Revised Model of eBook Sales
Executive, plain language summary: If you sell a book to everyone in a room on the first day, you won't sell any more until you go outside the room, or the people in the room finish the book, go outside, and talk to their friends. Long-run sales volume should roughly equalize no matter what you do, but short-run sales may be very different based on promotion.
I (along with lots of others) have predicted that because digital books will stay on bookshelves essentially indefinitely, that sales won't go down. It's not hard (including my own reports on the first volume of The Crimson Pact) to find that sales slump off after an initial spike. You can also look at the numbers reported by Jim Hines and Tobias Buckell, who are also experimenting and being very forthcoming with data.
So what gives? Was I wrong?
I've got a significant background in the sciences 1, so I'm okay with my hypothesis being wrong. But I'm not okay without knowing why I'm wrong. Especially in the social sciences, finding out why (or how) you're wrong often leads to the greatest breakthroughs.
One part of the initial hypothesis seem valid. The first volume of The Crimson Pact is still selling, which mirrors the experience of Jim and Tobias. That wouldn't be the case if they'd been rotated off the shelves in a physical bookstore, so that's a point for the original hypothesis.
The next step is to look for confounding variables. There's two that I've been able to identify.
The first confounding variable is in my business plan. Myself, Don Bingle, and Sarah Hans were all at Origins the other weekend. Because of the way I handle CD copies of the book (they're treated as "author copies", and are sold to the authors for the expense of making them), those copies are not counted in the sales numbers. So while I sold several CD copies of the book (mine get counted; I'm the publisher, not an author), any copies that Sarah or Don sold are not reported as sales of the book. While this gives the authors - especially the flash fiction authors in the anthology - a chance to make more money from their work, it means that sales are underreported by an unknown amount. So while I see spikes when I or others mention the book on our websites or blogs, I didn't see as much of a spike from Origins - and I think that's why.
The second confounding variable is that we publicize the book. As I pointed out before, I can tell from sales numbers when The Crimson Pact gets publicized. Likewise, when Tobias mentioned his book on his blog (and when he played with the numbers), sales went up (profit did NOT, which alone makes reading his post worthwhile). Therefore, I believe that we inflated initial demand for a short-run gain, which then fell back to a more organic demand curve.2 While I doubt that I drew the curves quite right (I did it in MS-PAINT, okay?), the area under the green curve should be equal to the area under the red curve. That is, the organic demand curve and the inflated demand curve should sell the same number of books by the time the inflated demand curve equalizes back out.
Of course, demand is not infinitely asymptotic. There is a finite number of people who are able, let alone willing, to read your book. (Of course, this is the INTERNET, so there's a damn lot of them.) Also, I've not attached numbers to the graph. I believe - given the evidence to date, that these curves are an accurate model regardless of sales volume. Now, perhaps for one author that original "spike" is five thousand units and for another it might be five.
As always, this is a marathon, not a sprint.
So while I think I was wrong in the details, that was because of a failure to take all variables into consideration. My original prediction made it seem like sales would never drop regardless of other factors, and that's patently not true.
It is possible to inflate your sales above the natural, organic demand for your work. While this is not necessarily a bad thing - the additional exposure will help - I would expect long-run demand to equalize out.
1While my degrees are in the social sciences, I try to maintain as close to the same rigor as I would for the hard sciences. And yes, economics is a social science, despite all the freaking math.
2I postulate - and this is a wild guess - that demand drops below the organic demand curve but does not fall to zero, and that the system will return to equilibrium. This would largely be due to the effect of blogging, word of mouth, reviews, and so on