Uncategorized

Streaming and Royalties: Analysis and Real-World Figures

Streaming platforms are notoriously opaque about how much they pay per stream. For Spotify, we already had unofficial estimates, which showed that stream counts had to be considered alongside factors such as the listener’s geographical location. A few weeks ago, Qobuz, the other French streaming platform alongside Deezer, announced the unthinkable: a figure!
A stream on Qobuz would generate €0.018 compared with €0.003 on Spotify (the figure generally cited in the music industry). I decided to investigate and analyse these digital-distribution figures to get a clearer picture.




1. Qobuz: making waves

It all started with a surprising Facebook post that prompted me to write this article: Qobuz sponsored by Europe 1. And on top of that, Qobuz was talking about artist remuneration and actually giving a figure. A double surprise.

Spotify vs Qobuz: very similar entry-level pricing

Qobuz is the first music streaming platform to officially publish its average payout rate per stream. Six times more than Spotify! How is that possible? Qobuz does not offer a free tier, so every listen comes from a user paying for a subscription: Qobuz’s entry-level plan costs €12.49 per month, which is not far from Spotify’s €11.12. For less than €1.50 more per month, you can pay for a streaming service that is more favourable to artists. Fair trade always costs more than the “standard” market, but rarely by so little compared with the major corporations dominating the market. Qobuz also offers music downloads, which pay more than streaming.

Using the data available to The Queen Is Dead Records, let’s see whether these claims hold up. Qobuz’s current burst of “transparency” nevertheless reveals a blind spot: the announcement of its payout rate does not specify whether it has always been in force or was only introduced this year. Transparency without any historical context is still a form of opacity. We’ll have to work with that!



2. The figures

Above is a complete table of data collected by The Queen Is Dead Records across all platforms, together with detailed figures and averages for the Beatport, Facebook, Spotify, TikTok, Trebel and Youtube subgroups.

Subgroups (examples: Beatport, Beatport Stream, Beatsource) are shown in italics and their data appears in paler-coloured cells. Groups (examples: BEATPORT, FACEBOOK) are shown in bold capitals and represent the average of the data from their associated subgroups. Subgroup and group data are gathered inside coloured frames (alternating pink and green).


3. Analysis conditions: The Queen Is Dead Records’ figures

Total streams measured for releases distributed by The Queen Is Dead Records


Before trying to read and interpret the information in the main table, it is important to understand how this data was obtained.

As of 18 April 2025 (when I started writing this article), The Queen Is Dead Records’ distribution service accounts for a total of 580,000 streams shared across 311 artists, 1,330 tracks, 140 singles, 61 EPs and 77 albums. These streams are split between two management systems operating in parallel:

• From July 2013 to the present:
150,000 streams generated through AWAL, a digital distributor that is not open to the general public and offers relatively little flexibility. This includes only label releases up to September 2021, as well as the first year of digital distribution handled by The Queen Is Dead Records, from September 2021 to August 2022. Since September 2022, no new releases have gone through AWAL, but releases distributed during that first year have remained within AWAL’s management system, so updated data for those releases continues to be collected.

• From September 2022 to the present:
430,000 streams generated through the current system, with direct management allowing much greater flexibility. Since this is the system containing the most data (three times more than the AWAL system), I based my investigation exclusively on this source, avoiding any mixing that might create alignment issues because the two databases do not use exactly the same structure or types of data.

These 430,000 streams provide a sufficiently substantial dataset to conduct an analysis with some perspective. All this data (an Excel document containing 75,000 rows and 20 columns) gives the observations presented in this article a degree of legitimacy and therefore a certain credibility.



4. Every choice means giving something up

Each geographical area has its own streaming platform (Source)


A table with that many rows doesn’t tell you much at a glance: visually scanning it would take too much time and physical effort. And mental effort. With so many values below 1, it is difficult to distinguish proportions and contrasts (“Are there five zeroes after the decimal point, or six?”). Whether presented as a table or a graph, exhaustiveness becomes less useful here. So I need fewer streaming platforms.

I am therefore going to exclude two types of streaming platform:

• Not the target: some platforms are largely unknown where we are because they target specific geographical areas outside Europe. Those platforms, along with any for which I have never seen a public mention (article, advertisement, documentary, discussions with musicians, etc.), will therefore be excluded.

