A timely post from Mark Coker, the founder of Smashwords. Mark broke down a 100,000-strong sample of January's eBook downloads from Smashwords by popularity. The results showed PDF to be the most popular format at 35%. Quite aside from the fact that I've been questioning the wisdom of offering my short stories exclusively in the format, what's interesting here is that a sample of half that size around twelve months earlier had PDF at number two with just 19%.
The format that PDF is playing toe to toe with is EPUB, an open format specifically created for eBook readers and the official standard of the International Digital Publishing Forum. As such most eBook readers (including the iPad) will accept the format with the Kindle as a notable holdout.
The Kindle prefers the Mobipocket format, probably not a problem in itself except maybe for the punitive DRM lathered over purchased files.
Despite the impressive sample sizes, I'm not convinced at how representative the statistics are. When I see counter-intuitive numbers, I wonder about what kind of sample has been taken. Are the randomly selected eBooks of similar length and content? A loading up of shorter works or non-fiction works (think how-to titles) might well be more popular in PDF as they are quickly scanned or read on a standard computer and forgotten about. Depending how many such titles make it into your sample will depend how much skew they apply to the stats, one way or another. It's possible the vacillating stats are an artefact of the Wild West of digital publishing—the audience is changing rapidly as is their preferred format for reading—but I doubt it.
What I'd like to see is far more targeted numbers, broken down by the length of the work or its content or whether it is being offered by a publisher or directly by the author. As interesting as Mark's breakdown is, I suspect there's little of use we can take away from what we have so far.