The dramatic growth of music streaming over recent years is potentially very positive. Streaming provides consumers with low cost, easy access to a wide range of music, while it provides music creators with low cost, easy access to a potentially wide audience. But many creators are unhappy about the major streaming platforms. They consider that they act in an unfair way, create an unlevel playing field and threaten long-term creativity in the music industry. This paper describes and assesses the basis for one element of these concerns, competition between recordings on streaming platforms. It argues that fair competition is restricted by the nature of the remuneration arrangements between creators and the streaming platforms, the role of playlists, and the strong negotiating power of the major labels. It concludes that urgent consideration should be given to a user-centric payment system, as well as greater transparency of the factors underpinning playlist creation and of negotiated agreements.

By Daniel Antal, Amelia Fletcher & Peter L. Ormosi1

 

I. INTRODUCTION

Over recent years, music streaming has grown very quickly, with streaming revenues growing at around 50 percent per year globally. According to IFPI figures, in 2014, streaming accounted for just 14 percent global recorded music revenues. By 2019 this figure had increased to 56 percent, and the figure for 2020 ­is likely to be higher still.2 Some streaming revenues derive from advertising, but the vast majority for most platforms is derived from subscriptions. The revenue growth from streaming has helped to revitalize a flagging music industry, with global recorded music revenues now increasing again, after hitting a low point in 2014.

So far, so positive. But it means that streaming is now a critical part of the recorded music industry. A chunk of its growth has come at the cost of revenue streams from other forms of legal digital distribution and physical distribution. Over the medium term, it is also likely to threaten the viability of many commercial radio stations and music television, as the latter lose listeners to streaming services. All of this means that streaming has already become the most important route to market for recorded music, and this position is likely to strengthen over the near future.

It is therefore both timely and important to reflect on the extent to which streaming services provide a level playing field for those involved in producing recorded music: labels, publishers, composers, performers, producers. This is also topical since streaming has received substantial criticism, with many artists considering that they have been treated unfairly by streaming, despite the overall boost it has given to industry revenues.

Fair treatment is important not only for artists but also, over the longer term, for consumers. As we discuss later, if competition is distorted, this risks inhibiting innovation, variety and the prospects of upcoming and more niche artists. This in turn risks harming the long-term success of the music industry. A thriving independent sector has long been an important breeding ground for musical innovation, whilst niche areas of music ­– including music which reflects local culture and experience – can be exceptionally important culturally.

In this paper, we focus on one particular aspect of streaming platforms: whether they provide for a level playing field in terms of competition between (types of) recordings. We also briefly summarize some of the wider competition concerns raised in relation to streaming.

 

II. INTRODUCTION TO THE VALUE CHAIN: HOW STREAMING REVENUES ARE SPLIT

Streaming platforms obtain streaming rights for recorded music (i.e. the “fixation” of the recorded performance) from record labels (and sometimes individual artists) and from publishers or collective management societies for the music works. The three major record labels (Universal, Sony BMG, and Warner) account for around 67.5 percent of the overall global recorded music market, and each has historically had strong negotiating power with the streaming platforms, since their catalogue is vital for any credible streaming platform. The influence of the major labels and the deals they receive may also be enhanced by their small shareholdings (direct and indirect) in the streaming platforms. For example, Universal holds a 3.5 percent stake, and Sony Music a 2.9 percent stake, in Spotify,3 while Deezer is part-owned by Access Industries which in turn owns Warner Music Group.

The contracts that are signed between the platforms and the major labels are confidential and complex, depending on factors such as geography and whether streams were advertising-funded or subscription-funded. But effectively the platforms pay the labels royalties that are calculated ­on a pro rata basis, as a proportion of the revenues associated with the streams of their content. Independent labels and artists make up the remainder of the industry, and their royalty payments are also calculated on a pro rata basis. An organization called Merlin negotiates on behalf of a number of the larger independent labels, which account for around 12 percent of digital streams. The remainder of the industry has less negotiating power, and terms are effectively set by the platform.

