Tracks of My Tears: Reconstructing Digital Music

At the dawn of the digital music era, record labels went along with a pricing scheme devised by Apple that they are still paying for today. The idea to "unbundle" albums into separate tracks sold for 99 cents each suddenly allowed consumers to bypass higher profit-margin albums.

Sure, you could still purchase an entire album. But consumers found more value in cherry-picking favorite tunes for much less money. Fans of Strawberry Alarm Clock, for instance, could buy the 1960s hit Incense and Peppermint while easily avoiding ghastly tracks such as Sit With the Guru and Rainy Day Mushroom Pillow.

So despite selling record numbers of individual songs on online services such as Apple's iTunes, the labels are in an era of declining revenues and consolidation. What happens next?

"When consumers start buying music online, they switch from buying full albums to cherry-picking their favorite songs."

Harvard Business School professor Anita Elberse, who does much of her business research on the entertainment industry, looked at the clash between bundles and digital distribution, and the effect on media and entertainment firms. We asked Elberse about her recent working paper, "Bye Bye Bundles: The Unbundling of Music in Digital Channels."

Sean Silverthorne: First, what gave you the idea for this study? How did you do the research?
Anita Elberse: As with most of my research, the idea emerged after a conversation with an executive in the media and entertainment industry. Many products in those sectors are sold in a bundled form, so bundling strategies have always been a key topic for media and entertainment firms. But I learned that the rise of digital channels is introducing new questions. Executives are wondering whether and how to bundle products offered online, or even whether to sell their products via those channels at all if online retailers impose certain bundling policies.

I thought the music industry would be an ideal research setting because it is a sector strongly associated with bundled products—music was traditionally sold mostly in the form of albums—and because it is among the sectors most strongly affected by digital technology. So I reached out to Nielsen SoundScan, the company that tracks recorded-music sales in North America, to obtain data for a random sample of over 200 artists for the period January 2005 to April 2007. I analyzed those data using an econometric model that relates the growth in online music buying to the revenues per bundle.

Q: Could we have real-world definitions of what you mean by bundle, pure bundle, mixed bundle, and unbundling?
A: Sure. A "bundle" is any set of products sold together. Think of songs on a music album, television episodes on a DVD, or chapters of a book. "Pure" and "mixed" refer to the condition under which those products are sold. Under a pure-bundling strategy, a firm sells only the bundle, while under a mixed-bundling strategy, a firm sells both the bundle and (all) the products separately. "Unbundling" refers to products that were previously sold as pure bundles being sold as separate items.

Take rock band U2's latest album, No Line on the Horizon, one of this year's top-selling albums. People can buy the full album with its 11 songs as a bundle, but they can also download one or more songs separately through online services like Apple's iTunes. Before the emergence of the online channels, it was not economically feasible to sell all the songs on an album separately—just imagine labels having to print and distribute CDs with numerous subsets of the songs on the album—but when the process is fully digital the costs of reproducing music are much lower. The Internet makes mixed bundling feasible.

Q: You note that while demand for individual songs on services such as iTunes is growing, record label revenue is shrinking. What's happening here?
A: The core idea is pretty simple. My research shows that when consumers start buying music online, they switch from buying full albums to cherry-picking their favorite songs on those albums. On average, each album no longer bought is "traded in" for one, perhaps two, individual songs. And because song prices are relatively low—labels typically have to sell 8 to 10 digital songs to generate the same kind of revenues as they do with one digital album—this causes a sharp reduction in revenues over time. In fact, I estimate that, over the course of the study period, a drop of around one-third of the total weekly sales across the album and its associated songs is directly attributable to people switching to buy music online.

And it might be helpful to point out that my model controls for any trends in illegal downloading, and for other possible changes in marketing strategies (such as, say, the labels placing more emphasis on less popular music genres over time), that could also put a downward pressure on music revenues. This makes the estimated drop in revenues all the more dramatic.

Q: Are these forces affecting every artist and album equally?
A: No, that is an important point. There can be sharp differences. My study was primarily aimed at uncovering what factors may affect the magnitude of the impact. For instance, I expected my results to show that albums with a larger number of songs would be more insulated from the drop in revenues—I figured that if all songs are priced equally, those albums are a particularly "good deal" for consumers. But that is not what I found: The number of songs on an album does not really matter.

