Woodcut of the execution of King Charles I I’ve been meaning for a while to comment on the Guardian article “A musical tail of hits and misses,” which attempts to address the question of whether the much-discussed “Long Tail” effect actually operates in the real world of digital music services (most notably the iTunes Music Store). Recently David Harrell posted an interesting article “eMusic and the Long Tail” that renewed my interest in this topic and prompted me to get off my rear end and post this.

One of the interesting things about eMusic is how it illustrates how the reality of Long Tail marketing differs from the ideal theory. As presented in Chris Anderson’s original Wired article and in subsequent discussions, the theoretical treatment of the Long Tail makes two simplifying assumptions (whether explicitly or implicitly):

  • The cost of offering items for sale is negligible, and does not vary from item to item.
  • Consumers vary in their tastes (i.e., in which items they might be inclined to purchase), but not in other relevant ways.

Together these assumptions and others lead to the conclusion that in order to maximize revenue and profits retailers should offer as many items as possible, from the most popular to the most obscure, and should find ways to induce consumers of the most popular items to purchase additional less-popular items that might also be of interest (“drive demand down the Long Tail”). As Anderson puts it, in contrast to MP3.com (“only Long Tail”) or Movielink (“Offering only hits is no better”),

The success of Netflix, Amazon, and the commercial music services shows that you need both ends of the curve. Their huge libraries of less-mainstream fare set them apart, but hits still matter in attracting consumers in the first place.

However in the real world the ideal assumptions above do not necessarily hold, and the ways in which they break are directly relevant to eMusic.

First, the cost to offer items for sale is not necessarily negligible and may vary significantly from item to item, perhaps in systematic ways. For example, in the case of eMusic and other services there are both upfront and ongoing administrative costs to offer a particular label’s releases for sale: Label agreements need to be negotiated and re-negotiated as appropriate, each label’s sales need to be tracked over time and licenses paid as needed, and so on. More importantly, the major labels impose a major “cost” in the form of requiring that DRM measures be put in place for their releases.

Together these real-world costs mean that eMusic does not in fact offer as many releases as it could (or, according to Anderson, should), and in effect truncates both the “head” and the “tail” of the full spectrum of releases: At the tail end of the spectrum there are releases that eMusic could in theory license but in fact is not motivated to do so based on the potential benefits in sales vs. the costs of licensing the material and dealing with the labels or artists. As David Pakman noted in an interview with MP3.com:

Together Europe and the US put us at about a million and a half [independent label] tracks. Then you go to the rest of the world and, you know, there are millions more tracks. . . . Are there a few more million tracks out there that we should get over time? Yeah. But they’re not as essential.

And of course at the head end of the spectrum eMusic (unlike most digital music services) avoids the “cost” of DRM simply by not offering major label releases. Again it’s a simple cost/benefit analysis: The cost to eMusic of not having major label releases available for sale is outweighed by the benefit of being able to sell to users of the dominant digital music device, namely Apple’s iPod.

Second, real world consumers differ in ways beyond their taste, ways that are quite relevant to retailers like eMusic. In particular, people who are primarily interested in downloading chart hits are qualitatively very different than people who are primarily interested in indie music and in “non-mainstream” genres like jazz, blues, and classical music. The former group is dominated by teenagers and twenty-somethings who make up the bulk of people downloading music “for free” using P2P networks, while the latter group skews much older and is both more willing and more able to pay for legal music downloads. As David Pakman has commented regarding eMusic customers (in the same MP3.com interview referenced above):

They’re a much more pleasant consumer to deal with because they are far less fickle. They are interested in value but they’re not starving for dollars. They have credit cards that are valid and don’t max out all the time and they can afford to buy both a $400 hardware device and spend, you know, 100 bucks a year on music. So, we like that consumer a lot better.

By forgoing sales of major label hits and thus “chopping off the head” of the overall spectrum of music releases, eMusic has in effect changed the composition of its customer base in a way that is financially beneficial to it, leading to lower rates of credit card fraud, higher subscriber retention rates, higher overall revenue per customer, and so on.

Having said that, eMusic does in fact have its own equivalent of “hits.” The Decemberists, the Gin Blossoms, and other “eMusic Top 10” artists mentioned in the Guardian article are certainly not mainstream acts, but within the context of independent label music they are well known quantities. (I’m sure David Harrell wouldn’t mind at all if the Layaways had a significant fraction of their popularity.) In his post David notes that (depending on genre) roughly half to two thirds of eMusic albums have tracks downloaded in a given month; according to the Guardian article “70% of eMusic’s catalogue sells more than once every quarter.” This is reasonable evidence of the Long Tail at work, but (as David notes) doesn’t come up to the expectations raised by Chris Anderson and other Long Tail evangelists.

I suspect that in many respects eMusic is “hit driven” as well, it’s just that the “hits” are in the context of particular non-mainstream genres. Certainly from my own perspective I disproportionately download albums from eMusic based on their promotion on the eMusic front page, by the columnists for eMusic Magazine, and in the various “eMusic Dozens” lists. By contrast I rarely use the “so-and-so also likes” recommendations on individual album pages, and have never downloaded anything based on what my supposed “eMusic neighbors” are listening to.

Thus I’d conclude that eMusic is far from being an examplar of the Long Tail model, at least in its pure form. Instead it’s something less trendy but (in my opinion) more interesting: a good example of a traditional specialty retailing strategy adapted to the realities of today’s music business and executed well.

A final note: I had some fun finding the image for this post. A special prize (well, not really, but it sounds good) goes to the reader who can identify the connection of this image to alternative music (but not, unfortunately, alternative music available on eMusic).

Avant Music News » Blog Archive » Newsbits: September 15th, 2006 - 2006-09-16 11:33

[…] Swindleeeee!!!!! has an analysis of whether or not eMusic fits into the “Long Tail” media hyptothesis. Kayo Dot has an ipcoming show. One month from today, Birdsongs of the Mesozoic play the Boston area. The NOW Ensemble plays chamber music in New York on September 28th. On October 3rd, Northwestern University will hold an evening of Steve Reich music honoring the composer’s 70th birthday.   [link] […]