As we saw in part 14 of the series on the streaming economy, most music creators receive so little revenue from streaming music that it is economically irrelevant. To understand this better, we need to look at the accounting mechanism for music streaming revenues. The most common model is the pro-rata model, which distributes revenue according to the market share of the rights holders. This is contrasted with the user-centric model, where actual usage by streaming customers forms the basis for distribution. This part of the series examines how these two systems work and the distributional effects they have.
The Music Streaming Economy – Part 15: Pro-Rata versus User-Centric
In the music streaming business, the pro-rata distribution model predominates. In this model, the music streaming service calculates the total monthly streaming revenue (after tax) according to the pricing models offered and retains an administrative share of around 30 per cent, as we have already seen. The remaining total amount to be distributed to the rights holders (labels, music publishers, collecting societies, digital music distributors, etc.) is divided by the sum of all payment-relevant streams (more than 30 seconds playing time) per month. This value is then multiplied by the repertoire shares attributable to the rights owners, resulting in the monthly payment amount per rights holder. If this value is higher than the upfront payment contractually agreed by the streaming service, the money is distributed in that amount. However, if the distribution is less than the contractually agreed upfront payment, the rights holder receives the higher amount.[1] The difference between the upfront payment and the actual payout is known in the industry as ‘breakage’, but more on that later.
The pro rata model can be mathematically formulated as follows:[2]
RX … Revenue for streamed track x
SX … Sum of all streamed tracks x
TS … Total number of streams
RG … Total streaming revenue
A very simplified example illustrates how the royalty model works. In 2022, the song “As It Was” by Harry Styles was streamed around 2.9 billion times (SX) on Spotify worldwide.[3] In the same year, a total of 1,600 billion streams (TS)[4] were recorded on Spotify and Spotify distributed around EUR 8.2 billion (RG) to all rights holders.[5] Applying these figures to the formula above, the share of the payout for “As It Was” is EUR 14 million. Harry Styles is signed to the Sony label Columbia Records. If we add up all the payouts for Harry Styles’ songs and do the same for all artists signed to Sony Music, we get the total payout from Spotify to Sony Music Entertainment. The payments are not calculated on a global basis, but separately for each country, considering the different pricing models. However, the simplified calculation should show how the money collected by Spotify is distributed.
In his meta-study on the distribution of streaming revenues, Rasmus Rex Pedersen shows that in the pro-rata model royalties are paid per stream, even though users pay a monthly flat rate, “which results in a cross-subsidization from low-streaming user to heavy-streaming user.”[6] He illustrates this with a highly simplified model based on two users, each of whom streams only one song, with different levels of usage intensity (fig. 1).
Figure 1: Revenue distribution in the pro-rata model
Source: After Pedersen (2020, p 5)
User 1 and user 2 pay the same monthly subscription fee of EUR 10, with user 2 being an intensive user of song B with 90 streams per month and user 1 only occasionally streaming song A with 10 streams per month. Since the revenues of both users are pooled in the pro-rata model, after deducting the share for the music streaming service, significantly more is distributed for song B than for song A. The aforementioned redistribution from occasional users to intensive users occurs.
The User-Centric-Model
The user-centric model has long been discussed as an alternative. As the name suggests, this model takes into account the listening habits of the users and distributes according to usage, as can be seen in fig. 2.
Figure 2: Revenue distribution in the user-centric model
Source: After Pedersen (2020, p 6)
This means that songs listened to are remunerated regardless of streaming intensity, with songs listened to intensively are worth less per stream. This would encourage niche repertoire because the payout per stream would be higher. Fans would feel that they are paying for the musicians they like, which is seen as fairer.
The user-centric model can be represented mathematically as follows:[7]
n … Number of all users
RX … Revenue for streamed track x
Si … Number of streams of track x by user i
Ti … Total number of streams by user i
Ri … Revenue generated by user i
The total revenue for a song is then calculated by adding up all the revenue from users who streamed the song over a given period.
