The Long Tail (12 page)

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 carry
ing 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.

Amazon started by placing industrial printers in its own warehouses. Then, in mid-2005, the company massively expanded its capacity by acquiring BookSurge, a leading print-on-demand business. A few months later, it did the same for movies, buying CustomFlix, a DVD-on-demand company. Now Amazon can retain an inventory that takes up no space and has no cost at all: These books and movies remain files in a database somewhere until they’re ordered.

Of course, Amazon didn’t invent the notion of print-on-demand. It has long been a dream of the book industry, but until recently, print-on-demand was hobbled by technical and economic constraints. Printing a paperback that looks good is not, surprisingly, the problem. Unless you know what to look for (mostly the reproduction of images on inside pages), you probably can’t tell if that paperback that just arrived from Amazon was printed in a batch of 50,000 by the publisher or in a lot of one by a laser printer in one of Amazon’s warehouses.

Despite the compelling economics, the publishing industry is still far from a widespread shift to print-on-demand. Traditional printing is
considerably cheaper for large batches. It is still costly to turn a book manuscript into a file that’s formatted correctly for print-on-demand. As I write, print-on-demand is also limited to a few set paper sizes, meaning that books whose pages are larger or smaller than certain dimensions absolutely must be redesigned and reformatted. Then, there is the gnarly question of rights. For older books, the author’s permission is required to make a book available as a print-on-demand edition. Yet many authors are afraid that the price premium now required of on-demand printing (a few dollars to cover the slightly higher production costs compared to bulk printing) will suppress sales; so they resist.

But the potential of print-on-demand is extraordinary, and not just for the onesies and twosies. The biggest cost to publishers is the cost of returns from booksellers, which the publishers freely accept as a matter of industry practice. The reason booksellers over-order is that they want to make sure they don’t run out between print runs, and since the cost of any excess is borne by the publisher, there’s little risk in ordering a bit more than they might need. But if the booksellers knew that demand could be filled via small print-on-demand batches between big print runs, they might be willing to order no more than they actually need, potentially reducing returns radically.

Thus the economic efficiencies of print-on-demand wouldn’t just be extending the Long Tail, but also improving the economics of the head, where there are far more dollars at stake. This is, needless to say, a powerful attraction and will only accelerate the adoption of the technology.

THE END OF INVENTORY ALTOGETHER

The ultimate cost reduction is eliminating atoms entirely and dealing only in bits. Pure digital aggregators store their inventory on hard drives and deliver it via broadband pipes. The marginal cost of manufacturing, shelving, and distribution is close to zero, and royalties are paid only when the goods are sold. It’s the ultimate on-demand market: Because the goods are digital, they can be cloned and delivered as many times as needed, from zero to billions. A best-seller and a never-
seller are just two entries in a database; equal in the eyes of technology and the economics of storage.

Today this is the model that iTunes, Rhapsody, and the other digital music services are so dramatically demonstrating. But the opportunity goes much farther than just music. The overwhelming trend of our age is to take products that were once delivered as physical goods, find ways to turn them into data, and stream them into your home.

For video, pure digital markets range from commercial video-on-demand services provided by cable companies to Web-based video aggregators such as Google Video. Peer-to-peer file trading technologies such as BitTorrent are the underpinnings of hundreds of noncommercial digital video markets, while iTunes is building a thriving pay-per-download video business for its video iPod. Some of this is television content, making these network-based digital video markets a sort of TiVo in the sky. Other aggregators offer movies, a market that will someday take Netflix’s massive selection and make it all instantaneously available (a move that will presumably be led by Netflix itself).

Video games, once delivered on cartridges and then on DVDs, are now increasingly streamed as bits to game consoles in the living room. This creates a new market in everything from older titles and niche titles to supplemental content such as new characters and levels. Nintendo is putting this at the core of its next console, code-named Revolution, which will be backward compatible with its previous consoles, making most of its back catalog available as Long Tail content—fun or nostalgia downloadable and playable for a small fee.

