Build the path where you walk
A design idea I've always liked is sometimes used at universites and other large campuses.
Rather than build sidewalks where they think the people will walk, they simply wait until people walking around creates many paths from place to place, and build the sidewalks there.
I think this principle has much broader application than sidewalks. With my new office I'm trying to wait on setting up some things until I can tell from experience the best spot (wait..maybe that's just PROcrastination in my case, but not a bad idea.
For internet marketing clearly an experimental approach is called for. Try things out, do more of what works and less of what fails. For organic search rankings his approach is challenged by the fact that Google, Yahoo, and MSN take some time to incorporate changes into the results, but generally experimentalism is a good concept in marketing, and vastly underutilized. Google Blogger Matt Cutts has often noted this in his excellent observations about search strategy.
I've always been amazed how often marketing managers use intuition and even whimsy rather than math to determine advertising buys. With salespeople pressure this leads to a lot of wasted ad dollars.
Online pay per click advertising offers instant feedback which is one of the reason's it's becoming so popular and effective for those who "do the math".