This only makes sense for high ticket items. For low ticket items, it's simply not scalable because being able to effectively frame product relevance means being about to truly understand the customer and his problem. In many cases the only information you have to go on is the customer's search query. Beyond the search query, all the other information necessary to determine relevance is effectively trapped in a buyer's head. Unless you have an effective way of extracting enough data about the customer's problem or need, you are pretty much S.O.L. when it comes to determining product relevance.

For low ticket items, the questions now are "How do we capture more information beyond a simple search query?" and "How do we further promote product darwinism and eliminate the influence of advertising, marketing, salesmen and other product misrepresentatives?"