Bowne won't release precise figures for start-up costs for the site, but Mr. Caglayan does say the eGenie site took four peopleÅ"a software engineer, scripting engineer, interactive architect and a graphics designerÅ"three months to build it. Maintenance is done by one person, who handles content management and technical support.
Plans for the site include enhancing privacy features and enabling the learning agent to become even more sophisticated about personal preferences based on site activity history.
"We have learned a lot in what is needed in building an online business where personalization is the strategic architectural framework," Mr. Caglayan says.