• Too little data: some platforms have very few data points, making it impossible to identify meaningful trends. Most of these platforms target specific geographical areas outside Europe, but a European platform with too little data would also be left out.

Taking these two points into account, my selection will be limited to platforms we know in Europe and for which I have enough data to draw an interpretation. I am therefore keeping 12 streaming platforms: Beatport, Qobuz, iTunes, Amazon, Tidal, Deezer, TikTok, Facebook, Youtube and Instagram, to which I will add a thirteenth. I cannot ignore Tencent, which may be the leading music streaming service in China (the only territory it covers), but nevertheless reaches the remarkable figure of 555 million users out of a population of 1.5 billion. By comparison, Spotify scores only slightly higher, with 675 million users worldwide.
Tencent therefore deserves to be included in the analysis.



5. Don’t trust payout tables

Example of a typical table presenting fixed figures that are therefore completely unusable and never hold up in practice.

Now that I have selected the 13 platforms to consider, I can produce a data table like the ones we are used to finding online (see above). To carry out an analysis, you need to know which data you are going to examine, extract and compare. The payout for a single stream seems like the most obvious and easiest thing to look at. Everyone does that online. But this approach leads straight into a wall.

For every streaming platform, there is no single payout for one stream. Depending on the listener’s subscription and geographical location, a stream does not generate the same amount for the artist (and we will see later that other “surprising” parameters can also come into play). This is why, even though we have all done it for years, it is simply not reasonable to look for a table containing fixed values for each platform, because those figures do not hold up in the real world.

Moreover, the per-stream payout figures found online are rarely dated. Consider the table above, whose source is Digital Music News. The exact same figures can be found online in publications from April 2020, December 2021, July 2024 and November 2024, while lower figures were already visible in April 2020. What explains the difference between these different figures: the date or the source? The lower figures are probably the more recent ones, but all of this data is certainly outdated because all of it was already being circulated in 2020.

In the end, no data table can be considered “the right one”. We therefore cannot rely on this kind of document for precise information to use in calculations. Of course, you could argue that it still gives a rough idea. But within a single platform, the differences can be so large that the average becomes an approximation that is simply unusable. The wider the range, the more reference points you need to understand what you are looking at from enough different angles. Several figures therefore need to be considered for each platform.



6. Analysis method: average and median

Let’s do some maths!

The more data points there are (numerical units of information belonging to a dataset), the more reliable statistical analysis can become. If there is only one data point for a particular streaming platform, the minimum, maximum, average and median will inevitably all have the same value. That is of no use statistically, because no meaningful statistics can be produced from one isolated value.
Among the statistics that can help reveal an overall trend, we need to distinguish between averages and medians.

The average is the sum of the values divided by the number of values. The median is the value located in the middle of a dataset sorted in ascending or descending order: there are as many values above it as below it.
For example, consider the following data: 1, 2, 3, 4, 100.
The average, influenced by the extreme value (100), is 22; the median is 3. Quite a gap! Neither figure is more true or accurate than the other. Considering both at the same time remains important because they provide two different angles, two overlapping perspectives that offer relevant material for a synthesis while still taking exceptional outlying values into account and trying to understand them.

To study internal variations within each service, we need both to consider extreme values as significant information and to avoid letting them distort the general trend. For my analysis, I will therefore not use the modal value (or “mode”).
The main weakness of modal analysis is that it focuses only on the highest frequencies, without taking the overall distribution of the data into account.
For example, consider the following data: 1, 1, 1, 99, 99, 100, 100, 101, 101.
The mode is 1, because it is the most frequently observed value (3 occurrences), thereby ignoring more distant values (99, 100 and 101) that occur less frequently (2 occurrences each). Although less frequent and different from one another, these values form a coherent set that is more representative of the overall distribution.
Looking at modal values would therefore lack perspective, focusing too narrowly and selectively on details whose relevance can be somewhat arbitrary, or even questionable.
This raises a crucial question: should some data be grouped together to calculate an average? And if so, which groupings would be relevant? The answer is far from simple.

Because the median is resistant to extremes while the average is influenced by them, and because the aim is to obtain a view that is both precise and comprehensive, combining the perspectives of the median and the average seems to me the ideal recipe: simple and effective, and providing a fuller view of the dataset being examined.