Pro-Rata versus User-Centric: The Studies by Pedersen and Maasø
What would a change from the current pro-rata to a user-centric system mean for the distribution of music streaming revenues? Scientific studies have already been conducted on this, but they have not produced clear results. In 2014, Rasmus Rex Pedersen conducted a study using data from the music streaming service WiMP (now Tidal) and highlighted that the user-centric model would not change the distribution of income between the most streamed artists and those who serve a niche. He even shows that in the segment of the 5,000 most streamed artists, the user-centric model would lead to a redistribution in favour of the top 100 artists. Only domestic artists in the top 5,000 segment would benefit from the change, receiving 33.9 per cent instead of 30.8 per cent of the streaming pie.[8]
Arnt Maasø came to a similar conclusion in 2014, using the same methodology and also WiMP data for Norway, showing that there would be virtually no redistribution effect between top artists and less established artists if the pro-rata model were replaced by the user-centred model. Maasø also analysed redistribution effects between major and independent labels, which are not statistically significant. For Norway, however, there would be a redistribution from international to local stars if the user-centred model would be introduced. Artists working in Norway would receive more than a quarter of the streaming pie under the user-centred model, instead of the 22.5 per cent under the pro-rata model.[9]
Pro-Rata versus User-Centric: The Study by Muikku
A study conducted by Pradeep Durgam of Aalto University on behalf of several music stakeholders in Finland, written and published by Jari Muikku in 2017, comes to different conclusions than Pedersen and Maasø. However, the Finnish study is based on a different methodology, using all streams (more than 8 million) of Spotify Premium tier subscribers in March 2016 as a data basis. From this, a random sample of 22,496 streams from 8,051 user IDs was drawn and analysed. The artists behind the streams were divided into three groups. The top group consisted of those artists who had more than 100 streams in March 2016. This group represented 0.4 per cent of the sample. The second group consisted of those artists who had between 10 and 99 streams per month. This group represented 9.6 per cent of the sample. This left 90 per cent of the artists in the third and lowest group, who did not achieve more than 9 streams per month. A comparison of the two payment models, pro rata and user centric, shows a statistically significant difference in the top group. While artists in this group (more than 100 streams per month) were able to generate almost 10 per cent of all revenue from music streaming, the revenue share for this group in the user centric model was only 5.6 per cent. This means that the pro-rata model benefits top artists, while less established artists tend to benefit more from the user-centric model. However, the evaluation of the second and third groups of artists does not show clear results for or against the user-centric model, but the the study’s author still concludes that “the user centric model gives more direct power to users to target the money they pay for the service to the artists or tracks they favour compared with the pro rata model, which is not transparent from their point of view.”[10]
Pro-Rata versus User-Centric: The Study by the Centre National de la Musique
The contradictory results for Norway and Denmark on the one hand and Finland on the other are probably due to the different research methods used in the studies. In order to get a clearer picture of the distributional effects of the switch from the pro-rata to the user-centred model, the Centre National de la Musique commissioned a wide-ranging study with data from the year 2019, analysing not only the effects of the switch on the distribution of streaming revenues, but also on the diversity of music genres streamed and on streaming fraud. To this end, the authors of the study sought the cooperation of Spotify France and Deezer, but only Deezer fully implemented the methodological requirements. Therefore, the results of the study can only be considered valid for the Deezer data. The first and expected result of the French study was that switching to the user-centric model would result in a massive redistribution of Deezer’s streaming revenues from intensive users to average and occasional users. Instead of 69.2 per cent of all streaming revenues, heavy users (3rd quartile) would contribute only 31 per cent of the streaming pie in the user-centric model. In contrast, average users (between the 1st and 3rd quartiles) would generate 61.7 per cent (instead of 30.7 per cent in the pro-rata model) and occasional users (1st quartile) 7.3 per cent (instead of 0.1 per cent) of total streaming revenues. This is very plausible because, as we have shown above, the pro-rata model gives much more weight to heavy users.[11]
Music titles from the back catalogue, i.e. songs that are older than 18 months after the release date, would also benefit from the switch to the user-centric model. While back catalogue titles can account for 48.9 per cent of streaming revenues under the pro-rata model, this figure would rise to 52.1 per cent under the user-centric model – an increase of 3.2 percentage points. The five largest rights holders of back catalogues would benefit most from the switch, receiving a share of 42.7 per cent of the streaming pie instead of 39.6 per cent. On the other hand, the streaming share of the top 5 rights holders for current music titles would fall from 40.5 per cent to 38.7 per cent. The bottom line is that this would result in only a slight increase of 0.8 percentage points to a streaming revenue share of 88.5 per cent for the five largest rights holders. A switch to the user-centric model would mean a slight loss of 1.8 percentage points for the domestic French repertoire, from 44.3 to 43.5 per cent.[12]
However, the implementation of the user-centric model would have the greatest impact on the distribution of revenues to artists. The switch would result in a redistribution of 17.2 per cent of revenues from the top 10 artists on Deezer to the other categories, with the highest increase of +5.2 per cent going to artists outside the top 10,000. In absolute figures, the redistribution effect is even more evident. The top 10 artists would lose EUR 4.5 million in streaming revenue, which would correspond to an average loss of EUR 457,422 per artist. Of this, around EUR 810,000 would flow to the artists in the group of the top 11-100, who would receive around EUR 9,000 more per person. The group of top 101-1,000 artists would see the highest increase, with EUR 2.6 million, which would translate into an average of EUR 2,311 per person. EUR 348,000 would go to the top 1,001-10,000, but they would receive only EUR 39 per person, and finally EUR 1.6 million would go to the artists outside the top 10,000, but due to their large number this would translate into income of less than EUR 10 per person.[13]
Figure 3: The pro-rata and user-centric model in comparison
Source: After Jari Muikku, 2017, p 24
In addition to the income effects, the French study also calculated the impact on the diversity of music genres by changing the distribution model. The authors of the study identified a total of 28 different genres, with rap/hip-hop (25.8 per cent), pop (25.0 per cent) and rock (10.3 per cent) receiving the largest shares of the streaming pie in the pro-rata model. The long tail is made up of 21 genres that receive no more than 3 per cent each, including soul (2.2 per cent), jazz (1.1 per cent), metal (0.8 per cent) and classical music (0.7 per cent). Both pop and rock music would benefit from a switch to the user-centric distribution model, with increases of 13 per cent and 12 per cent respectively. However, classical music would see the largest increase at 25 per cent. Other niche genres would also benefit from a switch: blues (+18 per cent), disco (+17 per cent), jazz (+10 per cent) and folk (+8 per cent). The biggest loser in the user-centric model would be rap/hip-hop, which would lose 40 per cent of all streaming revenues. The results suggest that niche music could certainly benefit from a switch, with pop/rock music also likely to see growth, due to the strengthening of the back catalogue, as seen above, through the user-centric model.[14]
Finally, the study examined whether a switch from pro-rata to user-centred distribution could reduce streaming abuse. The authors’ assessment is not particularly euphoric. While the user-centric model would reduce the classic fraud of streaming farms that play certain music tracks around the clock, it would also create new opportunities for manipulation, such as hacking into underused and inactive accounts that could be used for fraudulent streaming.
