Calling all authors!
If you have written a book, you have probably wondered: What’s the connection between amazon rank and number of books sold? Well, wonder no more. Below is a graph based on The Shangri-La Diet. The copies-sold information is from Nielsen BookScan. Their website says:
Most of the nation’s major retailers for books are included in our panel of reporting book outlets: Borders and Walden, Barnes & Noble Inc., Barnes & Noble.com, Deseret Book Company, Hastings, Books-A-Million, Tower Music and Books, Follett College stores, Buy.com and Amazon.com. Weekly sales information is also tracked from Mass merchandisers like Target, Kmart and Costco, along with smaller retail chains and hundreds of general independent bookstores.
The graph shows that the relationship between books sold and amazon rank is linear on a log-log scale (as so many things are — the physicist Per Bak wrote a whole book about such relationships). Each point is a different week. To illustrate the formula of the line,
ln(copies sold/week) = 9.67-0.53*ln(amazon rank),
the amazon rank of Send In The Idiots: Stories From the Other Side of Autism by Kamran Nazeer, a masterpiece about the adult lives of autistic children, is now 35,758. Predicted sales is 61 books/week.
Sweet.
To first order I see how this works. But. The total volume of sales must have tremendous seasonal variation. e.g. I’m flux’d that a rank of 100 in late November would sell the same number as the a rank of 100 in late January. I wonder if that might explain some portion of the variance?
So I predict many of the points below the line are in quiet seasons and those above are in the holiday season.
Coming at it from the other direction, I wonder if you can use your data to discover in what months diet books move. I bet people rarely give diet books as christmas presents
.