And so, too, for ebooks and audio books, online newspapers and magazines, and software. All were once delivered on paper or plastic, necessitating all the complexities of physical inventory and delivery. All are now joined by digital versions, with corresponding digital economics. The experience is not always the same, which is why paper books and magazines are still the preferred version for many. But the functional gap is shrinking. And the distribution advantages of the digital versions are irresistible.

Once upon a
time, there was really only one way to launch a hit album: radio. Nothing else reached as many people, as often. Getting on a radio playlist was tricky (especially after payola was outlawed), but once a song was in heavy rotation it had a high probability of selling. Then, in the 1980s, came MTV, which became the second way to create a hit. It had even more limited capacity for new music, but its influence over a generation was unparalleled. For the music labels, those were good times. It was a brutally competitive business, but it was a business they knew. They understood the rules, and they could earn their keep by working them.

But now rock radio is in seemingly terminal decline and MTV doesn’t show many music videos anymore. So how to market music? Labels know the answer lies online, tapping the word-of-mouth forces that are replacing traditional marketing in creating demand, but they’re still trying to figure out exactly how best to do it.

We’re entering an era of radical change for marketers. Faith in advertising and the institutions that pay for it is waning, while faith in individuals is on the rise. Peers trust peers. Top-down messaging is losing traction, while bottom-up buzz is gaining power. Dell spends
hundreds of millions each year on promoting its quality and customer service, but if you Google “dell hell” you’ll get 42,000 pages of results. Even the word “dell” returns customer complaints by the second page of results. The same inversion of power is now changing the marketing game for everything from individual products to people. The collective now controls the message.

For a generation of customers used to doing their buying research via search engine, a company’s brand is not what the company says it is, but what Google says it is. The new tastemakers are us. Word of mouth is now a public conversation, carried in blog comments and customer reviews, exhaustively collated and measured. The ants have megaphones.

The question of how to drive demand in such a world is a key one, and in this chapter I’ll describe many of the techniques that work best. But first, I’ll start with the music industry, ground zero of the Long Tail explosion. Three bands tell the story of an era where the power has shifted from music executives to fans, to the consternation of suits everywhere. The results are mixed—one is a disappointment, another a success, and the third a sobering lesson in how bands may soon not need labels at all—but together they illuminate the challenges of selling in a new era of empowered consumers.

BONNIE MCKEE

In September 2004, the record label Reprise (a subsidiary of Warner) released the debut album by a then-nineteen-year-old singer named Bonnie McKee. It was a rocky start. The record had been recorded twice and delayed a year while the label tried to figure out what to do with it—and her. Although young, McKee had a mature, throaty voice, wrote her own songs, and had had a troubled adolescence that involved drugs and sexual experimentation. She had married at eighteen but openly dated other men, sometimes those twice her age. Her hero was the delightfully unhinged Fiona Apple, another artist whom record labels have had trouble categorizing.

Based on her hard-luck story and rough edges, Reprise eventually
decided McKee fit into the singer-songwriter rock category that included the likes of Sheryl Crow. They titled her album
Trouble
and began a marketing plan that would pitch her to so-called adult contemporary radio stations, which appeal mostly to women in their late twenties and early thirties.

Such guesswork is risky—even the labels can’t predict whether and where an artist will resonate—yet for new acts without a touring history there have been few alternatives. But today radio is no longer the only way to launch new artists. So while it was preparing its radio rollout, Reprise prereleased several tracks to online music sites, including Yahoo!, which has a free Internet radio service called LAUNCHcast. One of the most popular features of LAUNCHcast is its customized radio station, which allows its millions of users to select bands and genres they like and then listen to those bands and others like them for free. Reprise decided to see if this audience could help them find out where McKee fit in.