The gap between average and median should be read by asking two questions:
– Are the values identical/close, or not?
From a 10% difference upwards, the gap can be considered notable.
– If the values differ significantly, which one is higher?
• Average > Median (right-skewed distribution):
a few isolated high values are artificially pushing the average upwards
• Average < Median (left-skewed distribution):
a few isolated very low values are pulling the average downwards



7. A first chart

Click the chart to enlarge it

Here is the chart for my selection of 13 streaming platforms, highlighting the average and median “payout per stream” values (based on the data in the initial table – an important clarification).

The first thing that stands out is the gap between the average and median values, which shows that a stream does not always generate the same revenue within a single platform.

Platform names written in capitals refer to groups combining several types of data: these datasets are made up of different entities and programmes, with different payout rates. For readability, I chose to merge the data within these groups in order to obtain a simple figure (an overall average) that can be presented intelligibly. The downside is that this overall average is an amalgam, resulting in a highly abstract figure. These groups will therefore need to be examined in greater detail. And they will be in this article.

We also need to distinguish between platforms providing access to streams, those offering downloads, and those offering both. Downloads generate more revenue because they are outright purchases, whereas streaming is closer to a form of rental. Platforms doing both will therefore see streaming pull their payout rate down, or downloads push it up – it all depends on how you look at it. Not every platform should be evaluated in the same way: where the data allows it, distinctions can be made for a fairer analysis.



8. Platforms with multiple faces

In the initial table, the data is reported separately for different programmes or entities belonging to the same group. For greater clarity (and a few interesting revelations) than the first chart, which combines subgroups, a slightly deeper analysis is needed, particularly to avoid two things:

– lumping together subgroups attached to the same group (for example Beatport, Beatport Stream and Beatsource within BEATPORT) even though they have different payout rates

– analysing insufficient data hidden by overall averages (if one subgroup contains very little data, its weighting would be too strong compared with another subgroup containing a huge amount of data).

Let’s take a closer look at all the subgroup data, averages and medians, supplemented with minimum and maximum values to get a more precise picture of the differences between entities operating under the same brand.



9. Beatport: electronic music, but not only

The “BEATPORT” entry combines three entities:

a – Beatport
This is the electronic-music download platform you probably already know.
Minimum: €0.06 / Maximum: €1.14
Average: €0.13 / Median: €0.10
The median is close to the minimum: this means there is a large majority of clustered low values, along with a few downloads at a more attractive rate generating 64 cents per track.
Such a wide gap for downloads may seem surprising: paying only 3 cents for a track is clearly low, especially when you know that a track costs at least €1. That is quite a margin!

b – Beatport Streaming
A lesser-known service, this is Beatport’s streaming version, offering the same content as standard Beatport but operating through a subscription.
Minimum: €0.08 / Maximum: €0.09
Average: €0.08 / Median: €0.08
The median and average are very close, and there is very little difference between the minimum and maximum values: this streaming service is clearly more even-handed than some other platforms, to the point where a stream can potentially pay more than a Beatport download (3 cents vs 4 cents).

c – Beatsource
A little-known platform that, unlike Beatport, is not exclusively focused on electronic music (it leans more towards pop and mainstream music) and is primarily aimed at DJs, who can access special versions of tracks through DJ software, such as extended intros, instrumentals or a cappellas.
Minimum, maximum, average, median: €0.16
Unlike the other platforms, for which I have hundreds or even thousands of data rows to analyse and use to identify trends, for Beatsource I have only one stream for one track. Its value, almost 10 cents, is very high, and this general level of remuneration is consistent with what the platform itself claims: it reports an average of 15 euro cents per stream. We do not know the gap between minimum and maximum values, nor the criteria used to establish any scale. In any case, it is far more than Spotify, which makes sense because this platform targets a niche audience paying for subscriptions.



