Small businesses have always had one big advantage over their larger brethren: the close attention they can give to everyone who passes through their doors. But on the Web, that's been turned upside down. It costs a lot of money to teach a site to treat each visitor as an individual. The bigger a company is, the better it can afford the technology that lets it personalize content and features and the better its business will do. It's a paradox that all small businesses must struggle with, and there aren't any easy answers.
On the Web, individualization means highly customized, one-to-one relationships with visitors. It means doing simple things like calling them by name. It means setting up complex sites where pages magically mutate to show the information they are most likely to find useful. Most of all, it means using a database to drive the site.
It's the database stupid!
Most e-commerce sites are already using databases, but in a different form. A shopping cart system relies on a database, but it keeps track of data by putting that data in front of the user and asking what changes are needed. All the interactions involve asking for various lists of what is or what could be in the cart.
A Web site with individual customization, on the other hand, retrieves preferences that the user has expressed sometime in the past and then makes decisions about what content to display. There are lots of rules about what is or isn't relevant to the visitor, and these are combined with the preferences they've either wittingly or unwittingly indicated. The customer sees neither the rules nor the preferences, but instead the results of applying the rules to the preferences. Ideally, they don't know it's happening.
How important can a database be to a Web business? "I really can't imagine how any business Web site could do without good database customization," says Rob "Commander Taco" Malda, who founded the popular and highly customizable alternative news site at www.slashdot.org. Slashdot's database system is very complex and has been widely imitated. Readers can tailor the display to show what kind of news items they want to see, and the system can also alter pages automatically depending on the speed of their Internet connections.
Malda, of course, is plenty comfortable creating his own elaborate database programs. Why should anyone else go to the trouble? For Tim Carter, a syndicated newspaper columnist and former construction contractor, it was simply a question of giving visitors to his site easy access to the information they wanted. His Ask the Builder Web site (www.askthebuilder.com), which launched in December of 1995, brings together tons of information and advice. Carter says running his site from a database "was absolutely part of the plan" from the beginning.
Carter uses databases to provide quick access down several layers deep into the hierarchy of information stored on his site. The two biggest sections of the site, "Tim's Library" and "Tim's Tips," are driven by databases. For instance, select the topic "chimneys," then choose "construction" or "downdrafts" from the list of possible subtopics on the next page. Soon you'll know more than a London chimney sweep. What's happening is this: As users move down a hierarchy, the system "remembers" where they came from and can give them ever more specific information about the topics they're interested in.
The Absent Minded Database
But even though askthebuilder.com's database can remember the path a user takes in a given session, it doesn't hold onto that information from session to session. In other words, it doesn't remember if a user's been there before. That puts it one step behind most of the big sites. There are several different ways to improve a site's long-term memory, and none of them are cheap.
First, the site needs to collect information about who the visitors are, or give them some way to identify themselves to the system. Then that information can be added to the rest of the information in the database. Most sites collect this information by either asking users to register and log in every time they visit, or by dropping a file (called a cookie) onto a user's hard drive and checking for that file every time the user accesses the site. Many of the biggest sites do both.
The Cost of Competetiveness
The problem for small sites is that creating a database system and getting it to work with features like cookies will almost certainly be a major outsourced project. "Right now, your best bet is to hire independent contractors to code your database pages by hand," says Michael Schneider, vice president of IMC Online, an Atlanta-based Web presence provider that has worked on the sites of many small businesses. "And it takes a lot of hours expect it to cost several thousand dollars, at least."
Carter says the initial cost of coding the custom database pages for askthebuilder.com was $5,000 but don't expect it to always come that cheap. "I was fortunate and found creative programmers who built the mechanical aspects of my site," he says.
For now, IMC's Schneider advises small sites to wait to add fully integrated databases. "They get badly burned because they try to use the latest technologies and end up hiring developers who are inexperienced with those technologies," he says. "The projects go so far over budget that small businesses simply run out of money to finish the projects, and they fold."
Schneider advises small businesses to wait on personalization until the process has been hammered out, packaged up, and offered at discount prices. This is already starting to happen: One company, called GuestTrack, is pushing a product that lets Web designers embed specialized markup tags to insert different content according to how the system recognizes individual users. (For example, a tag like "#user[name]#" will insert the current user's name into the page that user sees.)
GuestTrack runs on Unix and Linux servers and costs $6,000, but according to the company that should by no means be considered the total cost. Most of its customers use it in combination with other products and technologies, and the average total cost of a customer's site is around $50,000. That may be beyond your company's reach or may seem like a bargain but keep it in mind as a guideline.
One way or another, the personalization paradox means that small businesses end up paying big. They can wait until the tools have been refined and work at a competitive disadvantage in the meantime or they can take the plunge, pay for the privilege of being early, and hold their breath until they find out if it's come in over budget. What's the right choice? That's a personal decision.