A Tale of Two Cities: What Sweden and the US Tell Us About the Outlook for Streaming

Streaming is the digital zeitgeist, that much is clear.  How it will shape the future of the music business, from artists through to labels is less clear and things are not helped by an increasingly confusing and diverse set of data, each suggesting a slightly different outlook.  A look at two very different digital music markets – Sweden and the US – gives some sense of what the next couple of years should hold.

Notes: for sake of readability the term ‘streaming’ is used to refer to subscriptions and ad supported streaming combined. Also all current year figures are 2013, extrapolating half year figures to create full year estimates.

Two Very Different Streaming Stories


Sweden is streaming’s heartland, home of Spotify and the stand out good news story for music subscriptions. Streaming now represents a whopping 95% of digital revenue in Sweden and 67% of all recorded music revenue while downloads make up a paltry 4%.  Streaming growth has been equally impressive (see figure one) and has propelled the total Swedish music market into growth for two successive years.  That growth came at the direct expense of downloads (which declined by 15%) and it accompanied a dramatic 51% collapse in CD sales.  But 2013 revenues look set to come in at just a little below 2003 levels, no mean feat.   Although we need to bear in mind that a surge in growth can easily reverse (as the experience of South Korea shows us) it is clear that streaming has been a strong positive force on total Swedish music revenues.


The picture is very different in the US however, where streaming has grown less dynamically and only represents 23% of digital and 14% of overall spending.  As I previously noted, the strength of Apple and the download sector have acted as a pronounced brake on streaming growth in the US.  Neither, however are invincible, and some of Spotify’s 2013 growth has come at the direct expense of download spending which looks set to decline by a percentage point in 2013 (see figure two).  Little wonder Apple has launched iTunes Radio, though ironically the app may well spur a resurgence in download sales.  So in the US streaming is becoming an increasingly important part of the market but shows no sign of suddenly acquiring Sweden-like ubiquity.  Which in part explains a 5% decline in total music revenues between 2010 and 2013.

CONCLUSION: streaming can quickly drive strong growth in markets where downloads never got a foothold but takes more time to impact strong download markets.

The Impact on Total Digital Revenue

Streaming’s impact on the total digital market and indeed on total music sales is of course what counts most, and it is here we see a really interesting divergence between Sweden and the US. Over the last 6 years streaming drove a comparable rate of overall digital growth in Sweden that downloads powered in the US in the mid 2000’s.


But when we plot the growth of digital as a percentage of total music sales in the US between 2005 and 2010 against the same data for Sweden between 2008 and 2013 a stark contrast is immediately apparent (see figure three). Whereas digital share growth remained strong throughout the 6 years in Sweden it slowed markedly in the US.  Though growth returned later it didn’t ever replicate those pre-2008 levels.  The number one slowdown factor was the end of iPod sales growth (see this figure to see just how strong the effect was).  Interestingly digital share growth looks likely to slow moderately for both Sweden and the US in 2013.  In Sweden some level of slowdown is to be expected (there isn’t much physical market left to transition!) but there is still a lot of CD ground to be made up in the US.

CONCLUSION: streaming has driven market growth in Sweden and accelerated transition away from the CD and the download. While in the US the CD and the download both still hold much greater sway, culminating in something of a worst of both worlds, with streaming apparently eating into downloads but not having enough headway to transform the market.

The Artist Conundrum

But what does all this mean for artists?  It often feels that something doesn’t quite seem to add up when artist income is brought into the equation. For all the growth in streaming income, a vocal minority of artists and songwriters feel that streaming is damaging, destroying even, their ability to earn a living from music sales.  As I have argued before, a rounded understanding of streaming income for artists must both put streaming in a revenue continuum (i.e. compare it to radio not just downloads) and consider the life time value of a song (i.e. think of the income it will generate over a period of years instead of the revenue full stop a download represents).  In this context streaming is still worth less than a download, but nearer to 5.5 times less valuable rather than 280 times (see my Consumption Analysis piece for more on this).


There is however an added complexity, namely the amount of artists that get revenue from streaming versus downloads and streaming (see figure four).  If we take Spotify’s reported US metrics from 2012 as a benchmark and assume that the average subscriber listens to a modest 5 different artists a month then this is equal to 60 different artists per year per subscriber.  Working with an average total royalty pay out of $0.01 per stream this translates into an average royalty per artist per subscriber of $0.72 in the US.  When applied to the 3 million reported US Spotify subscribers this would equal an average annual royalty of $2.17 per artist.  (Though it is crucial to note that this refers to the total royalty payment made to rights holders and not to whatever share is eventually shared with the creators themselves). Also, there is of course no such thing as an average artist, and in practice a comparatively small number of artists would earn much more than that and most much less (there are after all 27 million tracks’ worth of artists so the tail is super long).