10. Spotify: no mercy

The “SPOTIFY” entry corresponds to two different calculation lines:

a – Spotify
Everyone knows Spotify, but it is highly complex and its payout structure deserves some detail.
Minimum: €0 / Maximum: €0.025
Average: €0.0042 / Median: €0.0037
• €0 in cases of streaming fraud, an exceptional case but one known to The Queen Is Dead Records and therefore factually present in my dataset (with an additional “Spotify artificial charge” line and a negative value corresponding to the fine paid by the digital distributor, which the artist then has to reimburse).
• €0 when a track has not reached 1,000 streams over the previous 12 months, in which case it generates no royalties.
• €0.0000000016 per stream for Belarus (the lowest rate recorded); at that rate, it would take 1 billion streams to buy yourself a chocolatine.
The average figure of €0.0035 quoted on various websites makes no sense: this figure is unusable because there are simply too many variations. A student who scores 0, then 10, then 20 out of 20 is not simply “a student with an average of 10”: each result may reveal something distinctive that needs to be interpreted. This is a case requiring deeper analysis because each value could represent an anomaly within the trend.
The geographical location of the listener and the listener’s status (paid or free subscription) can greatly affect the payout rate. So can the artist’s level of fame: the majors (Universal, Sony, Warner) have historically signed global licensing agreements with platforms, sometimes securing guaranteed minimums, but at catalogue level, not per artist. It is impossible to trace precisely how much of these sums is allocated to artists independently of the number of streams generated. Given these parameters, it would have been interesting to calculate the average payout per stream (even though, as you will have understood, that is not the underlying basis of the calculation).

b – Discovery Mode
This is a promotional programme launched by Spotify to increase the visibility of selected tracks through its recommendation algorithms (particularly playlists such as Artist Radio, Autoplay – automatic playback after another track ends -, Discover Weekly and the user’s personalised Daily Mix). Editorial playlists created organically by Spotify’s teams are not included in this programme.
Minimum (deducted): – €0.00001 / Maximum (deducted): – €0.0023
Average (deducted): – €0.0012 / Median (deducted) : – €0.0016
Negative values? Yes! For every stream of a track participating in this programme, the royalty calculation involves deducting between 30 and 50% of the expected rate (between €0.000008 and €0.0014 per stream): tracks promoted through the programme are paid less.
If you gain streams that you would not have obtained through your own promotional strategies, that is entirely beneficial to you. However, if your fanbase listens to your tracks through Discovery playlists rather than through your Spotify profile, you may lose money. The same track can therefore have two different payout rates depending on how it is listened to: this means that joining Discovery does not reduce that track’s payout rate outside the Discovery programme. In other words, the reduced rate applies only to streams generated by this promotional programme.



11. TikTok: scaling back its ambitions

The “TIKTOK” entry includes data from two different services:

a – TikTok
TikTok’s video service has excellent figures for a video platform. The numbers are similar to Spotify’s.
Minimum: €0.00014 / Maximum: €0.018
Average: €0.0054 / Median: €0.0037
To better understand these rates, let’s compare them with those recorded for Instagram, its direct competitor:
Minimum: €0.00000009 / Maximum: €0.07
Average: €0.0022 / Median: €0.000015
You might conclude that TikTok is more generous to artists. However, remuneration is not just about the figures: the payment method matters too. TikTok’s distinctive feature is that the platform does not pay artists according to the number of plays of their tracks (in other words, the number of “views” on videos using their music), but according to the number of videos using their music. It is therefore better for your track to be used once in 100 different videos than 100 times in a single one. (Anyway… you get the idea). In the end, this model is highly disadvantageous for artists, especially since this video platform genuinely needs music in order to exist: “88% of TikTok users say that sound is essential to the TikTok experience”, and even that “it is almost impossible to separate a TikTok from its music, otherwise the video no longer makes any sense.” (Source)
Earlier, I explained that per-stream payout tables were unreliable. TikTok takes the problem one step further: TikTok’s model does not allow a stream to be associated with a payout value, because what is counted is the number of videos created using the music. Millions of listens can therefore generate only a handful of euros.

b – TikTok Music
A service that aimed to compete with Spotify.
Average: €0.0028 / Median: €0.0031
Minimum: €0.0016 / Maximum: €0.0036
TikTok Music was launched on 7 July 2023 by ByteDance, the group that owns TikTok. In 2020, the group had launched Resso, an app based on a similar concept and available only in India, Indonesia and Brazil. A way to learn the ropes. TikTok Music was an ambitious project intended not only to replace Resso but also to expand into new territories such as Australia, Singapore and Mexico. TikTok Music was shut down after just over a year, on 28 November 2024, so resources could be redirected towards new projects.



