The Prepublication Ratchet
Update 8/25/17: The theory of occasional dumps to Amazon rankings was wrong; instead, individual book sales at these levels of rankings make a real difference (see near bottom); this was more or less confirmed by the “XX left in stock!” admonition as jumps coincided with a drop in XX by 1 (or on one day, drop by 2).
Sorry, this was just too amusing to pass up. GG noted awhile back that his book was listed for preorder at Amazon. Kind of a signpost that this was really going to happen. But the startling thing was that they listed the book’s sales rank, within books as a whole and geology books and history books. This seemed amazing seeing as no books had been sold yet. Curious, GG kept track of this and the picture that emerged was a bit surprising:
The book seemed to slowly decline in popularity until July 3rd, when it leapt up before declining again until, once more, on July 31st. This doesn’t seem to be the case with published books. The first time, GG wondered if one brave soul had preordered the book. But the second time, exactly 4 weeks later? Hmmm…. So GG is guessing that every 4 weeks, Amazon’s system dumps any accumulated preorders into the ranking system, producing an artificial spike. If so, presumably the curve will continue to decay again until August 28th, when it will spike again–but then what happens on Sept. 5th when the book is published?
Yes, GG can be easily amused….
More update 8/5.
So could those jumps be single book orders? Short answer is yes.
Consider the rough relationship between overall ranking and book sales suggested some years ago by Theresa Regan (though I am including a slightly different list posted here). If we plot this up we get these points:
Ignoring the hard turn at the bottom left, which is probably tied to physical limits on numbers of readers, we might be seeing a log-log relation we could extrapolate (May’s original numbers are very linear here). So let’s do that into the realm of Mountains that Remade America territory. This month it is hanging out in the 300,000-1,000,000 area (shaded), which would then range somewhere between 1 and 0.04 books a day, or a book a day to a book every 25 days. If that is really what is going on here, then a single order would indeed seem capable of jumping the ranking by that much.
Put another way, the jump on July 31, if it was one book, would suggest about 0.03 books/day. But add in a book on the 5th of August and we are in the area of a book a week, or 0.15 books/day, which is almost exactly the distance along the top line of the shaded rhombus. Toss in the forward-looking aspect of the book ranks and the failure to jump as high on the 8/5 book makes sense (and also explains why the ranking is higher than 0.15 books/day would suggest on May’s line).
Feel free to test this by preordering the book and letting GG know and maybe we’ll gain insight into how this all works…and GG thinks you can cancel preorders…