The study also assessed the costs of switching from the pro-rata to the user-centric model, which was repeatedly cited as a counterargument.[15] However, qualitative interviews with representatives from Spotify and Deezer revealed a mixed picture. While Deezer representatives stated that the current distribution model could already provide sufficient funds for a switch to the user-centric model, Spotify interviewees estimated that a switch would lead to an increase in operating costs of 2 to 3 per cent, which is still a lot of money in absolute terms.[16]
SoundCloud’s User-Centric-Model
In practice, however, only the German artist and streaming platform SoundCloud has so far introduced the user-centric model in the form of “fan-powered licence payments” on 1 April 2021. The SoundCloud FAQ page explains how the model works.[17] This payment model is available to artists who are directly registered with SoundCloud in the “Premier Programme” or with “SoundCloud for Artists” (Next Plus or Next Pro subscriptions). As soon as users start streaming their music, the artists receive royalties based on the listening time or the number of adverts used. The use of a paid subscription for SoundCloud Go+ also plays a role. Artists who fulfil the requirements and have a large fan base that streams their music intensively benefit from the “fan-powered” licence payments. One year on from the launch of the user-centric model, SoundCloud has reviewed the situation and announced that a total of 135,000 musicians have benefited from the ‘fan-powered’ model. On average, these artists have earned 60 per cent more under the new revenue model than under the old pro-rata system. SoundCloud uses two examples – “Chevy” and “Vincent” – to demonstrate the increase in royalty payments. Chevy” received an average of 217 per cent more in monthly distributions, while “Vincent” saw his payout increase from US $120 to US $600 per month.[18] These are, of course, only anecdotal values, but they support the study results published by the Centre National de la Musique for France.
Endnotes
[1] See Music Managers Forum (MMF), 2015, Dissecting the Digital Dollar, part 2. Ashford: Hartley Brothers, pp 22-24.
[2] See Rasmus Rex Pedersen, 2020, A Meta-Study of User-Centric Distribution for Music Streaming, Report of the Roskilde University, Denmark.
[3] See Wikipedia, “List of Spotify streaming records”, version of July 4, 2024, accessed: 2024-07-05.
[4] This figure is derived from the total revenue that Spotify generated in the financial year 2022, divided by the average payout per stream of EUR 0.005. This results in approximately 1.6 trillion streams generated by Spotify worldwide.
[5] The 30 per cent administration fee is deducted from the total streaming revenue of EUR 11.7 billion (according to the 2022 annual report).
[6] Pedersen, 2020, p 3.
[7] Ibid., p 6.
[8] Rasmus Rex Pedersen, 2014, “Music Streaming In Denmark: An Analysis of Listening Patterns and the Consequences of a per User Settlement Model Based on Streaming Data from WiMP”, Working Paper, Universität Roskilde.
[9] Arnt Maasø, 2014, “User-centric settlement for music streaming: a report on the distribution of income from music streaming in Norway, based on streaming data from WiMP music”, Working Paper, Oslo University.
[10] Jari Muikku, 2017, Pro Rata and User Centric Distribution Models: A Comparative Study. Study commissioned by Digital Media Finland, p 12.
[11] Ibid., p 22.
[12] Ibid., pp 22-23.
[13] Ibid., p 24.
[14] Ibid., pp 25-26.
[15] This is the line of argument used, for example, by Will Page, Director of Economics of Spotify und David Safir, Vice-President International of ASCAP in the Music & Copyright Newsletter “Money in, money out: Lessons from CMOs in allocating and distributing licensing revenue” of August 29, 2018.
[16] Muikku, 2017, ibid. pp 27-28.
[17] SoundCloud, “Fan-powered FAQs zu Lizenzzahlungen”, n.d., accessed: 2024-09-23.
[18] Music Business Worldwide, “135,000 artists are now getting paid through SoundCloud’s Fan-Powered Royalties platform”, April 26, 2022, accessed: 2024-09-23.
Comments