Book Sales Studies by Academics and Publishing Industry Groups

I read a lot about book sales, I've subscribed to various industry publications for both publishers and authors, and I never cease to be amazed at the shoddy quality of research that passes for "data." Unfortunately, the work I've seen from academics never appeared to be much better, though it's a lot more difficult to prove that since they use a lot of math I haven't seen since graduate school. My recent discussions with Chris Anderson about the assumptions that went into his now famous Long Tail article have led me to look into a few of the sources he originally relied on.

The source he quotes as the best analysis of Amazon sales to date was published at the M.I.T. Center for Ebusiness, certainly a reputable outfit. However, when I looked for a checksum in their article, all I could find was their estimated sales rate of 99.4 million titles per year being sold by Amazon. To confirm it, they used the estimated 2.5 billion sales for 2001 published by the B.I.S.G, of which 6% was reported as online, and Amazon was given a 70% share. All that math works out to 105 million sales for Amazon, or well within 10%. The problem is that I'm familiar with B.I.S.G. studies and I've never been impressed. As my own checksum, I looked at Amazon's total US media sales for 2001, which includes Books, CD's, DVDs/Video, which was $1.688 billion. Then I adjusted it with their 2000 number for the proportion of non-books in the number, which leaves 1.17 billion for books. This is probably high, since the non-book sales were the faster growth segment at this time. This yields an average Amazon selling price of $11.77 per book. The same M.I.T. study gives the average selling price of a book at Amazon as being between $29 and $41, depending on the popularity. Somehow, this checksum is off by a factor of 3.

I don't claim to have the monopoly of insight into Amazon, all of the analysis I've done over the years is based on estimates and hand drawn curves. The one area where I do feel I have all of the academic observers beat silly is patience. I've been visiting daily since the late 90's, and as many as a dozen times a day when I'm watching sales ranks. I've observed peculiarities in their system that could only be detected by tracking a title for years, like the hysteresis effect in their old ranking system that I'm sure was unintentional. I don't dress my results up in Greek notation and footnotes, I just do my best to call it like I see it. Maybe the fact that I have no thesis to prove helps me keep an open mind, but the whole point of this rant was to say, don't believe what you read about book sales just because it's in the New York Times or The Economist. They can check facts from morning to night, but in the end, they're only as good as their sources.