Dani Rodrik tries an experiment:
Princeton University Press ran a small ad for my book last Sunday in the New York Times book review. I was curious if it would have any effect on sales, so I ran a little experiment. I checked the book’s sales ranking in amazon.com at periodic intervals starting on Saturday afternoon.
But the ad didn’t matter so much (see also the comments on the post). I would note a few points of speculation:
1. Below the top tier, a book can rise rapidly through the Amazon rankings without selling so many extra copies.
2. Amazon buyers are better educated and not representative of the market as a whole.
3. It is an open question whether the Amazon rankings are "honest," or strategically designed to sell what is hot at the moment, by making it look especially hot.
4. The best question to ask is: Is your book in Wal-Mart and Costco?
5. The next best metric is to check its location in Barnes and Noble.
6. Success of a book in Borders is less representative of overall success than it used to be; Borders (which is on the verge of going under, I might add) is now closer to an "indie" book store in many ways than it is to B&N.
Addendum: Chug writes in the comments: "display ads for books are not to sell books. they are for good relations between the publisher and the author…."