On Bookscan

I was writing a comment on this post at The Beat, and got carried away. So I figured that, since it was long enough to be a post, I might as well make it a post. And so it is:

The problem with Bookscan -- and, particularly, with making any big pronouncements based on Bookscan numbers -- is that Bookscan isn't consistently any percentage of retail sales.

For some books and some publishers, it's around 65%, for others, 75%. In some cases, it could be close to 100%, but it can also be 25% or less (especially anything that gets into Wal*Mart, which sells huge numbers of a few books and which doesn't report to Bookscan).

I work for a publishing line of mostly technical, professional books, and recently did an analysis on one particular product that showed that Bookscan registered about 40% of the sales of that product through the channels that Bookscan covers. (Leaving out all of the other ways those books are sold -- directly by the publisher; through organizations, corporations, or governments; by non-book stores; to college students; and so on.) That's an extreme case...but those cases do exist. And there are probably similar cases in the comics world.

So when someone who can see Fantagraphics's real sales figures says that they don't resemble Bookscan numbers, I believe him. Bookscan is best for parallax; if you know what your books are selling (for real and on Bookscan), and you know what the competition is selling on Bookscan, you can work out, roughly, what the competition is really selling. But without real numbers for comparison in the middle there, Bookscan figures alone are dangerous to rely on.

And the idea that indy comics are failures because they don't sell at the level of long-underwear projects is just silly -- in "real" publishing, five thousand copies of a 23-dollar book isn't bad at all for a literary project. Thrillers sell better, yes -- in comics and outside of them. This is news?