The Long Tail of Amazon Gets Shorter:-)

Last week I finally read the now famous article about the "Long Tail" phenomena at many Internet retailers by Chris Anderson, the editor of Wired Magazine. The article and the blog that Chris started for a book he'll be writing on the subject are well worth reading, but his initial assertion that the long tail accounted for over 50% of Amazon sales immediately struck me as wrong, and I've been working on the question from a couple different directions. A warning for authors and self publishers reading this blog. Long tail effects are beneficial to large retailers, such as Amazon, but they do nothing for the bottom line of authors or publishers. By definition, only the retailer makes a living on the long tail.

Several factors complicate the analysis of Amazon sales, but the primary factor is that the data is confidential. What Amazon does make public in its annual reports is the total media sales for North America (Books, DVDs, CDs) and a sales ranking of any product that has sold on Amazon is available on their site. Under the old sales ranking system which was in force through October of 2004, the rank expressed a combination of sales momentum and total sales. As of November, the new system is almost entirely based on recent sales, with the main effect of sales history being to slow the decay rate of the ranking after the first 24 hours without a sale.

My analysis so far has led me to the conclusion that any discussion of Long Tail phenomena is dependent not only on the amount of time you look at, but on where you declare the tail as beginning. In his initial analysis, Chris used an inventory of 130,000 as the "body" for Amazon, and all slower selling titles as the tail. The 130,000 represented a typical bricks and mortar superstore inventory. However, the titles included in that inventory change throughout the year, not to mention from year to year. While I'm working on an all-time historical book sales analysis for Amazon, it wouldn't make sense to contrast an up-to-date superstore inventory with eight or nine years of Amazon sales.

This led me to do a short term analysis based on a single day's sales at Amazon. The conclusion I reached using real time data was that on any given day, the "body" sells 189,000 books, and the long tail 70,000, or about 27% for the long tail, using the 130,000 break point. If we moved the break point a little closer to the 200,000 mark, or just 5% of Amazon's active inventory, the long tail would shrink even further. I'm not sure it's even arguable that Amazon benefits more from Long Tail sales due to the profusion of used books, because used book availability is excellent through Amazon's system.

I'm working obsessively on these questions and hope to post an update later this week. I did just come across a breakdown of Amazon's book sales vs. music and DVD/video, something I may actually have known from five years ago when I was first plotting Amazon ranks. In the 4th Q of 2000, with a couple years of selling both music and video under their belts, the ratio was 2.23:1 in favor of books revenue - for every $2.23 they took in selling books, they took in $1.00 selling music and DVD's.