The streaming platforms are also required to obtain licenses for mechanical rights and public performance rights in relation to the musical compositions. These fees ­– which go to publishers and songwriters ­– are often negotiated collectively through rights organizations, and typically also take a percentage form. The total payment for these rights is substantially lower than (around a quarter of) the payment made to labels.

 

III. WIDER COMPETITION CONCERNS

There is not an obvious lack of competition between platforms for consumers, at least currently. The largest platform is Spotify, with 320 million monthly active users, 144 million subscribers across 92 different geographic markets, and annual gross revenues of €7.6 bn (as at September 30, 2020), but it faces stiff competition. Apple Music is the next largest with around 70 million subscribers, while Amazon, YouTube (Alphabet), Deezer, Pandora (in the U.S.), and Tencent (in China) also provide successful streaming services. In addition, while YouTube offers an audio streaming service, its more popular video streaming service is also an effective competitor for listeners, with IFPI estimating that it accounted for more than 45 percent of all audio listening hours in 2017.4

However, the presence of vigorous competition between streaming platforms does not mean that there are no competition concerns.

A first competition issue highlighted by many parties, including Spotify and Deezer, relates to the commission fee charged by Alphabet and Apple for subscriptions that are made through Android or iOS devices, and the restrictions that are imposed (by Apple) to inhibit or prevent services from circumventing these fees. This arguably creates a competitive disadvantage for independent third-party services like Spotify and Deezer, relative to the vertically integrated services of these major tech firms. In the EU, the European Commission’s proposals for a Digital Market Act appear likely to address this issue, via the proposed obligation at Article 5(c).5

A second concern that has been highlighted is that the video streaming activity of YouTube effectively competes with specialist music streaming services, but (and in contrast to YouTube’s own music streaming service), it does not obtain proper licenses for the recorded music it streams, nor does it pay fees analogous to those paid by other streaming services. Instead it offers content publishers a share of the advertising revenues that it earns from their content. It is not straightforward to compare the resulting payments made by YouTube’s video streaming service on a per stream basis, but they appear to be substantially lower than are paid by the specialist audio streaming platforms (including YouTube’s own). As such, YouTube video arguably competes unfairly against specialist audio-streaming services, as well as against radio stations and television, and also treats creators unfairly.

Concerns have been raised about the transparency and lack of auditability of revenues and payments from YouTube’s video streaming service. These YouTube-specific concerns are analogous to similar concerns raised about Google’s conduct more widely, as discussed in the CMA’s market study into Digital Advertising.6 In the EU, the situation looks set to be improved by Articles 5(g) and 6(1)(g) of the European Commission’s proposed Digital Markets Act.

A third area of concern, expressed by many artists, is whether the total payments made by the platforms to third parties are fair. Comparisons between streaming and prior sources of income are far from straightforward, given the diverse sources of income ­– physical product, digital sales, radio plays ­– that streaming is replacing. However, streaming does appear to pay out less than many of these. For example, assuming an average streaming payout of around 0.004 Euros.7 By contrast, a single sale of a digital track via Bandcamp could easily generate a payment of 0.8 Euros. That individual purchaser would therefore need to stream the track an unlikely 200 times for it to generate the same royalties.8

On the other hand, while the precise payments made under its third-party licensing agreements is confidential, Spotify’s overall cost of sales is around 75 percent of its revenues, and it claims that the vast majority of this relates to such payments. Although transparency is limited, and payments differ across countries and types of streams, it appears that the other popular streaming platforms (apart from YouTube video) pay slightly more than Spotify on a per stream basis. While there is clearly potential for the revenue share offered to third party suppliers of content to be higher, it not necessarily low enough to be viewed as abusive under competition law.

Of course, another route to offering higher royalties to artists would be for the streaming platforms to increase gross revenues, either by charging more for the service, or by attracting more subscribers to the paid-for service.

