The Long Tail: Big Hits and Big Mistakes
The “Long Tail” is a colloquial name given to various statistical distributions characterized by a small group of high-amplitude events and a very large group of low-amplitude events. Coined by Wired magazine writer Chris Anderson in 2004, the long tail of the web has stumped academics and challenged online marketers. Think of Hollywood movies: there are big hits that really make it and thousands of movies that no one knows about. In economics, it is the Pareto principle: 20% of anything produces 80% of the effects. It is these misses without hits that make up the Long Tail. Anderson claims to have discovered a new 98% rule no matter how much content you put online, someone, somewhere will show up to buy it. eBay seems to be a perfect example. The online tag sale contains millions of items pulled from every Aunt Tilly’s closet in the world and still seems to find a buyer somewhere for just about anything.
On the Internet, where storage and distribution costs are next to nothing, Amazon can offer 3 million books for sale compared to a typical large bookstore with 40,000-100,000 titles. The same goes for CDs, DVDs, digital cameras, and portable MP3 players. Everywhere you look on the Web, you’ll find huge inventories and lots of items that few people are interested in buying. But someone is almost always looking for something. With a billion people online, even a one in a million product will find 1000 buyers. According to Anderson, online music sites sell access to 98% of their titles once a quarter. According to Netflix, 60% of its 85,000 titles are rented at least once a day by someone. Unlike physical stores like Wal-Mart and Sears, online merchants have much lower overhead costs because they don’t have physical stores and have lower labor costs. So that can build up in inventory, including items that are rarely sold.
There are several implications of the Long Tail phenomenon for web marketing. Some writers like Anderson claim that the Internet revolutionizes digital content by making even niche products highly profitable, and that the revenues produced by small niche products will eventually exceed the revenues of hit movies, songs, and books. For Hollywood, and all content procedurals, this means less focus on budget-busting blockbusters and more emphasis on steady-based hit titles that have smaller audiences but make up for it in number of titles. The Long Tail is a democratizing phenomenon: even lesser-known movies, songs, and books can now find a market on the Web. There is hope for your blog and garage band! To the economist, Long Tail represents a net gain for social welfare because customers can now find exactly the niche content they really want instead of accepting the “greatest hits” on the shelf. The long tail of the Web makes more customers happy, and the chance to make money off of niche products should encourage more production of “Indy” movies and music.
The problem with all these bugs in Long Tail is that few people can find them because they are, by definition, largely unknown. Thus, in its native state, the revenue value of low-demand products is locked in collective ignorance. This is where recommendation systems come in: they can guide consumers to obscure but wonderful jobs based on the recommendations of others.
In many cases, recommendations are based on the user’s past purchasing behavior, which may or may not reflect the current user’s needs or preferences. However, the ability to narrow down the list of potential options makes the information gathering process more efficient and very useful for many users.