For downloads, extrapolating from Nielsen mid year numbers, the average downloader buys 2 albums and 27 single tracks.  If we assume each of these is for a different artist then we end up with 26 artists per downloader and an average royalty of $1.22 per artist per downloader (using a 70% royalty assumption).  This isn’t actually that much higher than streaming, but things change when it is applied to the total number of download buyers (which at 63 million far outstrips paying subscribers) and results in an average royalty per artist of $76.34 (again total royalty before distribution to creators).

In Sweden though, where there are more subscribers than downloaders the picture is very different.  Applying the same Spotify metrics to an assumed subscriber base of 2 million in Sweden (which feels about right based on survey data and IFPI numbers) we see an average royalty per artist of $1.44 compared to $1.22 for downloads.  (The average royalty per buyer is higher in Sweden because a smaller number of people are buying a smaller number of downloads resulting in the revenue being split fewer ways).

CONCLUSION: streaming can generate meaningful revenue at scale but will still be lower than downloads because of the above mentioned life time value factor and because revenue is split more ways across a wider selection of artists.

The Cost of Democratization of Artist Income

Thus artists are effectively paying the price for the democratization of music: more artists are getting listened to more regularly and as a consequence the pie gets cut into smaller slices. Which raises the interesting dilemma of whether artists speaking out against streaming are also indirectly speaking out against a more equitable distribution of income among artists?!  The core question though is whether the pie can get large enough for those slices to represent anything more than an apetizer for the average professional artist.

All of this extra data may appear to add as much fuddle as it does clarity to the debate, but it is crucial that debate is based upon reasoned understanding of the most complete grounding of data available.    The next couple of years will see streaming go from strength to strength but its impact on global music revenue will be less dramatic than it has been in Sweden, if perhaps more vibrant than it has in the US.

13 thoughts on “A Tale of Two Cities: What Sweden and the US Tell Us About the Outlook for Streaming

  1. I don’t understand how you get the figure of $2.17 average royalties per artist. I can get your figure of $0.72 per artist per subscriber (13 billion streams divided by 3 million subscribers gives 4333 streams per subscriber; 4333 divided by your assumed number of 60 artists listened to by each subscriber gives approx. 72 streams per artist, and multiplying by $0.01 gives $0.72), but I don’t see how you get from that to $2.17. I can see that $2.17 is approx. 3 times $0.72, but what basis do you have for multiplying by 3? Your graphic does indeed give the number of US subscribers as 3, but that is short for 3 million! Am I missing something? I don’t see any way to get an average royalty per artist without knowing the total number of artists on Spotify, which is not stated. It is probably a very large number, which would give a low average royalty per artist, but the average is pretty meaningless when the tail is so long and thin.

    Sorry to nag you (again!) about your figures, but it is important to see the basis for your conclusions, which are all that the casual reader is likely to come away with.

    I’m inclined to agree that with streaming the total revenue is likely to be spread across a larger number of artists, but I don’t know of any firm evidence for that, and it doesn’t follow necessarily from the assumption that each individual listener is listening to more artists than if they bought CDs or downloads. Consider the simplest possible example: 2 listeners, A and B, and 2 artists, C and D. Without streaming, A buys an album by C, and B buys an album by D. With streaming, A and B both listen equally to C and D. Assuming that total spending is the same, the revenue of C and D is unchanged.

    Assuming that listeners are indeed listening to more artists, I still don’t see that artists are ‘paying the price’. Obviously if some of them are getting a larger share of revenue than before, and total revenue is constant (which the streaming advocates would dispute), it may be that a large number of artists are benefiting at the expense of a small number of pop superstars, which would be no bad thing for music in general.

  2. “The core question though is whether the pie can get large enough for those slices to represent anything more than an apetizer for the average professional artist.” – This sentence here is pretty much the real unadulterated crux of the matter. There are different ways in which the contributors to the size of the pie can be characterized – higher payout per stream, more subscribers etc. Unless the pie gets sufficiently large – which I personally don’t see happening – the artists will suffer. It is not that the rich (popular band) get poorer (receive lesser plays because other bands are getting their share), it is that the poor (startup bands who are apparently picking out a small share off of the popular bands) will never ever get a chance to see any REAL riches (It is still a small share).

    Also, as far as the artist is concerned, the ONLY THING that matters (with regard to payments) ultimately is that number on the check that they are getting – the TOTAL INCOME from ALL ROYALTIES. Is it higher than what they were getting before? Over the period of a year, is the TOTAL money they are getting from streaming higher or lower than what they were making with download and CD sales before?

    What DOES NOT matter to the artists is if they are apparently getting a higher ‘share’ of the receipts (as pointed out in this post). This ‘higher share’ becomes completely irrelevant – to the point of being misleading – if the artist is making LESSER TOTAL money. What also does not matter to the artists’ income is how a stream’s value is rated against a download. The choice of the metric that is used to show the value of a stream against a download – one way or the other – is completely pointless if they are receiving LESSER TOTAL money from all royalties.

    There is going to be some TIPPING POINT in the future unless the business models of both the subscription services and the record companies are revised. What is on the other side of the tipping point is anyone’s guess.

