The Long Tail

by Chris Anderson

  Used textbooks are a model of an efficient market—every year millions of students buy and then resell expensive volumes they need only for a single semester. The set of books with resale value is determined by the published curriculum of core classes; the price is set by what competition there is between campus bookstores; and the supply is replenished twice a year.

  Textbook publishers don’t mind this very much because it means they can actually charge more for new copies, since the buyers know they have a predictable resale value. Indeed, the economic model at work here is more like a rent than a purchase. Typically, stores buy books for 50 percent of the cover price and then resell them for 75 percent. Depending on whether the student is buying new or used, that “rental fee” is between half and a quarter of the list price of the book. This arrangement works so well that the used-textbook market in the United States is now a $1.7 billion enterprise, accounting for 16 percent of all college store sales.

  Publishers ensure that the used books don’t circulate forever, which would depress new book sales, by releasing new editions with different page numbers (so the old ones can’t be used). This purges the market of old inventory from time to time.

  In the case of the non-academic used-book market, however, there were few of these efficiencies. The typical used-book store’s access to secondhand books is limited to whomever happens to be local and selling volumes from his or her own collection. As a result, the selection at these stores tends to be pretty random, reflecting the taste of the proprietor and the luck of the catch rather than any comprehensive slice of the book market. For patrons of used-book stores, this randomness is part of the appeal, providing a serendipitous sense of exploration and discovery. But if you’re looking for a particular book, that process of cruising around the store and browsing the shelves can be unrewarding.

  In economic terms, what makes the textbook market work is ample liquidity. There are so many sellers and so many buyers of a relatively small set of traded commodities that the odds of finding what you want at the right price are excellent. By contrast, what ailed the non-academic used-book market was poor liquidity—not enough sellers and buyers of an unbounded set of commodities. The result of too many products and not enough players was that the odds of finding what you want were poor. Thus, most buyers simply never consider a used-book store when they’re shopping for something specific.

  Weatherford had realized that although the economics of each individual bookstore didn’t make a lot of sense, together (with all the bookstores combined or linked up) the overall used-book market made a huge amount of sense. The collective inventory of some 12,000 used-book stores could rival the best library in the world. The individual store owners uploaded their inventory, and Alibris collected them all together and ensured that the used books were displayed right alongside the new ones at the online booksellers that used Alibris data.

  It made that database available to the big online booksellers such as Amazon and bn.com, which integrated the used-book listings alongside new books, effectively making “out of print” obsolete and offering a low-price alternative to new books. By bringing millions of customers to the used-book market, this gave used-book stores even more incentive to computerize their inventories, which, in turn, gave Alibris (and by extension its online retailing partners) even more inventory to sell. It was a classic virtuous circle, and the effect supercharged used-book sales. After years of stagnation, the $2.2 billion market is now growing at double digits, with all that growth coming from a $600 million online market that’s growing by more than 30 percent a year, according to the Book Industry Study Group.

  ENTER THE AGGREGATORS

  Alibris is a Long Tail “aggregator”—a company or service that collects a huge variety of goods and makes them available and easy to find, typically in a single place. What it did by connecting the distributed inventories of thousands of used-book stores was to use information to create a liquid market where there was an illiquid market before. With a critical mass of inventory and customers, it tapped the latent value in the used-book market. And it did it at a tiny fraction of the cost that it would have required to assemble that much inventory from scratch, by outsourcing most of the work of assembling the catalog to the individual booksellers, who type in and submit the product listings themselves.

  That’s the root calculus of the Long Tail: The lower the costs of selling, the more you can sell. As such, aggregators are a manifestation of the second force, democratizing distribution. They all lower the barrier to market entry, allowing more and more things to cross that bar and get out there to find their audience.

  There are literally thousands of other examples, but I’ll give just a few here. Google aggregates the Long Tail of advertising (small-and medium-sized advertisers and publishers that make their money from advertising). Rhapsody and iTunes aggregate the Long Tail of music. Netflix does the same for the Long Tail of movies. EBay aggregates the Long Tail of physical goods and the Long Tail of merchants who sell them, right down to the millions of regular people getting rid of unwanted birthday presents.

  It goes far beyond selling, too. Software, such as Bloglines, that collects “feeds” of online content using the RSS standard are also referred to as “aggregators,” and for good reason—they pull together and coherently order the Long Tail of online content, including millions of blogs. Wikipedia is an aggregator of the Long Tail of knowledge and those who have it. The list of examples goes on and on, aggregating everything from ideas to people.

  In this chapter, I’ll focus on the business aggregators. They fall mostly into five categories:

  Physical goods (e.g., Amazon, eBay)

  Digital goods (e.g., iTunes, iFilm)

  Advertising/services (e.g., Google, Craigslist)

  Information (e.g., Google, Wikipedia)

  Communities/user-created content (e.g., MySpace, Bloglines)

  Each of these categories can range from massive companies to one-person operations. A single blog that collects all the news and information that it can about a topic, let’s say needlework, is an aggregator, as is Yahoo! Some aggregators attempt to straddle an entire category, such as Netflix (films) or iTunes (music), while others simply find their niche, such as services that aggregate only SEC filings or techno music.