LAUNCHcast is built around an “adaptive” recommendation system that decides based on your preferences what else you might like. While each song is played, a little window display encourages you to rate the song, artist, and album on a scale of one to five stars, from “Never play again” to “Can’t get enough.” As you listen to music and rate it, Yahoo!’s software is getting to know you and changing the playlist of upcoming songs accordingly.

But it’s not just software. LAUNCHcast is also learning from other listeners and using their opinions to guide its recommendations. Because this is an online service with millions of users, Yahoo! is able to record hundreds of millions of likes and dislikes each year, measuring the taste of its listeners with remarkable precision. This tells it something not only about each of its users, and how to provide them with more music that they’ll like, but also about the music itself. LAUNCHcast, along with being a free music service, is a polling machine of remarkable size and fine-grained resolution. It is, in a sense, constantly taking the pulse of the culture, learning how artists fit into it through the clicks of millions of music fans.

If enough people say they like Groove Armada as well as The Crystal Method, there may well be a stylistic connection between them,
despite the fact that one’s categorized as “downtempo” and the other “beats and breaks.” Such strong associations tell Yahoo! to put the two on the same playlists more often, and if the positive ratings continue to come in, that connection is reinforced.

As Yahoo!’s software makes custom playlists for each listener, it occasionally sprinkles in a few new artists and tracks to see if they resonate. Radio stations do this, too, but typically only with artists who have a good track record, and even then only after much pretesting and record-label marketing. The difference is that Yahoo! has literally millions of radio stations, each one of them a stream customized for a user. It effectively has infinite broadcast capacity, and thus, just as with infinite shelf space, it can afford to be a lot less discriminating. So it can try to break more new artists and albums—thousands of new songs each year, almost all of which will get no airplay on traditional radio.

If a new song gets high ratings from the few listeners who first hear it, Yahoo! will add it to more playlists. Unlike a traditional radio station, Yahoo! knows quite a bit about those listeners who liked the song. It knows their gender, age, zip code, and a lot about their musical taste from having tracked their listening behavior and ratings. These data streams, used cleverly, can unlock a powerful new way of marketing music—word of mouth amplified by the feedback effect of adaptive recommendations.

This is what drew Reprise to the service. Unsure of where to find an audience for the talented McKee, Reprise decided to use Yahoo!’s ability to test new artists by pushing her first single, “Somebody,” to adult contemporary playlists, which were similar to the listenership of the radio stations they intended to market her to. The label paid for extra placement and promotion to push McKee out to more listeners, hoping that the ratings feedback would support their instincts about her natural audience. And after a few weeks, Yahoo! did indeed have its answer. “Somebody” was very popular, but not equally with all demographic groups—and not, surprisingly, with the 25–35 female group the label had aimed it at.

The report from LAUNCHcast showed the following demographic information about McKee’s listeners:

AUDIENCE COMPOSITION

The lesson was clear. Reprise had guessed wrong. Instead of appealing to women in their twenties and thirties, Bonnie actually appealed to a far younger audience, with nearly half of her listener base under the age of seventeen. Instead of showing an affinity to artists like Sheryl Crow, this listener constituency most commonly searched for artists like Avril Lavigne, Britney Spears, and Gwen Stefani. It turned out that many teenage girls could relate to the troubled adolescence and bruised romance story in McKee’s lyrics.

By the middle of November 2004, “Somebody” had become the tenth most played song on LAUNCHcast. Finally, as a result of the promotional campaign, Bonnie McKee became a Top 50 search term on the service.

This data prompted label executives to make a major change in how they marketed Bonnie McKee. They gave her a makeover, emphasizing her edgier side, a sort of bubblegum Lolita-gone-wrong look. She was neither a Sheryl Crow nor a Britney Spears, they decided; she was the rebel anti-poptart, appealing to an angsty subset of the teen girl audience.

