Additional thoughts on per-subscriber streaming (from Medium)

Posted · Add Comment

https://medium.com/@dhuman/additional-discussion-about-per-user-on-demand-streaming-models-43afb4618472

I’m pleased to see discussion growing around per-user streaming models— my compliments to Sharky for putting together a strong presentation regarding this discussion. This model is one I and several others who do digital music strategy work have been kicking the tires on and discussing online in various newsgroups for quite some time. Streaming is unlikely to disappear, hence conversations about how to modify streaming models to produce different results are timely and valuable.

It’s been difficult to find actual modeled data from an existing streaming service to prove or disprove the hypotheses about revenue re-distribution in this. Yesterday I found one, an analysis of per-artist streaming on the WIMP digital music service by Rex Rasmus. While the author points out that his report does not take into account individual deals between streaming services and individual rights owners, and his report lands at the artist level and not at the individual user level, this is still illustrative. Here’s the link:

http://www.koda.dk/fileadmin/user_upload/docs/Analysis_Music-Streaming-In-Denmark_2014.pdf

I have several observations related to the conclusions of Sharky’s and Rasmus’ articles:

1 —It appears that Rasmus’ conclusion from his analysis is that per-user streaming tends to move money from the far end of the streaming music tail closer to the front end. The revenue movement has to come from somewhere — if the service makes the same amount of money, on the same amount of income, then any internal money shift has to happen from within the existing pool of distributed revenue.

I would make the argument that the intersection of this money shift with mid-level professional artists is exactly where this money shift is needed, if this is in fact where it occurs. Rasmus’ data seem to suggest that the front of the tail is going to encompass not only top-level artists (who seem to benefit from this metric because mainstream music streamers appear to stream less than niche music streamers and thus raise their per-user pro-rata payouts) but also a percentage of these mid-level artists — loosely defined as artists with a thousand true fans — who are also professionals and more dependent on streaming / download / digital income because many do not have day jobs and make music their primary career.

This is in contrast to starting bands and hobbyists who are potentially living at the far end and may be more interested in the promotional aspect of building an audience via participation in music services than in the income itself, at least until they start to grow to becoming mid-level artists as defined above. The end of the tail is likely composed of artists who are not making a living now on whatever money they’re making from streaming / downloads / record sales, and never have. I managed several of these artists in the 90’s; the main place they made money on their music was either from shows/performing live, from songwriting / sync / publishing income, or, if they were signed to a label or publisher, from their initial advance. They all had day jobs. It isn’t easy, but it’s never easy to turn a hobby into a career, be it in music or elsewhere.

2 — I wonder about what happens with breakage (users who do not stream anything in a particular month) in the per-user model? Many digital services address this by adding some portion of revenue into the total revenue pool which raises the overall rate for all artists by a small bit (more revenue compared to the same number of plays). Perhaps breakage continues to be dealt with in the per user model by a hybrid model of per-user for actual plays, and the traditional pro-rata of total revenue for breakage.

3 — In the per-user model, as discussed by Sharky, there’s much room for artist engagement with the process, but it needs to be coupled with information/data going back to the rights owner and to the artist. Let’s suppose the artist in question offers to play an internet concert for superfans who agree to stream primarily their music in a given month on a given service, for instance around release date. If the artist or label can gain access to information about who streamed their music — without running afoul of privacy laws, a real concern—then this creates a huge incentive for artists and labels to participate in driving streaming revenues. This is a big net positive for all players in the system — service, labels, artists, and consumers. Everyone gets something. Given the number of plays it takes to generate $7 under the current streaming paradigm (1000-ish plays at an average pro-rata overall rate of .7¢/play), it’s easy to see ways to use this system to reward artists that a consumer likes.

4 — Ad revenue based on a freemium (free limited subscription with hope for subscription upgrade) model could benefit and grow in this system, especially if it was also distributed per-user. If a freemium user streams primarily niche music—and it’s likely that niche music listeners are more likely to be responsive to new artists in general (niche music listeners on average are demonstrably the bigger buyers of music)— certain types of advertisers are certainly likely to look for people predisposed to try new things, and be willing to pay more for direct advertising access to that freemium individual than for a mainstream music listener who might be less responsive to trying something new.

Put this another way. Existing ad platforms like Facebook allow selection of advertising criteria based on, for instance, individual band Likes. Advertising to fans of Creeper Lagoon, vs. advertising to fans of U2, is going to provide a much more targeted group of consumers, and actual click-throughs vs. display ad views are generally better on average for more targeted advertising.

With data and per-user ad compensation models, individual artists are more in the driver’s seat of asking their fans to click through and support advertisers who are advertising around their music. Might not those increased CPMs also generate increased money for individual artists from freemium users on top of per user subscription fees?

5 — the main hole I find in all of this is the effective disincentive to stream a lot of music if you want to raise stream rates long term. This probably flies in the face of user behavior as bandwidth gets cheaper, interfaces for easy access to digital music (Sonos) get better, etc. But even if there’s a declining reward proposition to the artist or label for heavy plays built into the per-user system, what needs to be “beat” is the overall pro-rata on-demand streaming rate, which depending on the streaming service at this point in time largely fluctuates between .4¢ / 2¢ per stream. Sharky points out that the inflection point where more streaming makes the per user rate actually less than the pro-rata rate is approximately 1000 streams at a .7¢ rate. 1000 streams a month. Right now, and probably for the forseeable future, that would almost certainly define a serious power user.