The book Touching the Void, an account of an ill-fated rock-climbing trip in Peru, was written decades ago, sold few copies and was already out of print when Jon Krakauer's Into Thin Air, about a disastrous expedition up Mount Everest, hit the best-seller list. When Amazon referred readers of the latter to the older book, orders went through the roof.

This episode was an inspiration for The Long Tail theory, espoused by Wired editor Chris Anderson and built around the notion that the lowest common denominator to which most products are marketed does not reflect what most people want. In the realm of statistics, "the long tail" is used to describe a graph in which a high-amplitude event or population is followed by a "long tail" in which the amplitude trails off but can cumulatively outweigh the peak in the graph. Translation? A business might have more to gain from niche products, for which customers will pay higher premiums, than from common goods.

Had the publisher known ahead of time about a resurgence of interest in climbing survival stories, it could have reissued the book sooner and marketed it alongside related books.

On the Web, the ability to identify niche interests requires an ability to flag certain types of site traffic. BI and Web analytics can be used to build benchmarks so enterprises can pick up on unique behaviors and develop niche offerings that bring customers into the long tail. "Web analytics can track preferences and behaviors so companies can build on segmentation dimensions," says Bill Gassman, Gartner's research director of BI. "As the dimensions build, you can correlate them against behavior patterns to gain insight into what future marketing campaigns should look like."

Companies should recognize that their brand is not the only bridge by which customers can find their products. Pursuing the long tail means seeking broader visibility beyond the head event and across the tail of the graph. The theory works if there is a head of "hits" and a tail of many niches. Traction builds as people are methodically led from a trusted, known place to one that is unfamiliar. "The day will come where enterprises will no longer be able to change Web sites to make people behave the way they want; rather, they will have to look at behavior to understand what people want at the head and then lead them down into the tail," says Gassman.

Search technology, too, could be used to lead enterprises to the long tail. The technology is evolving to support unconventional means of finding information, such as searches of blogs and discussion threads. This could help point the way to highly profitable niches.

"Today, a few keywords produce the majority of a site's current traffic, but most Web sites contain and can target thousands of different keywords," says Stephan Spencer, president of Web design and consulting firm Netconcepts, who says the "unbranded" search potential for the average retail Web site is nearly 40 times greater than that of branded search. In other words, if the keyword "Pfizer" brings in ten visitors a day through search engines, there is the potential to attract 400 visitors a day through long tail search terms.

The BI and Web analytics technologies needed to discover the long tail exist, but can they be used effectively? If companies don't trust their data, or if there have been too many false alarms or blips that didn't mean anything, a company could end up investing in market quarks rather than real potential customers. --Susana Schwartz