In my previous post I discussed the difficulties that independent labels were having in adapting to the new landscape of the music business, and whether eMusic management was being sufficiently sensitive to that fact in their dealings with labels. Beyond the emotional aspects of the indie labels’ issues with eMusic and the music business in general, there are also some serious questions as to whether the current eMusic model is an overall plus or minus for labels. There seem to be three related but distinct concerns here:

  • Selling through eMusic is not profitable, period.
  • Selling through eMusic isn’t profitable enough.
  • Selling through eMusic lowers overall profitability by diverting sales away from the iTunes Store and other higher-priced outlets.

The first concern is that dealing with eMusic is simply unprofitable; for example, on the eMusic message boards Wesley cited this as a reason for Red House Records leaving: “According to a contact at Red House they felt it was time to leave because after paying the artist and songwriter royalities they were losing money on every track.”

As frogkopf has noted on the eMusic message boards, this problem is primarily due to traditional music industry royalty arrangements, in particular the practice of paying fixed mechanical royalties (X cents per track):

For along time, having a fixed rate for mechanical royalties worked in favor of the record labels, since it was inflation proof - as retail prices climbed the rate stayed the same, resulting in the payouts being a smaller and smaller percentage of the overall sales. Now that the price per unit is dropping under new pricing models such as eMu’s the labels are on the losing end of the equation, and they don’t like it a bit.

I see two things happening as a result of this trend. First, many individual artists will publish their own works, perhaps using someone like CDbaby as a distributor and keep what is left over as their profit (I believe CDbaby keeps 9% off the top and forwards the rest to the artist). Second, labels will have to restructure their contracts with artists into more of a profit sharing arrangement, where each party keeps X% of the sales, after original costs (recording, etc.) are re-couped.

(Long-time eMusic customers may remember that the problem of fixed per-track royalties was also what doomed eMusic’s original “all you can download” unlimited subscription model.)

Unfortunately this is not a problem that can be addressed in the short term, given that labels and music publishers aren’t going to change business practices overnight. In these circumstances it’s understandable why labels with unfavorable cost structures might want to leave eMusic if those cost structures prevent them turning a profit at all, and it’s not necessarily a slight on eMusic when they do so.

The next concern is that although selling through eMusic is profitable, it’s not profitable enough according to some measure (typically as compared to selling through the iTunes Store). Such concerns could be addressed either by eMusic raising prices across the board or by eMusic instituting some sort of variable pricing scheme that would allow labels to have more control over per-track prices.

Regarding across-the-board increases, it’s worth noting that eMusic has already raised its prices over the past few years since the unlimited subscription model was abandoned. For example, the eMusic basic plan for the US, which originally offered 40 tracks at $9.99, now offers only 30 tracks at the same price. This corresponds to an effective price increase of 32 per cent (from $0.25 per track to $0.33). Users in the UK and Europe also saw a price increase when the UK- and Europe-specific services were introduced. At some point further price increases would likely tend to drive customers away from the service, as David Pakman implied with his comments on price elasticity:

It’s the basic economic concept that says, for certain goods, when you raise the price, sales will fall disproportionately, and so the increased revenue doesn’t make up for the lack of sales. And if you lower the price, sales will rise disproportionately. Music is an elastic good, and we have now seen that by raising prices, the industry in fact did not make up the revenue, and, in the end, only slowed sales.

Regarding variable pricing, it’s difficult to see how variable pricing could be introduced in the eMusic subscription model without over-complicating it. One approach would be to change the subscription from being track-based to credit-based, with most tracks requiring one credit but some counting as two or more credits against the limit. This model has been adopted by Audio Lunchbox (another service offering DRM-free music from independent labels), but I question its viability. In particular, a credit-based model imposes considerable mental accounting costs on customers trying to decide what to buy, and thus likely dampens their enthusiam for buying anything at all.

(Audio Lunchbox takes this to insane levels: For a particular album you can purchase individual tracks at a cost of 1 to 4 credits depending on the track, you can purchase a whole album for a cost in credits that’s not necessarily that the same as the sum of the credits for the individual tracks, and you can purchase an album for an a la carte price in dollars that doesn’t necessarily match the cost of the album in the equivalent number of credits. A price-conscious customer would almost need to keep a calculator near to hand in order to use the service.)

The final concern is that the low prices at eMusic are attracting customers who would otherwise have bought either CDs or higher-priced digital albums at the iTunes Store or elsewhere. This in turn allegedly reduces label revenues and profits below what they would have been in the absence of eMusic.

To use Drag City as an example (a label that joined eMusic relatively recently): I was initially introduced to Joanna Newsom’s music on, and then bought her debut album Milk-Eyed Mender on CD (from Amazon) because it wasn’t on eMusic. However after Drag City came on board with eMusic I then got Ys from eMusic, even though I liked Newsom well enough that I would have almost certainly bought the CD if the album hadn’t been on eMusic. So in my case Drag City did in fact paid a price for releasing to eMusic.

However there are other cases where I’ve bought albums from eMusic that I would have never bought on CD or as full-price downloads. In fact, this was the case with the Joanna Newson and the Ys Street Band EP, which costs $10.99 from Amazon, $10 direct from Drag City (in CD form), $3.99 from the iTunes Store, but less than $1 from eMusic under the plan I’m on. Given that the EP had only one new song, in my opinion even $4 at the iTunes Store was too high a price to pay.

How to move beyond such anecdotal evidence? One approach is to ignore the underlying dynamics of the market and simply conduct controlled experiments: For example, suppose a label has two albums that from past experience it expects to be of roughly equal appeal to the market. It could release album A to eMusic and hold back album B, reserving it for the iTunes Store and other services that have price points higher than eMusic, and then compare the revenue and profits for each album. If album A does better overall then releasing to eMusic is a net plus, and if album B does better then releasing to eMusic is in fact a bad deal for the label.

However this wouldn’t necessarily settle the question, since we may be comparing apples with oranges: Album A may differ from album B in ways that would affect the relative revenue and profits. In particular, the set of people interested in album A may have different buying behaviors than the set of people attracted to album B. The goal is to measure the impact on releasing to eMusic or not, all other things being equal, but it’s hard to arrange the other things so that in fact they are equal in reality.

An alternative approach is to apply what’s known about economic principles and come up with some sort of business strategy vis-a-vis releasing to eMusic that is both easy to implement and likely to be effective. I’ll defer discussion of this to my next post. This post is long enough at this point, and this particular topic is important enough to deserve a post of its own.