  3. David I always value your insightful contributions and probing questions. Here are my responses:

    – you are correct in your assumptions about how the royalty per artist per subscriber is calculated. The average revenue per artist is also exactly as you say, calculated by multiplying the number of paying subscribers (with the average per subscriber already accounting for number of streams per subscriber per year per artist). This is not meant to represent a definitive measure of how much each artist gets paid but instead a directional indicator of an average that allows us a like-for-like comparison with downloads because we apply the same approach there. (Though the extra life time value lens also should be considered for sake of completeness).
    – this is a bottom-up, not top-down analysis. Therefore it does not attempt to estimate wheat share of 27 million total tracks are played / bought on download stores or streaming services. Instead it assess the total number of artists played per artist. Therefore, and this is crucial, a single artist’s income (for both downloading and streaming) will compromise this income from multiple subscribers.

    The intent here is to create comparable measures that give a sense of the different effect of each digital revenue stream, not a definitive measure of income for artists

  4. Akshay – thanks for your comment. You raise a lot of valid points. However they were beyond the remit of this post. I was already at over 1,200 words and had to try to keep the focus tight. The issue of the amount of money that makes it back to artists its something I have discussed at length elsewhere and will cover in detail in my forthcoming book. It is also complicated by a vast number of variables. But the purpose of this blog post is to create an extra measure of how streaming and downloading stack up against each other in directly relative terms. From there artists and songwriters have a further piece of evidence to help assess how much their label and publisher deals are impacting the difference in income they get from each source.

  5. Interesting post, Mark. How much, though, do you think that the average US consumer’s streaming habits have been formed by access to free, ad supported, DMCA compliant streaming radio like Pandora? I would imagine this has affected overall market growth and also skews the numbers in comparison to Sweden and Spotify.

    Anecdotally I hear that Vodafone are struggling to communicate the value of Spotify to their UK 4G customers, so I can imagine that given the different evolution of services in the US, on-demand subscription streaming could be an even harder sell over there.

  6. Very good point James – yes almost certainly a big additional factor, though we do see the download brake effect in markets without a big internet radio player. Interestingly it is looking like a few Lat Am markets are beginning to get transformed by the arrival of recent free streaming tiers, and not necessarily in an entirely good way either.

  7. I still don’t understand the numbers, but I won’t pursue that.

    I do just want to stress again that there is no clear evidence for a ‘more equitable distribution of income’ among artists. Even if with streaming each individual user listens to more artists than previously, it does not follow that total revenue at the population level is more evenly distributed than before among different artists or different genres. Suppose that 10% of the population previously bought and listened mainly to country music records. With streaming, quite possibly a larger proportion of the population will sometimes listen to country music, but by the same token the country fans may now devote more of their listening time (and revenue) to non-country genres. There is no way of determining a priori where the balance of gain and loss falls. Streaming could even lead to greater concentration of revenue in the pockets of the most commercial pop artists, since these are the ones that are constantly in the mass media. I would not dream of buying a Miley Cyrus record, but I am willing to check her out on a streaming service just to see what all the fuss is about.

    Having said all that, there is a plausible argument that ‘other things being equal’ streaming should produce a more even distribution. If we divide the listening audience into two groups, a majority who listen mainly to ‘mass market’ music, and a minority who listen mainly to ‘niche’ genres of music, then on the ‘other things being equal’ assumption that each group shifts the same proportion (say 10%) of its listening (and revenue) to the other category of music, then the minority category gains more than it loses. Suppose the majority is 80% and the minority 20%, then a 10% switch in both directions gives the minority genres a net gain of (10 x 80) – (10 x 20) = 6%, while the mass market genres have a net loss of (10 x 20) – (10 x 80) = 6%. But the assumption of ‘other things being equal’ is obviously a hazardous one.

  8. Good points David. Perhaps a more grammatically correct explanation would be ‘more widely distributed’ rather than more equitably distributed. i.e. there will still be a key concentration among the head rather than the tail, but across a broad swathe of the catalogue (probably mainly in the body i.e. middle) that income will be distributed among a greater number of artists. If so this mid-tier of artists who already have pressure on their income could well see the paradoxical situation of more people paying to listen to their music than ever before, but ending up with less income than ever before, while at a macro level revenue grows.

  9. This is a fascinating analysis, Mark. Do we know what the album life cycle looks like on streaming services? It seems to me that if we are going to compare streaming revenue to download revenue, we need to discount the streaming cash flow (using a simple present value formula should suffice).

  10. Streaming is the future for music. It’s a great way for artists like myself to get noticed. I signed up with Ourtunez new mobile app and I found out that they shuffle my independent music with mainstream music. It’s positive for both the listener and the artist.

  11. Pingback: ipods prices A Tale of Two Cities: What Sweden and the US Explain to Us About the Outlook for …

  12. Pingback: A Journal of Musical ThingsThe Recording Industry Issues First Look at 2013 Sales. It Ain't Pretty » A Journal of Musical Things

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s