This is the fourth and final in a series of posts (following parts 1, 2, and 3) speculating on Amazon’s rumored entry into the digital music market. In this post I discuss two additional music-related services Amazon might offer, particularly to customers already signed up to a subscription plan. The standard disclaimer applies: This is all fevered speculation and nothing more; I do not have any inside information about a possible Amazon acquisition of eMusic, nor about other future plans of Amazon or eMusic. Here follows my final attempt to play foolish prognosticator:

Amazon will offer a backup service for digital tracks purchased through Amazon or otherwise, leveraging the infrastructure used for the Amazon S3 offering. As noted in my first post, CDs can serve as a form of backup for people buying digital music. However at the same time many people (including current eMusic customers) have digital music collections for which they do not have the corresponding CDs. Amazon already has a “Simple Storage Service” offering whereby (sophisticated) customers can store data on Amazon’s servers, at a price of $0.15 per GB per month for the actual storage used and $0.20 per GB of data transferred from the customer to Amazon. This existing infrastructure will be leveraged to provide a custom service whereby Amazon customers can backup all or part of their music libraries to Amazon’s servers.

The service will work as follows: When purchasing digital music, whether by itself or in conjunction with a CD purchase, Amazon customers will be offered the option of having Amazon keep a copy of the track(s) in a “music locker” tied to their existing Amazon accounts. No separate registration step or account setup will be required, and nothing will actually need to be transferred from the customer to Amazon (since Amazon already has a copy of the track(s) in this case). For customers currently signed up to a subscription plan (as discussed above) the cost of backup will simply be added as an incremental charge to the customer’s existing monthly subscription fee. For customers not on a subscription plan the cost of backup could be charged as a one-time fee, based on Amazon’s future costs for the needed disk storage over the customer’s lifetime.

Amazon customers will also be able to able to backup digital tracks previously purchased through other services or ripped from CDs. In this case the backup will be done by Amazon-provided desktop software. For example, such capability could be included in an Amazon-specific download manager (like the current eMusic download manager) or a full-fledged Amazon digital music player (like iTunes or Songbird).

Assuming pricing comparable to current S3 pricing, a customer with a 40GB music library could back it up through Amazon for $6 per month at present, with this cost decreasing over time. (This assumes that Amazon actually stores separate copies of a given track for each customer. If copies can be shared among customers then the price could be reduced even further.) The music locker service will thus be a relatively straightforward up-sell for customers already on an Amazon music subscription plan.

Two final notes: First, this service will not be like eMusic’s current scheme whereby customers can re-download tracks previously downloaded. That service works only for tracks currently offered by eMusic; if a label pulls its offerings from eMusic then customers lose the ability to re-download previously-purchased tracks from that label. This is not acceptable in a true backup service.

Second, this service will not be usable as a way to share tracks with other users (at least not in practice). Each locker will be specific to a customer and will be accessible only using that customer’s standard Amazon userid and password, the same credentials used for one-click purchases and other actions. A customer allowing others to download from his or her account would be putting his or her own credit card account and personal information at risk.

Amazon will offer a personalized music recommendation service based on a combination of the current Amazon recommendation engine and human tweaking of recommendations done using Amazon’s Mechanical Turk service. Amazon already has a number of features to drive customers purchasing one item to consider purchasing another one, including automated recommendations based on a customer’s past buying history, information on related items (e.g., “people who bought this item bought these items too”), recommendation lists created by other customers, and so on. If it acquires eMusic then Amazon can also add eMusic’s editorial content (eMusic Magazine, the eMusic Dozens, and so on) to the mix.

As I noted in an earlier post, a major problem for many customers on a subscription plan is finding enough music worth downloading each month; this is especially true for busy people who are interested in discovering new music but do not have the spare time to explore the net reading music-related web sites and blogs and listening to sample tracks and Internet radio stations. On the one hand eMusic (and in future Amazon) is happy enough to have customers paying money and not actually using the service. On the other hand customers not getting value for money will eventually end up being ex-customers, and possibly disgruntled ones at that. Thus Amazon will have an incentive to keep subscription plan customers satisfied and wanting to come back for more.

In an ideal world each customer would have a real person serving as their dedicated guide to the world of new music, with that person taking into account the customer’s unique tastes. The eMusic message boards provide frequent examples of such personalized recommendations provided by live humans in response to requests, but unfortunately the message board approach won’t scale.

Perhaps the optimum way to approach this problem is to start with a machine-generated list of recommendations and then have a human review the list to vet it for compatibility with a given customer’s tastes, as well as to perhaps suggest additional items which the recommendation algorithms somehow missed. As it happens, Amazon already has the basic infrastructure needed to construct such a mixed automated/manual service, in the form of the Mechanical Turk service, its existing recommendation engine, and related technologies. As with music backups I think this is a service for which many customers will be willing to pay extra, for example as an additional charge (say of $2-3 per month) over and above the subscription plan price.

Ideally the personal recommendation service could be configured to have recommendations be automatically purchased and scheduled for download to the customer’s computer (e.g., when the customer nexts run the Amazon-provided software). That would minimize the total amount of work time-starved customers would have to do. However note that it will not be in Amazon’s interest to have a customer’s entire download allotment used up in this manner, for reasons noted earlier. It will be better to leave the customer a partial quota of downloads that they can either use at their own discretion or let expire unused (and thus subsidizing more active customers).

This completes the fourth and final post in the series. I’m all written out at this point, so I don’t have any grand final conclusions to draw, other than that I think that online digital music stores are past due for more customer-friendly pricing and useful customer-oriented innovations, and that based on its technology platform and music retailer experience Amazon is perhaps the most likely company to bring such pricing schemes and innovative services to market.

music promotion guy - 2010-02-20 16:33

how much of this has happened?

hecker - 2010-02-20 20:07

music promotion guy: Very few of my predictions regarding Amazon came true–pretty much only the conventional ones about Amazon selling only MP3s and trying out various pricing schemes. I think my other ideas are worthwhile, including the one about treating selling CDs as upgrades to MP3s and being able to get MP3 downloads as soon as you purchase your CDs. However existing licensing schemes by rightsholders will I think prevent this for the foreseeable future, until such time as the CD is a dead format in any case. P.S. Even if yours was a spam comment it was a relevant one, and worth responding to. Your prize is to get your link on my blog :-)