Lee Gomes’s Wall Street Journal column today discusses Chris Anderson’s new book The Long Tail. I should say up front I liked the book: it makes a serious point but remains quite accessible. Perfect beach reading. Gomes’s column suggests that he thinks that Anderson overclaims in the book; Anderson has a reply on his blog. And my former seminar-mate Tim Wu has a review at Slate. We should try to understand the core idea of the book and then consider the dispute.
Anderson’s focus is the unbundling of the shelf. Walk into a store and stare at a shelf. The shelf is both display—the way the products are visually presented to the consumer—and the fulfillment mechanism—you reach and pull the product off the shelf. Online stores unbundle the shelf: you see the product on Amazon’s web site and someone in a warehouse pulls the product off the shelf if you click and buy it.
Of course, catalog merchants have always done exactly the same thing, as Anderson points out in his discussion of the origins of Sears. The difference is that catalog display space is expensive relative to the price of online display space. Paper is more expensive than bits. For Jeff Bezos that meant the difference between a difficult business—selling books via catalog—and Amazon as we know it today, an online seller with an infinite amount of virtual display space.
But what would we display in this infinite space? That gets us to the second point, and one that Sears and Roebuck undoubtedly understood. This is the law of large numbers applied to inventory management and how that matters for product diversity. In the fringe—in the tail—local demand is highly variable. A customer who walks into Border’s or Tower Records will find what she wants if she is an average consumer, but if she has tastes outside the mainstream, she may walk out empty-handed. Local stores only have so many spots on the shelves and they are going to hit the big sellers.
An Amazon will stock a bigger inventory because it will cost them less to do so. Cost less in a physical sense—remote warehouses versus city rental rates—but at least as important, cost less because Amazon aggregates demand—it sums up idiosyncratic demand across the country—so it sells unique books all of the time. They don’t sit on its shelves gathering dust. The long tail.
So now the disagreement between Anderson and Gomes. The key issue is the size of the tail the relative to the head. The head is dominated by big sellers, while the tail is the domain of many many many more small sellers. Does the tail sum up to more than the head? Gomes thinks not and suggests that Anderson thinks otherwise; Anderson disputes that in his reply.
Movies are one of the interesting cases. Anderson notes that only about 100 films each year get into theaters in a big way, but that roughly 13,000 are shown in festivals each year in the U.S. Walk into a video store and you are likely to find 2000-3000 titles. Those represent that Theater 100 year-by-year, with a few others thrown in. Again, expensive local shelf space tracking local demand.
But switch to Netflix, a subscription service that rents DVDs by mail. We are on the three-at-a-time plan at my house, which means that for a monthly fee—under $20—we have out three movies from Netflix. When we finish one, we send it back via mail, and Netflix sends us the next one in our queue. How many titles are available at Netflix? 55,000. For many of the “extra” 52,000 movies, this is really the only practical way to see them.
But are we longtailing and if so why? The Netflix business model matters for this. Netflix has lowered the cost of sampling movies. Netflix does lump-sum pricing, that is, a fixed fee per month. There is no marginal cost for sampling films, only, as an economist would put it, an opportunity cost: if you have three German films out, you can’t have out three blockbusters. So you can sample easily, decide quickly that the German film really isn’t for you, and send it back. With a couple day turn around, this has cost you very little. Lower sampling costs should result in greater variety in consumption.
The Netflix model matters in a second way and this pushes towards the tail as well. Netflix is copy poor and title rich. Netflix manages its inventory by “underbuying” blockbusters. You manage your account at Netflix through a queue and you can see the waiting time for a film in the queue. There are often waiting times to get new blockbusters, and so if you wanted just those movies, you would sit around and waste your subscription. Netflix’s copy-poor/title-rich inventory strategy nudges you into the tail. It will be quite interesting to see what happens to tail viewing patterns when Netflix really becomes Netflix and switches from DVD distribution to online distribution directly into your home. Presumably we will walk away from copy scarcity and that may push us back from the tail towards the head.