  Many aggregators occupy multiple categories. Amazon aggregates both physical goods (from electronics to cookware) and digital goods (from ebooks to downloadable software). Google aggregates information, advertising, and digital goods (Google Video). MySpace, the hugely popular networking site for bands and their fans, aggregates both content (millions of free songs) and the people who listen to it and, in turn, generates more content about those bands in the form of reviews, news, and other fan ephemera.

  HYBRID VERSUS PURE DIGITAL

  Let’s start by contrasting the first category of online aggregator businesses, selling physical goods online, with the second, selling digital goods online. They’re both Long Tail opportunities, but the second one can extend farther down the Tail than the first.

  The online retailers of physical goods, from BestBuy.com’s camera selection to Netflix’s DVD library, can offer inventory hundreds of times greater than their bricks-and-mortar counterparts, but eventually even they hit a limit. By contrast, the companies that sell digital goods, from albums or songs on iTunes to TV shows or amateur clips on Google Video, can theoretically go all the way down the Tail, expanding the variety they offer to encompass everything available. (The other three categories of aggregator—services, user-created content, and communities—are largely based on digital information, so they share this quality.)

  We call the first type a hybrid retailer because it’s a cross between the economics of mail order (physical) and the Internet (digital). In this instance, the goods are usually delivered through the mail or FedEx, and the efficiencies come both in lowering the supply-chain costs with centralized warehouses and being able to offer an unlimited catalog with all the search an
d other informational advantages of a Web site.

  Take Amazon’s CD business. It lists about 1.7 million CD titles. Combined with the collective inventories of its many third-party “marketplace” sellers, the total is probably closer to 2 million. Still, there are limits to that catalog.

  Because CDs are physical items, somebody’s got to store them somewhere before they’re sold. As such, there’s some inventory risk associated with each Amazon listing. After all, a certain CD may never sell. Plus, there are shipping costs associated with each sale, so in practice, the price never falls below $3 or so. And, more important, the songs on a CD cannot be sold individually: it’s either the whole CD or nothing.

  Clearly, Amazon’s CD economics are a whole lot better than the average record store’s, which is why it’s able to offer as much as one hundred times the choice. That takes Amazon well down the Tail. But not all the way. According to SNOCAP, a digital licensing and copyright management service that tracks the usage of peer-to-peer file-trading networks, there are at least 9 million tracks circulating online. That works out to nearly a million albums’ worth—and that doesn’t even include most music from before the age of CDs, much of which will eventually appear in digital form. Plus, there are many thousands of garage bands and bedroom remixers who make music and distribute but have never released a CD at all. Together, all that music could easily amount to another million albums’ worth. So Amazon, despite all of its economic advantages, can actually get only a quarter of the way down the Long Tail of music (which is likely why they launched an MP3 download service in 2007).

  The only way to reach all the way down the Tail—from the biggest hits down to all the garage bands of past and present—is to abandon atoms entirely and base all transactions, from beginning to end, in the world of bits. That’s the structure of the second class of aggregator, the pure digital retailer.

  With the pure digital model, each product is simply a database entry, costing effectively nothing. The distribution costs are simply broadband megabytes, bought in bulk at fast-dropping costs incurred only when the product is ordered. What’s more, pure digital retailers can choose between selling goods as stand-alone products (ninety-nine-cent downloads at iTunes) or as a service (unlimited access music subscriptions at Rhapsody).

  Those commercial digital services have all the advantages of Amazon’s online CD catalog, plus the additional savings of delivering their goods over broadband networks at virtually no cost. This is the way to achieve the holy grail of retail—near-zero marginal costs of manufacturing and distribution. Since an extra database entry and a few megabytes of storage on a server cost effectively nothing, these retailers have no economic reason not to carry everything available. And someday (once they get past messy issues such as rights clearance and contracts) they will.

  Seen this way, there is no simple divide between traditional retailers and Long Tail ones. Instead, it’s a progression from the economics of pure atoms, to a hybrid of bits and atoms, to the ideal domain of pure bits. Digital catalogs of physical goods lower the economics of distribution far enough to get partway down the potential Tail. The rest is left to the even more efficient economics of pure digital distribution. Both are Long Tails, but one is potentially longer than the other.

  TRIPPING DOWN THE TAIL

  Let’s return to the Amazon story to see how this works in practice. Amazon embodies both the hybrid and pure digital models, which have emerged as it sought new ways to lower its costs and work its way farther down the Tail.

  Step one, as we’ve already seen, was Bezos’s original insight: that online commerce could have the basic advantage of a mail-order merchant’s centralized distribution, as well as the direct-buying advantage of a catalog retailer, without the corresponding costs of printing and mailing millions of catalogs. Thus, Amazon 1.0 (circa 1994–96).