It was a smart move, but it didn’t work. Her album sold fewer than 17,000 copies. Despite demographic and geographic data of where McKee’s most receptive audience could be found, she still got virtually no airplay. “What we’ve learned is that if a band builds an online fan base first, they have a better chance of selling CDs when the song gets on the radio or MTV,” says Robin Bechtel, who ran the marketing campaign. “Many artists who don’t do that either fail at radio or get on
the radio and only the hit song gets downloaded, rather than people buying the whole album. It seems the fans aren’t invested in the artist, just the song.”

She speculates that demand for McKee’s single was pretty much satisfied by all the free online access. Her appeal was apparently not deep enough to get people to go beyond the single they’d already heard online. The problem wasn’t positioning or marketing, it was a lack of authentic grassroots support. Getting online consumers to pay for music today takes more than a catchy single; it requires a real fan base, ideally one spreading the word online.

MY CHEMICAL ROMANCE

Reprise found a perfect example of just that kind of fan base with a punk-pop fivesome from New Jersey called My Chemical Romance. Although the band’s album
Three Cheers for Sweet Revenge
came out around the same time as McKee’s, it was their second album. The first, on an independent label, had sold 10,000 copies, which suggested a small but strong core following. So five months before the second album’s launch in May 2004, Reprise started giving tracks to Web sites focused on that core, such as Shoutweb.com and AbsolutePunk.net, to get the buzz going among the faithful in hopes that it would spread.

The label also pushed the band on PureVolume.com and MySpace.com, two relatively new (at the time) music-heavy social-networking sites with an exploding user base. It gave exclusive live tracks to PureVolume for promotions and premiered an Internet-only video for the band’s first single, “I’m Not Okay (I Promise).”

Once the tracks were out there, Reprise could watch how they did. Using BigChampagne file-trading data, the label could see growing interest in “Not Okay,” but also heavy trading and searching on the track “Helena.” On the basis of that, it made “Helena” the next single, and, helped by requests from the band’s core fans, the song got airplay. By the end of the summer, “Helena” had become the band’s biggest radio single by far.

As the band went on tour in September, Reprise extended the promotions to Yahoo! Music and AOL, including audio, video, and a heavily promoted live performance from Yahoo!’s studios. Meanwhile, fans flocked to the band’s Web site and MySpace page. My Chemical Romance now has Warner’s largest email list.

The album went on to sell 1.4 million units, making it one of the biggest hits of the year. Most of that came after radio and MTV embraced the band and brought it to a larger audience, but it all started online, where the band’s core audience had cemented its credibility.

What was the difference between My Chemical Romance and Bonnie McKee? Talent differences aside, My Chemical Romance had the advantage of an existing base of fans, both of its first album and its live shows. There were thousands of people already hungry for more from the band, and when the label gave them what they wanted, in the form of early online content, they returned the favor with strong word of mouth, including radio requests. And that, in turn, got the band the airplay that took it to the next level of popularity, acquiring a new, larger, set of fans.

McKee, by contrast, was a relatively unknown artist, who had rarely played live. Although people liked what they heard on Yahoo!, it wasn’t enough to trigger real fan behavior. They didn’t buy the album, and they didn’t clamor for more. On MySpace today, My Chemical Romance has more than 1 million “friends”; McKee has 12,000. Word of mouth makes all the difference.

BIRDMONSTER

This last example is a much smaller one, but one I know well, since it involves a former colleague. In the course of researching this book, I decided to track the progress of Birdmonster, an up-and-coming San Francisco band fronted by Peter Arcuni, an editorial assistant at
Wired
. The experience proved all too instructive.

Birdmonster is a prime example of how the three forces of the Long Tail are overturning the status quo in the music industry. Like all new rock bands, Birdmonster started by hustling for gigs. But rather than
pestering club owners for a break, the band members realized that there was now a smarter way. In club booking, the headliners are typically signed up first. Then, once the dates are set in the calendar, the club looks for opening acts to support them. Since virtually all club schedules are now online, opportunities for opening acts can be found simply by searching for the letters “TBA” and some other keywords to limit the search to local clubs. Then it’s simply a matter of contacting the club and offering to fill that gap in their lineup.