It is true that streaming revenues are still increasing, and platforms have strong incentives to increase subscriber numbers. However, average revenue per user has been falling steadily since 2013, for Spotify at least,9 reflecting that these increasing revenues can be largely attributed to adding new subscribers at lower subscription rates.10 Part of this is due to the growth of streaming in new territories where subscription prices are set lower. For example, Spotify Premium costs around a tenth as much in India as in Denmark. While these low prices partly reflect market conditions, they also feed into low royalty payments from these territories, which will be especially detrimental to artists who are especially popular in those territories.

We note that the percentage nature of royalty payments in fact incentivizes lower pricing than would a more conventional set-up with fixed unit costs, since royalty costs automatically fall as prices fall. By contrast with the audio streaming services, Netflix – which has a more standard cost structure ­­– has been increasing its average revenue per user over time. For music streaming platforms, the total content consumed per user has also been increasing, meaning that the average revenue per stream has fallen substantially over time.

It should also be noted that ­– for any particular recording – the share of the revenues which gets passed back to individual performers, producers and composers depends on the deals they have signed with their labels and publishers, as well as any charges levied at the distribution level for distributing the recorded music to the streaming platforms. To some extent, therefore, complaints about the streaming royalties received by artists and composers are targeted as much at their labels and publishers and the collective management organizations as at the streaming platforms themselves.

A final area of concern, linked to the YouTube-related concern discussed above, is that streaming platforms are able to compete unfairly against traditional audio or television broadcasters, and increasingly expand directly into these activities. First, if these platforms can acquire music content at an unfairly low per minute rate relative to radio or television broadcasters, then this gives them an undue competitive advantage against broadcasters that offer higher royalties to creators.

Second, a royalty system which is based on revenue-sharing could have the effect of subsidizing expansion by the streaming platforms into these alternative markets. After all, the more audio content that a platform provides to any given listener in these other formats, the lower the proportion of total streams which recorded music accounts for, and thus the lower the payment that would be made to recorded music (subject to any minimum payment guarantees). This would hold true even if the absolute number of recorded music streams remained unchanged. Any such reduction in payment for recorded music would in turn act a subsidy to such expansion and also put these platforms at a competitive advantage, relative other potential operators of such content which do not have the benefit of the same subsidy.

As well as potentially comprising unfair competition, any substitution by streaming for more traditional media also has the potential to circumvent existing domestic media regulation in several territories. Such regulation is typically designed to guarantee a certain visibility for national, regional, or upcoming artists. The replacement of regulated broadcasting streams with unregulated digital (media) streams can reduce the royalty income of these protected groups.

 

IV. COMPETITION ON STREAMING PLATFORMS

While we agree that these various concerns discussed above merit attention, and welcome the European Commission’s relevant proposals in this area, they are not the focus of this article. Instead, we examine the conduct of the streaming platforms themselves, and how this supports – or distorts – fair competition between different types of recorded music and their creators.

We note that the issue of ensuring a level playing field in music creation has a long history. As long ago as the 1950s, “payola” ­(effectively, paying for airplay) was ruled illegal in the U.S. In 1959 Berry Gordy was getting so much airplay for his Motown artists that he even created a second label, Tamla, to dilute this success and thus allay rumors that he was using payola to achieve it.

The line between marketing and payola can be somewhat blurry. We note the concerns expressed that Spotify’s new “Discovery Mode,” which is currently an experimental feature, is a form of payola.11 However, the issue we primarily discuss below is a broader one. We examine whether there is a risk to competition in recorded music creation arising from the combination of:

  • the pro rata royalty allocation method adopted by the streaming platforms and
  • the role played by playlists in driving streams (with particular reference to Spotify) and the way in which these playlists are created.

A. The Pro Rata Revenue Allocation Method

While a pro rata royalty allocation method might appear fair in principle, in practice it can skew revenues and distort competitive incentives.

