Search results are ranked according to a formula which supposedly measures how many users click on a certain site for a given search term. For example, the most popular site for the search term "strange carrot" would come up first in the rankings. This is different from the usual procedure on search engines like Excite, which rank sites based not on popularity but rather on a keyword relevance score. Direct Hit also seeks to take into account the length of time users stay on the site after they click on it. Apparently, Direct Hit follows you around!

I say "supposedly" and "apparently" because, although the idea here is fascinating, Direct Hit's technology doesn't seem to be entirely functional at this stage. A lot of the "same old garbage" (outdated links that have been ranking high on Inktomi-driven search engines for ages) seems to come up under a lot of search terms, so one tends to question if many of the sites in Direct Hit's current rankings are genuinely popular.

That's likely to change in the future. The company was recently acquired by Ask Jeeves for $507 million in stock. Direct Hit is becoming an important player for webmasters to take account of, because several major companies, such as Go2Net and ZDNet, have started to build its popularity ranking into their services. As Direct Hit is used more frequently, and as the company works to improve the technology, the results should become more reliable.

Popularity engines have enormous potential to help us sort through Internet clutter. For many users, they may act as a proxy for relevance. Why? If users are lingering on a site which comes up in a search engine result for a certain search term, it's safe to say that they're finding it relevant, or at least compelling.

Then again, if Direct Hit refers them to a mediocre site, they may still linger because they're not aware of what else is out there. Direct Hit could potentially create self-perpetuating high rankings for mediocre sites which just happened to rank on the first page in the early days. A remedy would be to cross Direct Hit technology with a selective human-edited directory, such as Looksmart, About.com, Suite 101, or other "Best Of the Web" rankings. The company claims to have something called a "directory engine". With further development and more usage, the goal of crossing a sophisticated popularity engine with specialized or human-edited directories is realizable.

But what of other popularity rating services? After all, various companies offer rankings of the most popular web sites, compiled through a variety of methodologies. Some, like WebSideStory, collect detailed information about surfing habits on behalf of webmasters. Their "Hitbox" product and their "popularity engine" Yep.com both have the potential to track site popularity in a manner similar to Direct Hit, finding out which sites for given search terms are most popular with users. The Hitbox is favored by mostly smaller webmasters, so its data may be useful for finding the best or most popular sites in certain categories which may nonetheless be off the radar screens of the larger stats gathering services. If you're in acquisition mode, finding the best-loved "little guys" was never easier...provided your quarry has the Hitbox installed.

Alexa is an industrial-strength version of Yep.com. Users of the Alexa toolbar can look at vital site info, including the number of Alexa visits, which may offer a rough guide to a site's popularity. Alexa, too, is a popularity engine, and many users of the toolbar voluntarily keep it switched on, possibly offering insight into surfing patterns.

In general, one wonders if the larger site ranking services like Media Metrix, PC Data Online, Nielsen Netratings, 100Hot.com, and so on, could be expanded or somehow crossed with Direct Hit-style technology to provide us with greater insight into which sites are most compelling and popular with users searching for given search terms. Or will they only sell this information to subscribers? We do know that such consultants can offer a ton of data on surfing habits; they track site stickiness and all the rest. But to link such data to search terms would be a real feat.

Finally, let's not forget Google and its secretive algorithm for measuring link popularity. Sites for given search terms which are linked by other reputable sites (measured by how many sites link to them, in a never-ending cycle of reputation measurement) are given the highest rankings.

These are clearly early days for the use of popularity ranking to assess the relevance of particular web sites to particular search terms. If the technology evolves far enough, we can look forward to the day when the best advice for those seeking higher search engine rankings will be: "build a wonderful site that people like."