05/04/19 A CYBERSHOPPER'S BEST FRIEND

Return to main story

A CYBERSHOPPER'S BEST FRIEND

Amy Rommel uses the Web a lot at work. But the 25-year-old public-relations director doesn't have time to compare prices when buying CDs online. Not, that is, until she discovered Lycos Inc.'s shopping service. Using the search engine's technology to scour the Web, Rommel found a way to come up with deals at the click of a mouse button. ''I knew there were other music sites, but the Web is so big, it's hard to keep up,'' she says.
Shopping services, which include helpers, or ''bots,'' are popping up all over the Web. They range from true bots--such as the one Lycos uses--that comb sites for prices each time a request is made, to hybrids, such as Compare.Net, a Web site that lists comparative product information.
These clever programs promise plenty of bargains for consumers--and headaches for merchants. Early online merchants often charged higher prices than physical stores because customers would pay for convenience. But bots undercut the convenience premium. A recent Ernst & Young study found that 87% of the 30 consumer products tracked could be bought online at the same price or cheaper than in retail stores. ''Margins are going to be pushed down and down,'' says Walter A. Forbes, chairman of Cendant Corp., which runs an online shopping club.
The timing is right. Web shopping is on the rise--10.3 million people bought online last year, up from 6.3 million in 1996, according to Jupiter Communications Co. That's prompting Web sites to add bots to their lineups. Search engines Yahoo! and Lycos offer shopping bots through partnerships with Junglee and Info-Space. Last fall, Excite Inc. acquired shopping service Netbot. And this month, Microsoft Corp. bought Firefly, which recommends products based on a consumers' tastes.
How do they work? Once a request is made, the agents hunt down product information as well as prices and reviews. Excite, for example, tracks products from 500 merchants. Consumers visiting the Excite shopping area type in the name of a specific product. The service then goes to different merchants' sites and hunts for current data and prices.
Yet bots have flaws. These first-generation agents can sometimes misidentify data on retailers' sites. A search for a Tickle Me Elmo doll on Yahoo! Inc. brought back information on that toy--as well as Elmo sheets, Barbie dolls, and Tickle Me Cookie Monster. Accuracy will improve in the coming months with a new language called XML, which lets site designers create ''tags'' to define data on a site, making it easier for bots to pick out the right information.
And there's a new generation of bots in the works. Massachusetts Institute of Technology's Kasbah can negotiate based on price and time constraints that it is given. And companies such as AgentSoft Ltd. create software that lets corporations build their own bots.
So what will happen when bots are really let loose on the Net? Sure, there will be bargains. But in computer simulations at IBM's Institute for Advanced Commerce, bots set off price wars. Some concerned companies already are blocking bots' access to their sites. But experts say the best solution is product differentiation and bots that take into account more than price. If that doesn't work well, bots may just turn out to be too much of a good thing.
By Heather Green in New York

Return to main story