Consider the following simple example. Imagine one premium subscriber who likes to play mainstream music in the background all day, while they do other things, and a second premium subscriber who only listens to a streaming service for a couple of hours each evening, when they can really focus on the more niche music that they love. Suppose both subscribers pay 14 euros per month, of which 10 euros is distributed as royalties. The first subscriber might account for around 9,000 streams per month, while the latter accounts for only 1,000.

Even though the first subscriber only loves mainstream music and the second only loves niche music, the pro rata revenue allocation system will allocate 9/10 of the total revenue (i.e. 18 euros) to providers of mainstream music, and only 1/10 to the providers of niche music. In other words, 80 percent of the royalty payments generated by the second subscriber’s subscription fee go to providers of music that they would never choose to listen to.

Of course, a “per stream” unit royalty payment would solve this issue directly, and is often advocated. However, there is an alternative approach to revenue allocation that also overcomes this issue. Under a “user-centric payment system” (or “UCPS”), the royalties generated by an individual user’s subscription is simply split between what they choose to listen to. In the above example, all royalties generated by the second subscriber would go to the providers of the particular music that they listen to. This would in turn help to protect the creation, over the long term, of exactly that sort of music which the second subscriber is interested in, and would help to preserve competition, variety and innovation in the creation of new music.

Of course, there will in practice be many different types of subscribers to platforms, and some of the heaviest users may well have the most catholic tastes. However, the empirical analysis carried out to date confirms that a move to UCPS would benefit local/national tastes and domestic artists, and more niche tastes, over the more international and mainstream stars.12 Thus a move to UCPS would be expected to help support a greater diversity of local/national music scenes, to the benefit of their economies and culture; and a greater breadth of types of music, which should enhance the long-term innovation and sustainability of the recorded music industry. Deezer, which is itself endeavoring to move to a UCPS allocation system, but apparently facing resistance from the major labels, has calculated that the system would also be better for upcoming artists.13 This should again help to engender new and more innovative music creation.

Another substantial benefit of UCPS is that it would eliminate the current cheating of the system through the “farming” of streams. Currently, individual accounts can be used to play particular labels’ music non-stop, thus increasing their total share of revenues. Although the streaming services have systems in place to try and police such activity, it is inherently difficult, and can lead to “takedown” actions which some consider disproportionate.14 UCPS would solve this in one stroke, since the more streams racked up by one account, the ever tinier would be the royalty payments made per stream.

B. The Role of Playlists

Music listening on streaming platforms can happen in three ways: listening to single tracks, listening to albums, and listening to playlists. There is typically also an “autoplay” function, whereby the service streams tracks by default if a listener has not made an active choice. These work similarly to playlists and can reflect either what the listener was listening to previously and/or their prior usage.

Playlists can be invaluable in introducing listeners to new music. However, as has been demonstrated by Aguiar & Waldfogel (2018), they are also an extremely important source of streams on these platforms.15 For example, they found that being added to Spotify’s “Today’s Top Hits,” a list with 18.5 million followers during the sample period, increased streams by almost 20 million. As such, fair access by recordings to playlists is important for ensuring fair competition.

Since playlists require less user intervention then listening to single tracks or albums, it is also likely that heavier platform users, who tend to have music playing in the background while they do other things, are particularly likely to use playlists. As such, the impact of playlists on royalty payments is likely to be accentuated under a pro rata royalty allocation system relative to UCPS.

There are three main types of playlists: user-created playlists; proprietary playlists created by the platform itself, which may be editorial or “algotorial” (algorithmically generated editorial); and third-party playlists. It is the latter two types that raise potential competition concerns, in particular in relation to the ability of new or niche recorded music to gain access to playlists and thereby attract streams.

