It’s been a while since I’ve done a fun with graphs post about the Amazon sales rank of How to Teach Physics to Your Dog, but that’s not because I’ve stopped tracking it. It’s getting to be enough data, though, that it’s worthwhile to look on a slightly coarser scale, so here’s the sales rank data binned by day:
This shows some clear structure, specifically two points with dramatic drops (that is, dramatic improvements in the Amazon rank– smaller numbers are better), followed by slow climbs. The two big improvements correspond to the immediate post-publication blog boost, and the publication of the AP review and Time Union feature.
One obvious question you might ask is whether those slow climbs have a consistent form. To check that, here’s a second graph showing the normalized sales rank during those periods:
For this, I divided the sales rank by the minimum value achieved during each drop, and then plotted that as a function of the day after that minimum. That is, both series start at 1 on day 1, and each day after is plotted as a multiple of that initial value.
These are surprisingly consistent, especially considering the very different sources of the two boosts in the ranking. I’m actually kind of amazed at how universal this behavior turns out to be. Fitting a line to it in Excel gives a slope of 0.548/day, meaning that the sales rank doubles after roughly 3.6 days.
This is a pretty common analysis technique in science, which is why it rates a blog post. If you have something that spikes up and then drops off, one way to try to tease out some information about the dynamics of the system is to look at how it evolves between spikes, through this sort of normalization procedure.
I don’t have anything brilliant to add to this, but I thought it was interesting. And it was a fun way to pass the time until my office hours are over, and I can go to the gym.
(I should note that even though the sales rank is dropping, the absolute number of sales is still quite good. My publisher is happy, booksellers are happy, etc. It could be better, of course, but there’s not a great deal I can do to create another such spike. If anybody reading this can get me national media appearances, and drive the rank back up again, please don’t hesitate to contact me…)