Boing Boing points to a study done at BYU that shows free eBook versions of book titles correlates to increased short-term sales of the physical version of the title — although not generally in the case of the eBooks released by Tor as part of its Tor.com promotion, in which each title was officially available for one week only. I thought this last part was a little weird, because I know my book sales numbers, and I remember receiving a fairly appreciable bump in my sales after the Tor eBook release of Old Man’s War in February of 2008. So I decided to check out the study, and figured out why their determination was different from my own.
The primary reason seems to be in the study’s methodology. In each case, the study looked at the Bookscan sales of the book eight weeks prior to the release of the free eBook, and then at the Bookscan numbers in the eight weeks after the free eBook release, and noted the difference between the two. This is a generally cromulent strategy, but in my particular case (and I would imagine Brandon Sanderson’s, as his eBook was out a week before mine) there was a factor I suspect should have been noted but was not, and that is that the “eight week prior” window included portions of December 2007, i.e., “the holiday season,” in which week-to-week sales numbers are artificially inflated as people desperately search for material objects to signify their affection/obligation toward other people.
This factor, I suspect, should have been noted, as it would very likely skew the data. I found my likely line in the data table and noted that the raw number drop in sales between the eight week windows was fairly small, and as a percentage was a single digit, which would appear to suggest that absent holiday sales as a factor, the influence of the eBook release might have been larger, and more in line with my own observations regarding the impact of the eBook release on my sale.
I wanted to look at the data another way, so here’s what I did. I threw out the weeks of December 2007, and averaged the weekly Bookscan sales of the mass market paperback of Old Man’s War for the weeks prior to the free eBook release of OMW by Tor. That’s seven weeks of data. I then averaged the sales of the seven weeks after the eBook release, and found on that on average sales were up 2% after the eBook release than prior.
What’s really interesting is what happens when you extend that window a little further, however. For example, in August 2008, Zoe’s Tale came out in hardcover and The Last Colony in paperback, and both boosted sales of the paperbacks of the previous books in the series. So let’s track the OMW paperback sales from the week the eBook came out through the last week of July 2008 (i.e., before those releases would have had an effect). What do we see? An overall 11.6% increase of average weekly sales from the seven weeks of 2008 sales prior to eBook release.
Were there other factors possibly relating to that increase? Sure; for example, in March of 2008 I was nominated for two Hugos, including Best Novel for The Last Colony. That might have had some influence, but I suspect if so it was tangential, since it wasn’t Old Man’s War up for the award. The major significant promotional push of 2008 for Old Man’s War specifically was the free eBook release. My own suspicions are that it was a significant factor in the overall average increase of sales of the book, at least until other releases in the series lifted its sales in their wake.
My point here is not to suggest the study here was done poorly; I think the authors go out of their way to note more and more rigorous data crunching is advisable, and that this is sort of an early swing through the numbers. My point is that when it comes to these free eBooks, quite a lot changes depending on the data set you choose to use to examine them — and this fact is probably one of the reasons everyone’s still in a tizzy about them.