Tuesday, July 20, 2010

Amazon Algebra

Amazon sent out another press release this morning, filled with glowing phrases attributed to Jeff Bezos and a whole slew of numbers carefully chosen so that they don't actually correlate with each other at all. Their clear aim is to show how their sales are growing immensely -- the underlying tone of every single Amazon press release dating back to 1995, though what, precisely, is growing hugely has changed often -- and that they have some big, impressive numbers to throw around.

So I wanted to try to quote some of those Amazon declarations, and put them into equation form, to show what it is we know, and what we don't know.

The full press release is here, for those of you who want to follow along at home.
the growth rate of Kindle device unit sales has tripled since we lowered the price from $259 to $189,"
So the first derivative of K1 (Kindle unit sales), has increased from x to 3x since June 21st. We do not know what K1 was before or afterward, nor what N1 (unit sales of Barnes & Noble's Nook device) is, nor what the similar sales are on any other e-reading device.

One month ago, Apple announced that they had sold 3 million iPads in three months. One may postulate that K1 has not yet reached this milestone -- because, if so, Amazon would mostly likely have said so -- but there is no solid data to support that assumption. (Amazon has sold four different physical Kindle units -- original, second generation, DX, and second-generation DX -- since November 19th of 2007, but total unit sales have never been released.)
Amazon.com customers now purchase more Kindle books than hardcover books
K2 (Kindle book units) > B2 (hardcover book units), over an unspecified unit of time.

Note that this implies that K2 is less than B1 (total book units), though that's unproven.
Over the past three months, for every 100 hardcover books Amazon.com has sold, it has sold 143 Kindle books.
K2 = 1.43 B2
Over the past month, for every 100 hardcover books Amazon.com has sold, it has sold 180 Kindle books.
No, wait -- K2 = 1.8 B2.

Amazon is clearly not willing to use a moving average to smooth out bumps in their data, and may well have timed this announcement to take advantage of specific bumps. (One possibility: the book business often slows down in the summer, and so Amazon is cherry-picking data to emphasize a recent peak.)

Also, see below -- the "past three months" may be Mar-May.
The Association of American Publishers' latest data reports that e-book sales grew 163 percent in the month of May and 207 percent year-to-date through May. Kindle book sales in May and year-to-date through May exceeded those growth rates.
The first derivative of K2 (not to be confused with the first derivative of K1, mentioned above) is larger than 163% for the month of May and 207% for Jan-May, compared to the same period the previous year.

It is unclear how, if at all, this statistic correlates with the "last three months" data above.
On July 6, Hachette announced that James Patterson had sold 1.14 million e-books to date. Of those, 867,881 were Kindle books.
Kindle is claiming a 76% market share in James Patterson e-books. (My own doubts about whether that Patterson number represents the true ceiling so far of e-book units by a single author still stand, with no further data on either side of the question.)

Amazon, as usual, is throwing around the largest numbers they can mine from their data, salting those numbers with their standard triumphalism, and declaring their obvious victory -- while, at the same time, not actually revealing the most important numbers. (K1, K2, the actual rate of growth of both of them.)

If you're impressed by that, and would like to join the bandwagon now before it leaves you behind, just click on the below banner and you can buy your very own Kindle -- and have your purchase be part of the next round of Amazon Algebra, sometime later this year:


Anonymous said...

My head hurts.

Liane Spicer said...

Another observer with a healthy dose of skepticism. I wrote about Amazon's algebraic shenanigans (though much less mathematically) in a December 2009 post here:

Post a Comment