  The next step was to reduce the company’s inventory risk even further by not paying for items that it kept even in its own warehouses. Amazon did that with a consignment program. Again, the company started with books. The Amazon Advantage program offered authors what at first blush sounded like a pretty one-sided deal: Pay a yearly $29.95 fee, ship your books to Amazon, and when it sells them, let it keep 55 percent of the proceeds. Why would an author do that? Because the consignment program was one more step away from the delays and uncertainty of special order. In a nutshell, it ensured that the author’s book would be in stock—and easily accessible—without having to beg a publisher to keep it that way.

  The third step toward even lower costs involved extending the virtual inventory model by bringing in other big retailers and their own existing relationships with manufacturers and distributors. Offering its sophisticated e-commerce technology to large retailers such as Toys “R” Us and Target, Amazon created storefronts for those big partners and let them deal with the inventory entirely. With each new partner, Amazon’s effective inventory grew by millions of items.

  Of course, not all big retailers were willing to put their digital future in Amazon’s hands, and those that were often asked to be the exclusive supplier in their domain of housewares or toys. While this limited how far Amazon could extend the model, in principle, being an aggregator-for-hire allowed Amazon to enjoy the sweet economics of a services business, free of the fuss of fulfillment. As eBay can attest, selling your software and servers for a fee is about the highest-margin business around.

  But the big growth in the virtual inventory model turned out not to be in moving up to larger and larger partners, but moving down to smaller ones. In 1999, Amazon introduced its “Marketplace” program, which extended its storefront service model farther into eBay territory by offering its services to all merchants. Retailers and distributors of any size, from specialty shops to individuals, could have their goods listed on Amazon.com just like the products in Amazon’s own warehouses—and the customers could buy either just as easily. By the end of 2004, Amazon had more than 100,000 Marketplace sellers, and these third-party sales represented nearly 40 percent of the company’s total sales volume.

  The rise of this virtual selling model turned the traditional inventory problem on its head. Again, a chain retailer like Best Buy has to distribute its supply of, say, digital cameras across all of its stores, hoping to guess roughly at where the demand will be and how big it might get. Needless to say, the people and the products must be in the same place—supply and demand must meet right there in the store’s aisles. But invariably the retailer will guess wrong, at least to some degree, running out in some stores and having surplus stock depreciating and taking up valuable space in others.

  With the Amazon Marketplace form of distributed inventory, the products are still on shelves around the country, but they’re collectively cataloged and offered in one central place—Amazon’s Web site. Then, when people order them, the products are boxed up and shipped directly to the customer by the small merchants who have held the inventory all along. Like the chain retailers, Amazon also connects centralized supply with scattered demand, but the genius of its model is that the store and customer don’t have to be in the same place. Ironically, this makes it more likely that the supply and demand will actually connect. Regardless, even if they don’t, Amazon bears none of the cost—the surplus stock simply depreciates on the shelves of a third party.

  As this program continues to grow, Amazon gets closer and closer to breaking the tyranny of the shelf entirely. It doesn’t have to guess ahead of time where the demand is going to be, and it doesn’t have to guess at how big that demand will be. All the risk within the Marketplace program is outsourced to a network of small merchants who make their own decisions, based on their own economics, on what to carry. (We’ll get more into the tyranny of the shelf in the “Short Head” chapter.)

  INVENTORY ON DEMAND

  Virtual and distributed inventory is a dramatic way to move down the Tail, but getting rid of physical inventory altogether can take you even farther. Amazon’s next step was to attempt to
get closer to this economic nirvana by building a business that kept inventory as bits until it was shipped.

  One of the problems with carrying books is that a lot of them sell only one or two copies a year. In that case, even orders of 10 copies—instead of 100 or 1,000—might not be viable. Even if it costs a retailer just a dollar to store a book until it sells (which could mean holding on to it for a whole year), the retailer is going to ask itself whether carrying that book is ultimately worth it if it is going to sell in such low numbers. What retailers need is an efficient, economically sustainable way to sell a book that sells just one copy per year. And that means near-zero inventory costs.

  Amazon’s solution was print-on-demand. In its idealized form books stay as digital files until they’re purchased, at which time they’re printed on laser printers and come out looking just like regular paperbacks. Since bits are turned into atoms only when an order comes in, the costs scale perfectly with the revenues. Or, to put it in the simplest terms, the production and inventory cost of a print-on-demand book that is never bought is zero. These economics are potentially so efficient that they may someday make it possible to offer any book ever made. If you’re a bookseller, that means you won’t have to be discriminating about what you do and do not carry in a print-on-demand edition, because the costs of making a mistake are also essentially zero.

  That’s the ideal form. The current reality is that most print-on-demand is used to top off inventory with small print runs of a few hundred. But the falling price of the technology is bringing that number down closer to idealized single-copy form.