12. Youtube: the roulette wheel

The “YOUTUBE” entry corresponds to 3 entities:

a – Youtube Content ID
This is the automatic system used to detect copyrighted content in uploaded videos.
Average: €0.005 / Median: €0.0003
Minimum: €0.00000004 / Maximum: €0.08
This is not strictly speaking a per-stream payment, but rather third-party use of your music in a monetisable video, with remuneration proportional to viewers’ exposure to advertising. For Premium subscribers (paying users) who do not see advertising breaks (100 million out of 2.5 billion users, or 4%), the share of their payment allocated to Youtube is distributed according to relative listening time.
In theory, under a user-centric model, if a paying subscriber listened to only one artist during the month, that artist would receive 100% of the relevant share of the €12.99 subscription fee, or around €9 (Youtube keeps a 30% commission). But the model is market-centric: subscription revenue is pooled and then distributed in proportion to streams (weighted by content type, territory, free/paid subscription and other parameters that are not always clearly known). The payout amount is therefore relatively unpredictable.

b – Youtube Shorts
The short vertical-video service, competing with TikTok and Instagram Reels.
Average, median, minimum, maximum: €0.000007
Although the figures collected appear to indicate a stable payout rate, the reality is that music is paid proportionally according to its presence in videos. Until 2022, rights holders received no remuneration for the use of their music in “Shorts”. Since 2023, part of the advertising revenue generated by the service has been pooled and redistributed between video creators and music rights holders.
As with Youtube Content ID, part of the advertising revenue generated by the service (after Youtube’s commission has been deducted) is pooled and redistributed between video creators and music rights holders. Here too, the amount of remuneration cannot be predicted in advance.

c – Youtube Music
This is Youtube’s least-known service: a music streaming service similar to Spotify, Deezer or Apple Music.
Average: €0.005 / Median: €0.0028
Minimum: €0.000006 / Maximum: €1.07
As with Spotify, remuneration depends on the number of streams and on revenue generated by subscriptions or advertising. As with every streaming service, the value of a stream varies. The estimated average sometimes found in online searches, ranging from €0.006 to €0.008, is overstated.



13. Facebook: Youtube, but slightly worse

The “FACEBOOK” group consists of 2 entities:

a – Facebook (Facebook Monetization)
Also known as “Facebook Monetization“, this service is Facebook’s counterpart to “Youtube Content ID” for videos created on Facebook using audio contained in Meta’s database (Meta Music Library).
Minimum: €0.00000006 / Maximum: €0.07
Average: €0.0027 / Median: €0.000016
When your music is used on Facebook or Instagram (in a Story, Reel, video, etc.), Meta can monetise that content by displaying advertising around or within the video (a video not monetised by its creator). This may take the form of a mid-roll ad, a pre-roll ad or an overlay banner. The beneficiaries of this advertising revenue include Meta, which keeps a share (often around 30%), and the music rights holders (you, your label, distributor, publisher). Note: if the content creator is an “eligible” partner, they can also receive revenue through the “Facebook Revenue Share” programme.
Several factors affect earnings: how long the music is used, the number of views on the video – and therefore the number of ad impressions -, the geographical region (for example, an ad in the USA generates more than an ad in India), and the type of advertisement (earnings vary from one ad to another).

b – Facebook Revenue Share
Whereas Facebook Monetization (“Content ID”) provides automatic remuneration, the “Facebook Revenue Share” programme applies to longer monetised videos made by partner creators.
Minimum: €0.0002 / Maximum: €0.04
Average: €0.0054 / Median: €0.0016
For each video, revenue is split three ways: Meta receives 30%, the video creator 55%, and rights holders 15%. If music by several artists is used in the same video, that 15% has to be shared between all the relevant rights holders. This arrangement may appear less attractive to musicians than Facebook Monetization, but the Facebook Revenue Share programme applies only to videos generating a substantial number of views.