Listener numbers for playlists are highly skewed. The tables presented below are based on preliminary analysis of UK Spotify data collected by Mariuzzo & Ormosi (2020).16 The sample for the data analysis is 24,000 songs from recent recordings by UK artists signed to both major and independent labels. These 24,000 songs have appeared on around 1.7 million distinct playlists at various points in time, although most of these playlists have very few (or zero) followers.

The most popular playlist is Today’s Top Hits with over 27 million followers, followed by Global Top 50 (16 million followers), and RapCaviar (14 million followers). Table 1 shows how skewed playlist listenership is thereafter. The 100th most popular playlist in our sample (when ranked in descending order of the number of followers) has around 2 million followers. The number of followers per playlist then drops off sharply, with the 1,000th playlist attracting around 260k followers, the 10,000th playlist having only around 11k followers, and the 100,000th having just 55.

Table 1 – Skewness of sample of Spotify playlist followers

 

Number of followers

100th playlist

2,050,319

1,000th playlist

259,862

10,000th playlist

10,949

100,000th playlist

55

Similar concentration numbers are reported by Aguiar & Waldfogel (2018). This is important because it means that access to the most popular playlists (and especially the top 100 playlists) can be especially valuable for driving total streams.

The most popular playlists tend to be at least notionally editorial (edited by a curator), although curators tend to pre-select on the basis of algorithms. Editorial playlists can be created by the platform itself or by third party editors. Purely algotorial playlists (derived through an algorithm without human discretion) are created by the platform itself. Both types of playlist play important roles. While editorial playlists tend to be the most followed, algotorial playlists are also important. Even if the listener numbers are lower, there are so many of them that they can have a significant impact. Spotify alone has around 4 billion playlists.17

C. Proprietary Playlists

As is shown in Table 2 below, the vast majority of successful playlists on Spotify are proprietary. It curates almost all of the top 100 playlists and 90 percent of the top 1,000 playlists. This is not unusual, and indeed it may even more true on the other streaming platforms. The problem with proprietary playlists of either type is that it is unclear what are the criteria for inclusion, and whether all producers and publishers have equal access to inclusion, or fulfilment of the criteria.

Spotify states that, for its own editorial playlists, songs can in principle be pitched by labels or artists, that followers and listening history do not play a role in their playlisting decisions, and that Spotify does not allow payment for playlist placement. In practice, however, playlists can be hard to access. While this is understandable, given that around 40 thousand songs are uploaded on Spotify each day,18 it nevertheless creates a significant risk of unfair competition, given the importance of playlist access for driving both revenues and future success.

Algotorial playlists, which are created by the streaming platform’s own proprietary algorithms, are designed to predict user preferences and select content tailored to users’ individual data, based on advanced data analytics. They can take various forms, including providing for a high degree of personalization. We understand that they draw on three primary types of data as “signals”: the metadata of the music itself (its musicological or engineering properties), user data about how users have been interacting with the music or the artist (liking, disliking, skipping, playing) and natural language text data provided by the artist (or their representatives) or drawn from the internet.

The algorithms are confidential, and so it is hard to assess what biases they may introduce. However, even though playlists can play a valuable role in introducing followers to music they might not otherwise have heard, and even though the platforms have a competitive incentive to offer appealing playlists as part of their commercial proposition, these algorithms could nevertheless easily be biased. For example, any algorithms which are based on past global performance could result in “success breeding success” and are likely to favor more mainstream, established and international recordings. If music is initially misallocated to the wrong playlists, and receives negative feedback (such as in the form of listener skipping), then the algorithms may make it hard for it re-establish itself with the right audience. Music that doesn’t fit easily within an established genre, or which is not in the English language, is also likely to be competitively disadvantaged.

In general, the largest distributors, owned by the major labels, may be more effective in gaining access to the platform’s proprietary playlists. Spotify’s incentives to playlist songs from the major labels may also be influenced by their contracts with those labels. While these are confidential, Spotify states that they include minimum payment guarantees, which require it to make payments even if that label’s recordings do not hit a specified level of streams.19 Putting more of that label’s music onto playlists would clearly reduce the risk of triggering such payments.

