Paul Wallich, Scientific American

If you're a typical consumer, your electronic mailbox has been sinking under a steady stream of offers from on-line merchants. By clicking on a link or typing a code, you can get 10 or 20 percent off your next order, free shipping for life, a $10 credit or any number of other incentives to shop on the Web. Sites have sprung up to disseminate the offers with names such as dealoftheday.com and virtualbargains.com, and informal networks of shoppers trade information on offers that, properly combined, can leave stores owing you money for having shopped there. Cumulative marketing budgets of these sites are in the hundreds of millions of dollars.

These kinds of come-ons have been part of the "brick-and-mortar" retail world since time immemorial: a loss-leader will get shoppers into a store, where they will buy a host of additional items on the spot rather than drive across town to compare prices. On the Web, however, commerce is supposed to be "frictionless"--competing stores are only a few clicks away, so there is little reason to avoid comparison shopping. There are even sites that do nothing but provide comparative price information.

So are on-line retailers crazy? Are they throwing their investors' money away hand over fist? On the contrary. Faced with the enormity of choices on the Web, people are less likely to make solid comparisons than they are in person, says Dan Ariely of the Massachusetts Institute of Technology: they retreat to the few sites they've already bookmarked and buy there regardless of who offers the best price or service. Last year an average book or compact disc cost about $2 more at Amazon.com (which commands about 80 percent of the market) than at Books.com (with about 2 percent), according to Erik Brynjolfsson, also at M.I.T. In fact, Brynjolfsson and his colleagues found that prices varied more widely on the Internet than at the brick-and-mortar stores they surveyed. Customers appear to be willing to pay for the security and familiarity that a well-known name gives them, especially because they don't have physical cues such as the condition of a storefront or the attitudes of the clerks. Profit, Ariely says, lies in reducing consumers' sense of uncertainty.

Indeed, Gal Zauberman of Duke University has run a series of experiments showing that people will stick with their first choice of search engine, bookstore or other Web service even in the face of evidence that another choice would be better or cheaper. Sites with low setup costs are much more attractive, he points out, because surfers want to get something done quickly, before a deadline strikes or their computer crashes. Many commercial Web sites require a lengthy registration process before delivering their goods--anywhere from 10 or 15 minutes to nearly an hour.

In addition, users must spend even more time to master the idiosyncrasies of a particular site's organization, page layout and search engine, Web guru Jakob Nielsen notes. You can't just wander the aisles of a virtual store to get a sense of where products are. Nielsen's studies indicate that most people are willing to spend only one or two minutes figuring out how to use a Web site unless it provides an immediate payoff of some kind. The only frictionless part of Web commerce, he quips, is people clicking away from a site they don't like.

To hook newcomers, a site must either follow the design rules embodied by existing sites or else be roughly twice as easy to use, he says. (Such an improvement is well within reach, he observes, because many sites are hard even for experts to figure out.) Or it must offer some other incentive. Seen in this light, a credit of $5 to $20 may be fair compensation for the time lost in switching to a new merchant portal or search engine.

Furthermore, once consumers have invested the time to find out whether a new site is better than the old one, they are unlikely to switch back, so it's in the interest of Web merchants to offer incentives to their current customers as well. Whoever is still standing when the venture capital runs out--so the current theory goes--will be among the winners. The total share value at stake among the Internet companies fighting these brand-recognition wars is about half a trillion dollars, Brynjolfsson states.

It's not just a matter of establishing brands before the money runs out and companies have to make profits, Brynjolfsson and his colleagues argue: there is a technological threat that could make current Internet brands obsolete. Ariely is one of those working on so-called intelligent agents that will not merely compare prices but levels of service--they may even choose new products for you based on knowledge of your preferences. For example, he predicts, you might tell your agent you want to buy some wine, and it could suggest a particular varietal and vineyard based on characteristics of your previous purchases. The identity of the wine merchant who fulfilled the order might be completely irrelevant.

Along with agents running on your computer (or perhaps in competition with them), there will also be "infomediaries"--entities Brynjolfsson and Zauberman liken to Consumer Reports or Underwriters Laboratories--that vouch for merchants to consumers, and vice versa. Such organizations would go well beyond the simple price comparisons available today to rate service quality, delivery schedules, reliability and so on. They would tell you everything you might want to know before entrusting your time and money to an anonymous bundle of bits.

Such agents, whether local or networked, would end up knowing a great deal about their users and exercising a powerful influence over their choices, Ariely explains, and the companies that build them should be highly profitable. So how will we choose the most useful and responsible agents, given that switching from one to another could be quite difficult? Let's hope that there will be more to the decision process than just relying on the comfort of a well-established brand-name.