As the search wars heat up, will relevancy be the make-or-break feature?

In my last column, I touched on the subject of “relevancy”, speculating that it might be Search’s contribution to the pantheon of metrics that include reach and frequency. In this column, I’ll take a closer look at the past and future of relevancy.

From a marketer's point of view, relevancy has always been with us. Sometimes called adjacency or propinquity, it is nothing more than the planner’s best practice of requesting that an ad run in close proximity to highly relevant content. An ad for a CD might run on the same page as a column by a respected music critic. But it is the instant interactivity of the Web that has brought relevancy to the fore. Advertisers can bid in real time for conspicuous placement under search terms relevant to purchase. This opportunity has fueled the explosive growth of Google, Overture and their competitors.

The search engines themselves think of relevancy in terms of customer satisfaction with their search results. Relevancy, therefore, is the holy grail for search engine code writers: coveted but elusive. Google’s initial and continuing success has much to do with the high relevance of its results. The “I’m Feeling Lucky” button on the Google homepage has always been a bit of implicit braggadocio.

First generation search engines used various criteria to construct relevancy ratings primarily based on the presence of the search terms in Web documents. This was known as “on the page ranking,” since the engine looked at content on the page to determine relevancy. Increasingly, search engines are ranking results by looking at “off the page” criteria such as inbound links and peer ranking. 

While the search engines compete fiercely with each other to deliver an ever more relevant crop of search results to the user, relevance itself turns out to be surprisingly hard to measure. This is because the very concept is highly subjective. Relevancy seems to be in the eye of the beholder, so automation has had little impact. The search engines all have panels of human editors who are employed to fine tune results. However, if customer satisfaction is the goal, the trend just might be going in the wrong direction. A survey by NPD new Media in 2000 detected an average search success rate of 81 percent, while a Nielsen-Norman survey from late 2003 showed a search success rate of only 56 percent.

Perhaps this is why we have a tantalizing hint that Microsoft’s newly launched challenge to Google just might feature relevancy. A search blogger recently caught a glimpse of what the new MSN Search product might look like when he stumbled across some cached pages he probably wasn’t supposed to see. The big news was that MSN was using “relevancy sliders” that allowed users to manipulate the ranking of pages by adding or reducing the weight of certain factors. The three sliders on the MSN test site allowed users to dial up or down the degree to which a site was static vs. recently updated, was very popular vs. less popular, and was an approximate match vs. an exact match.

This kind of relevancy personalization -- which has previously appeared on Yahoo! Smart Sort and the Kayak travel engine -- might just turn out to be a key front in the search wars to come.

Bob Heyman is Chief Search Officer at Mediasmith, Inc. and is the co-author of Net Results.2 (New Riders) and the Auction App (McGraw-Hill). He co-founded Cyberanutics (acquired by USWeb in 1997) where he is credited with pioneering Search Engine Optimization (SEO) and Search Engine Marketing (SEM), and he was also co-founder of CybeReps (acquired by Interep). His Rock Opera “Rock Justice”, co-written with Marty Balin of Jefferson Starship, is in the Book of Rock Lists as one of the Top Ten Most Obscure Rock Operas.

Mediasmith is a full service advertising media agency headquartered in San Francisco.