Recently, as mentioned above, Spotify has gone one step further towards commercializing access to playlists. In November 2020, it announced a new “Discovery Mode” feature, whereby it would be allowing artists and record labels to flag tracks that are a particular priority for them, in return for receiving a lower royalty rate on these tracks. These tracks will then be given a degree of priority when algorithms are creating playlists and “autoplay” tracks.20

There are obvious analogies here to other platforms, such as hotel online booking sites, which also allow business users to pay to gain preferential rankings. Concerns have been raised that such conduct effectively exploits the trust that consumers have in “natural” rankings, and the fact that they assume them to be user-orientated, rather than advertising-funded. This has led to calls for sponsored content to be clearly distinguishable from organic content.21 In the specific case of hotel online booking, the UK CMA has required sites to set out clearly how such commercial factors underpin their rankings.22 At the very least, it would seem reasonable to require this of streaming services too, and it is arguably required in the EU under the Unfair Commercial Practices Directive 2005.

D. Third-party Playlists

While third-party playlist curation could be potentially a competitive market, we note that the most successful third-party playlist operators would appear to be the major labels themselves. Filtr (Sony BMG), Digster (Warner), and Topsify (Universal) primarily play their label’s proprietary content. Aguiar and Waldfogel (2018) find that Filtr, Digster, and Topsify, respectively have 3.1, 2.7, and 0.9 percent of the top 1000 playlists’ cumulative followers. Table 2, based on the Mariuzzo & Ormosi (2020) data, shows similar prevalence. The first column in Table 2 shows that cumulative “follows” are heavily concentrated in the top playlists.23

Table 2 – Share of “follows” across different playlists24

Cumulative share of total “follows”

% Spotify playlists

% Sony playlists

% Warner playlists

% Umg playlists

% other third-party playlists

Top 100 playlists

26.8

98.4

0.6

0

0

1

Top 1,000 playlists

65.5

90.6

3

0.4

0.3

5.7

Top 10,000 playlists

92.7

73.9

4

1.2

1

19.9

Top 100,000 playlists

99.9

68.9

4

1.3

1.1

24.7

E. Implications for Playlist Access

The overall effect of the above factors is that major label recorded music has a greater share of the most popular playlists, which really drive streams, than they do in the less popular playlists. This is shown clearly in Table 3 for the Mariuzzo & Ormosi (2020) data. While this data sample does not provide a full description of the UK recorded music market, we note that total share of independent labels in the total UK recorded music market is around 30 percent . As such, a share of less than 25 percent of a playlist, for the top 10,000 playlists analyzed here (which account for over 90 percent of cumulative “follows”)­ would appear consistent with differential playlist access.

Table 3 – Share of major label recordings on sample of Spotify playlists

 

% Major label recordings

Top 100 playlists

81.0

Top 1,000 playlists

77.2

Top 10,000 playlists

77.7

Top 100,000 playlists

64.6

This preliminary analysis suggests that independent label artists are getting far less than their fair share of access to the most popular playlists. While the vast majority of these are curated by Spotify, the shares of the major labels’ own proprietary playlists may exacerbate the situation. This lack of access is likely to have a direct impact on revenues for independent labels and their artists today, and also an indirect impact on the sustainability of this important segment of the market in the future.

Given the importance of playlists for driving streams, any potential informational advantage to how they are created can provide a huge advantage for suppliers of recorded music. Providing greater transparency as to how each streaming platforms’ proprietary playlists are created would therefore seem to be vital for ensuring fair on-platform competition. In this context, we note that Spotify has an unusually open API, which provides artists with more detail on how their music is being represented than do its competitors. However, greater transparency and auditability would be valuable from all streaming services.