"The labels did themselves a disservice by granting a player like Apple such power in the channel."
Instead, the findings suggest that music buyers evaluate bundles in other ways. Consumers respond more favorably to a bundle if its items are more consistent in their appeal. That is, bundles that are highly uneven in how popularity is distributed across individual components see an even greater decrease in revenues over time. Perhaps this sounds intuitive, but it doesn't necessarily correspond with the popular belief that one or two popular songs can "make" an album. My findings suggest that as music consumption moves online, labels are less and less likely to get away with selling a bundle based on the strength of one or two components if the other items are far less appealing. The sharper the differences in appeal, the more consumers will know exactly where "to draw the line" in deciding which subset of items to buy, and thus forgo buying the album.

My study also highlights the ongoing role of brands: I find that a strong artist reputation helps to curb the negative impact of unbundling. Consumers are more likely to buy full albums from established bands like U2 with a strong track record of success.

Q: So what should record labels do to fix things?
A: The labels could simply refuse to offer their goods in an unbundled form online by avoiding retailers like iTunes that, with few exceptions, require that songs be made available separately. The band AC/DC has followed that strategy for years, and some insiders attribute its high album sales to that choice. However, it is difficult to see how this strategy would affect less-established artists, and I think the long-run effects are difficult to predict.

A better approach might be to continue to push for higher prices online and generally more flexibility in setting prices. The key for labels is to capture a high-enough markup on individual songs to make up for any lost revenues on albums. Another strategy worth considering is to sequentially release albums and songs so as to stimulate more loyal and eager consumers to buy the full bundle. Offering extras to consumers buying the full bundle may help, too.

More broadly, I think labels should rethink the essence of a bundle. An album with around 12 songs may be a fine format for some artists, but why would it necessarily fit the majority of musicians? Digital channels give labels great flexibility to try alternative formats. My results show that giving preference to quality over quantity and designing smaller, more consistent bundles may be beneficial.

In general terms, the same probably applies to other industries where digital channels could lead to an unbundling of products, such as book or newspaper publishing and television production. It should be about providing real value to consumers, not about tricking them into buying something they do not want.

Q: Apple appears to be making some changes in the iTunes store that could help labels. First, it has introduced three-tiered pricing, which allows higher prices for popular songs. Second, the iTunes LP format allows music companies to sell albums with bonus material such as photos. And the new Michael Jackson album will be available only in bundled format. Signs of things to come?
A: I think so, yes. The labels waited a relatively long time for more room to price songs as they see fit on iTunes, and they certainly have embraced the ability to price songs higher than $0.99. In a typical week, the lion's share of the iTunes Top 100 songs are priced at $1.29. The 30 percent higher price does not seem to prevent those songs from becoming popular among consumers.
And the Michael Jackson pricing strategy is another example of labels trying to protect themselves from the losses associated with unbundling, and rethinking what a bundle is. I think we will see a lot of experiments with different formats.

Q: Earlier this decade, Apple persuaded many record companies to agree to a pricing model ($0.99 per tune) that held for years. In retrospect, do you think the music companies missed the boat to rethink their digital music business at the very start? Are there lessons here for other industries faced with the same threat or opportunity?
A: I understand the reasoning, but I find that hard to say, even with the benefit of hindsight. Song prices certainly seem low, especially when compared with album prices. But we should not forget that labels were battling the threat of piracy, and before Apple came along it wasn't at all certain whether this new form of selling music online was going to take off. Had Apple opted for significantly higher prices for songs, more people might have stuck with or turned to pirated products. There is something to be said for letting people experience the advantages of a new channel for buying music first and then, when they have come to appreciate the value those channels offer, gradually raising prices to bring them more in line with the value created.

But I do feel the labels did themselves a disservice by granting a player like Apple such power in the channel. Apple now accounts for over 90 percent of music sold digitally, and the company isn't even primarily in the business of selling music—it mostly seeks to maximize sales of hardware like iPods and iPhones. I think that having such a player dictate pricing strategies is an undesirable situation for the record labels.

In fact, this may be the most important lesson for other content producers. They should consider which intermediaries they let into the channel and under which terms, or better yet, aim to be that intermediary themselves so as to maintain control over pricing and other marketing strategies. Initiatives like the Web video aggregation service Hulu, a joint venture between three television broadcasters, perhaps signal that other industries have learned from the music industry.

Q: What are you working on now?
A: I am working on a number of research projects that look at the impact of digital technology on strategic marketing, including a project on the pricing of digital songs, a study on the reach and effectiveness of online video advertisements for video games and movies, and a project on how online retailers of entertainment goods can best manage their "long-tail" assortments. I also just completed a case study on Hulu and its role in the television industry.

I've always been intrigued by the media and entertainment industry, but with the advances in digital technology it is turning out to be a particularly fruitful area for research!

This article was provided with permission from Harvard Business School Working Knowledge.