Note: start reading this thread on Amazon sales data with Scraping the Gorilla – Picking data off of Amazon
UPDATE: The below data is a bit cleaner than the previous data published earlier. This time I was able to collect all best seller lists and the associated 30,000 book pages in a single 24 hour period (6/19-6/20), eliminating most of the motion blur.
Amazon has around 460 fiction categories on five menu levels. The “Level 1” categories I am considering are:
|Science Fiction & Fantasy|
|Literature & Fiction|
|Mystery, Thriller & Suspense|
|Teen & Young Adult|
Below these are Levels 2, 3, and sometimes 4 and even Level 5 (Amish Romance, for example, which has less than 100 titles). On Amazon all roads lead to Rome so within Literature & Fiction is a category of Genre Fiction which somewhat duplicates the other four Level 1 categories. I will focus initially on the Level 1 genres above.
The first question that came to mind was, “What genre sells the most?” Unfortunately I cannot answer that, other than looking at the top 100 in each genre. That’s a start.
Note: See previous posts for caveats and background on how I arrived at these numbers.
Clearly Science Fiction and Fantasy is not as popular as I would like, given that it is my genre. Still, 22,000 books sold each day on Amazon is respectable. So how crowded is the field? Here are the title counts in each of the four genres. ( Including Literature & Fiction on this particular graph would have been misleading because it duplicates so many genre titles and includes categories like criticism and poetry.)
Ouch. SF&F, the smallest category in sales has a huge catalog. Good news for Teen & Young Adult writers though, as they have a big chunk of the sales and relatively fewer titles. Now I have a better feel for who the genre players are.
The next question that came to mind is “Other than the top 100, what sells?” Again there is no direct way to answer that just from the data I have gathered from Amazon’s website. There is a hint, however, in the pattern of sales among the top 100. If only a few blockbusters were selling, one would expect the top half of the 100 to greatly outsell the bottom half. If on the other hand sales were strong across the catalog, one would expect the bottom half to also sell well. Below is a comparison of estimated unit sales at the top (or near, anyway) and bottom of the list. I have ignored the top 10 because of the uncertainty in estimating sales of top ranked books. So this compares the top and bottom half of the 90 titles between 11 and 100.
Genres with higher dividing lines are weighted toward fewer big sellers, while genres with lower lines show a more even distribution of sales throughout the top 100. Keep in mind that the 50% mark would mean that all 90 sold at the same rate, an impossibility. So the fact that so many come close to that magic number indicates very healthy sales across the 90.
Series vs Standalone
Another question that came to mind is “Do series inevitably sell better than standalone novels?” I have often heard indie authors praise the marketing advantages of series: You make the first book free and, if it’s good, readers will buy the rest. Series are very popular, both in novels and on television. To get at this, I looked at the top 100 lists, crudely estimating which are part of series based on their title. For this purpose I consider books with a theme or continuing character (typically indicated by a parenthetical note in the title) as part of a series. In this case I was looking at money so used royalty estimates instead of unit sales estimates.
I confess I was surprised that Mystery, Thriller & Suspense was dominated by standalone novels, having so often had detective series recommended to me. And I was
horrified disappointed to see that standalone novels fared so poorly in SF.
Is Genre Rank Just Book Rank?
I assumed by default that books within a genre best seller list were ranked based simply on the overall book rank. This assumption is very close to what I am seeing.
There are a few anomalies. One romance book, not pictured because its rank is way off the scale, is ranked #2 overall on Amazon yet is ranked only #46 in its genre. This book turns out to be on pre-order. Amazon apparently allows pre-orders to boost overall book rank, but seems to heavily discount this factor when assigning a genre rank – perhaps discounting those numbers by as much as 90%. While not prefect, this correlation between estimated sales (which is just based on book overall rank) and genre rank is pretty good. I saw no decent correlation when looking at gross sales or royalties. Why isn’t the correlation of ranks perfect? Overall book rank is known to be a moving average, so perhaps genre rank is based on a somewhat different algorithm.