Finally, we note that only the streaming platforms themselves are currently able to offer algotorial playlists, as only they have access to the relevant data about every listeners’ streaming choices that is required to train the algorithms. But this situation is not inherent. With greater access to such data, for example through explicit data access requirements or through data portability provisions, there may be more potential for the development of an effective competitive market for third party playlists, even including enhanced algorithmic and personalized playlists. This should be given serious consideration, especially in the context of the wider policy debate on smart data access and portability.

 

V. CONCLUDING THOUGHTS

We have argued that fair competition in music creation risks being compromised by the combination of several factors: the nature of the royalty arrangements with the streaming platforms, the role of playlists, access to critical data, and the strong negotiating power of the major labels. These distortions tend to favor more mainstream, established and international music, in particular that which appears on major labels, and to disadvantage the more niche, the more independent, the more locally-focused.

This would be an important issue even in normal times, but at a time of a world-wide pandemic, which has effectively shut down many revenue streams for artists, and left streaming as their almost exclusive source of revenue, its significance is even more pronounced. Moreover, it has long-lasting potential consequences. Although currently consumers have low price access to an unprecedented selection of music, they will be harmed over the longer term if the current arrangements harm competition, variety, innovation and cultural diversity in music creation.

In terms of possible solutions to the concerns identified above, we recommend that urgent consideration should be given to moving from the pro rata payment system to a user-centric remuneration. We note that Deezer is already pushing for this change. We also believe that further analysis would be valuable into the role of, and access to, the main editorial playlists, as well as the algorithms underpinning the algotorial playlists, in order to ensure they do not create an unbalanced playing field. There should be greater transparency and auditability around how these playlists are created. Enhanced data access would also be valuable in facilitating the development of a competitive and innovative market for third-party playlist creation.

We would encourage greater transparency of contracts, once they are agreed, to help ensure fair treatment; or alternatively that competition authorities should allow industry-wide negotiation by labels, as is already carried out for performance and mechanical royalties on the composition side of the split. We note that some of the majors have residual equity stakes in Spotify. Requiring divestment of such stakes could also be helpful in ensuring that Spotify has the right incentives to ensure a level playing field.

Finally, while our focus here has been on competition considerations, we note that this is only one policy area relevant to streaming platforms. Copyright law and broader media regulation may also merit rethinking in the streaming era.


1 Daniel Antal, CFA, is the co-founder of the reproducible research startup company Reprex B.V. He has been advising rightsholder assocations and musicians’ organizations on royalty pricing and copyright infringement damages calculation since 2012. Amelia Fletcher and Peter L. Ormosi are affiliated with the Centre for Competition Policy and Norwich Business School, University of East Anglia. Amelia Fletcher is also a Non-Executive Director at the Competition and Markets Authority (“CMA”). Amelia is also a small-scale performing artist, and has music available on streaming platforms. This article reflects the authors’ own views, and not those of any organizations with which they are affiliated.

2 https://www.ifpi.org/wp-content/uploads/2020/07/Global_Music_Report-the_Industry_in_2019-en.pdf. In the case of the UK, the industry body BPI estimates that streaming now accounts for four-fifths (80.6 percent) of UK music consumption. See: https://www.bpi.co.uk/news-analysis/fans-turn-to-music-to-get-through-2020-as-a-new-wave-of-artists-fuels-streaming-growth/.

3 Warner Music Group sold its own equity stake in Spotify in 2018. See: https://www.musicweek.com/labels/read/warner-music-group-sells-final-spotify-equity-stake-as-digital-revenue-soars/073434.

4 Globally, excluding China. See: https://www.musicbusinessworldwide.com/more-music-is-played-on-youtube-than-on-spotify-apple-music-and-every-audio-streaming-platform-combined/.

5 Proposal for a Regulation of the European Parliament and of the Council on contestable and fair markets in the digital sector (Digital Markets Act), 2020/0374(COD), https://eur-lex.europa.eu/legal-content/en/TXT/?qid=1608116887159&uri=COM%3A2020%3A842%3AFIN.

