Research to Determine Relative Popularity of Games

I am trying to collect research on how popular certain games are, so it's easier to determine which ones to target through all the noise. The Playtools project will focus primarily on the games the developers are actually interested in, because that's how this works, but it's good for a project to keep its eyes open and examine other popular options to see what problems we might eventually have to solve when we're working on today's APIs.

Round 1: Barnes & Noble in Fresno (ad hoc)

I went to my local B&N, found the RPG section, and just counted all the books. These are counts of individual books, not distinct titles, so I count each copy of Expedition to Undermountain, for example. Here's my data:

Oddly enough, given the mindshare it currently has, I didn't see a single Hero-system publication there. There are several sources of bias here, starting with the fact that B&N will only carry what it thinks will sell in a brick-and-mortar store, and only from publishers it has deals with, and the location of the store, and so on. I consider this a very basic jumping-off point for this kind of research to set the stage.

Round 2: Amazon Numbers

I went to Amazon and looked up the most popular hardcover core rulebook for each of the following game systems. (Feel free to suggest others!) The number is the relative popularity on Amazon; lower is more popular.

Round 3: Borders in Fresno

Similar procedure to B&N, at Borders instead.