14. Qobuz: blind spots


The starting point for this article was an ad I came across on Facebook: it claimed a payout of €18.02 for 1,000 streams, or €0.018 per stream.
Here is what my data shows:
Minimum: €0.0046 / Maximum: €0.82
Average: €0.020 / Median: €0.016

The average and median seem to confirm the general range of €0.018, but several blind spots make it impossible to draw any firm conclusions.

a – Streaming and download data are mixed together
Qobuz also sells downloads, and the revenue reports I receive combine both types of activity. This distorts the average “per unit” if the aim is to isolate revenue per stream. The same problem applies to Apple and Amazon: my data does not distinguish Apple Music from iTunes, nor Amazon Music’s download offering from its streaming service. This makes it impossible to clearly separate streaming from sales and therefore to analyse payout trends accurately.

b – A market-centric remuneration model
An article published in March 2025 confirms that Qobuz uses a market-centric model, meaning revenue is distributed in proportion to overall listening activity.
1% of streams = 1% of total revenue
But this calculation depends on several variables:
– the type of subscription (Classique or Sublime)
– the total revenue generated during the month
– the total number of streams
These factors make the existence of a fixed per-stream rate impossible. And while the model is mathematically proportional, it is socially unequal: niche artists remain at a disadvantage even when they have a loyal audience, because their listeners have no control over how their subscription money is distributed.

c – Extreme differences between minimum and maximum payouts
A payout ranging from €0.0046 to €0.82 suggests that a significant share of the data probably comes from downloads, whose prices vary. For example, an album may sell for €18.99. With a Sublime subscription costing €199.99 per year, discounts of up to 60% may apply. This means the customer pays only 40% of the original price. For the subscription to pay for itself, the user would need to save €200. Since a 60% discount means saving €0.60 for every euro spent, they would need to purchase around €333 worth of albums at full price, or roughly 30 albums priced at €18.99 each. Not every subscriber will buy 30 albums a year. Revenue will therefore fluctuate not only according to the number of Sublime subscribers, but also according to how many paid downloads they purchase. In a market-centric system, these fluctuations add another layer of variability to an already unstable model, making any estimate of a fixed value per stream even less reliable.

d – An advertised rate whose stability is unknown
We do not know how long the €0.018-per-stream figure has been in use. It may be recent, experimental or not applied universally. If this figure had been stable for a long time, Qobuz would presumably have communicated about it before. Yet, to my knowledge, no previous campaign had highlighted such a figure.

e – The digital distributor: a grey area in the chain
Before the money generated by streaming platforms reaches the artist, it passes through an unavoidable intermediary: the digital distributor. This is one of the first sources of variability in the amounts ultimately received.
The exchange rate also plays a significant role. Depending on where the distributor is based, the reference currency may differ – USD (US dollar), EUR (euro), AUD (Australian dollar), CNY (yuan), etc. The final amount received by the artist may therefore fluctuate depending on when funds are withdrawn or converted.
But another question arises: how transparent are distributors really? Artists are often forced to trust their distributor when it says: “Here is the full amount of revenue we received on your behalf.” Yet it cannot be ruled out that streaming platforms may treat different partners differently, with some distributors potentially receiving more favourable payout terms than others. In an ecosystem where the data is rarely verifiable from end to end, every possibility remains open. And that only reinforces the need for transparency at every stage of the chain.



By communicating a precise figure (€0.018 per stream), Qobuz seemed to be opening a new era of transparency in the music industry. Unfortunately, on closer inspection, that figure exists within an opaque system. But Qobuz is not the black sheep of the industry. This study confirms just how far reality is from the average-value charts that circulate here and there across the internets: all the figures you will find are… wrong or, at best, highly debatable.
The data collected by The Queen Is Dead Records, and analysed here as transparently as possible, shows that remuneration per stream varies considerably from one organisation to another, and even within the same organisation (as demonstrated by the enormous gaps between some minimum and maximum values). The actual rules range from unclear to completely unknown. At best, the publicly presented rules are incomplete, and no notification is issued every time a parameter or structural element of the remuneration system changes. Everything remains abstract, leaving room for potential fraud (and conspiracy theories, Spotify, Deezer, Apple Music and the others would tell you).
As long as artists continue to take the supposed “per-stream values” circulating online at face value – often without reliable sources, but stated with remarkable confidence – nothing will change. If we, as artists and labels, stop treating these figures as universal truths, we may finally be able to demand genuine transparency from the platforms. Today, no detailed information is available, and we are still navigating largely in the dark.
Disappointed?

L.A

Leave a Reply