6 https://www.gov.uk/cma-cases/online-platforms-and-digital-advertising-market-study.

7 This is a generous estimate. Estimates of average recorded music royalties per stream on Spotify, as the largest streaming platform, range from 0.00284 to 0.0039 Euros (calculated using a 2019 average dollar/euro exchange rate of 0.8931). See: https://soundcharts.com/blog/music-streaming-rates-payouts.

8 We note that, based on data from a mid-sized indie label, it has been calculated that each individual song only receives on average around 350 streams per year in total (not per listener). See: https://thetrichordist.com/2020/03/05/2019-2020-streaming-price-bible-youtube-is-still-the-1-problem-to-solve/.

9 https://www.businessofapps.com/data/spotify-statistics/#:~:text=Playlists%20are%20the%20backbone%20of,listening%20to%20Spotify%2Dcurated%20playlists.

10 Moreover, this is true for streaming more broadly, as shown in Antal (2020), which reviews monthly streaming incomes from 330,000 sound recordings in the period 2015-2019, in the 20 advanced and emerging markets. Antal, D. (2020) Central European Music Industry Report 2020, available at: https://ceereport2020.ceemid.eu/market.html#recmarket.

11 https://www.musicbusinessworldwide.com/spotify-is-letting-record-labels-influence-personalized-recommendations-so-long-as-they-pay-for-it-in-royalties/.

12 See: https://www.fim-musicians.org/wp-content/uploads/prorata-vs-user-centric-models-study-2018.pdf;

https://www.koda.dk/media/224782/meta-study-of-user-centric-distribution-model-for-music-streaming.pdf;

https://www.hf.uio.no/imv/forskning/prosjekter/skyogscene/publikasjoner/usercentric-cloudsandconcerts-report.pdf.

13 https://musically.com/2019/09/11/deezer-steps-up-its-efforts-to-introduce-user-centric-payments/ and https://musically.com/2020/10/01/deezer-still-pushing-for-user-centric-payouts-we-will-continue-fighting/.

14 https://www.digitalmusicnews.com/2021/01/06/spotify-reportedly-pulls-indie-songs/; https://www.change.org/p/spotify-restore-our-music.

15 Aguiar & Waldfogel (2018) Platforms, Promotion, and Product Discovery: Evidence from Spotify Playlists. https://EconPapers.repec.org/RePEc:ipt:decwpa:2018-04.

16 This dataset has been collected for the purposes of a currently ongoing research project. The data contains information (streaming numbers, playlist activity, song features, etc.) on a large sample of songs by UK artists signed with major or independent labels. Details of the data are available in Mariuzzo, F. & Ormosi, P. (2020), Independent v major record labels: Do they compete on a level-playing streaming field?, available at: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3729966.

17 https://www.businessofapps.com/data/spotify-statistics/#:~:text=Playlists%20are%20the%20backbone%20of,listening%20to%20Spotify%2Dcurated%20playlists.

18 https://www.musicbusinessworldwide.com/nearly-40000-tracks-are-now-being-added-to-spotify-every-single-day/.

19 See p 12, https://s22.q4cdn.com/540910603/files/doc_financials/quarterly/2019/601c445e-1d37-4938-b854-e5344850c3f9.pdf.

20 https://newsroom.spotify.com/2020-11-02/amplifying-artist-input-in-your-personalized-recommendations/.

21 For example, in relation to Google and Facebook, the CMA’s recent market study report into digital advertising proposes regulation that would require platforms to “ensure advertising is presented in a way that is clearly distinguishable from organic content.” See Table 7.3 in https://www.gov.uk/cma-cases/online-platforms-and-digital-advertising-market-study.

22 https://www.gov.uk/government/publications/online-hotel-booking-principles-for-businesses.

23 We talk about cumulative “follows” since the same subscriber is likely to follow several playlists.

24 The playlist data was collected from www.spotontrack.com.