spotify

In 2022, Will Page, the former director of economics at Spotify, encouraged a U.K. committee looking into streaming economics to consider how collecting societies have divvied up fixed pots of cash for more than 100 years. A fairer system for paying royalties, he said, might consider how long a person listens. 
Page’s suggestion wasn’t a new, radical idea. Other royalty accounting systems already take listening time into account. In the U.K., collection societies such as PRS For Music and PPL apply a “value per second” rule to royalty payouts. So, Page explained, Queen’s “Bohemian Rhapsody,” which clocks in at 5:55, earns twice the royalty as “You’re My Best Friend,” which runs just 2:52. A similar approach is codified into U.S. copyright law: Songs over five minutes long receive a higher mechanical royalty than shorter songs.  

But streaming platforms have long paid royalties using a “pro rata” method that treats every song equally. At Spotify, for example, any two songs by Queen are treated the same. But there has been a movement in recent years to make royalty payments fairer to non-superstar artists. SoundCloud adopted a user-centric approach that pays royalties from each listener rather than pool all listeners’ revenue. Deezer has a “user-centric” approach — adopted by Universal Music Group, Warner Music Group and Merlin — which rewards professional artists at the expense of “functional” music. 

Trending on Billboard

Two years after testifying to the committee, Page has released a paper, “A Case for Completion,” that outlines how streaming platforms could reward songs that get streamed in their entirety. The idea is simple: For each stream, the streaming service asks whether the song was streamed to completion. If the song was skipped before the listener got to the end, a portion of the royalties are transferred to songs that were streamed to completion.    

The financial model looks like this: Labels earn about 50 million pounds ($64 million) for 10 billion streams. Page estimates that 10% of the songs will not be streamed to completion. Of those songs’ 5 million-pound ($6.4 million) royalty pool, 40%, or 1.3 million pounds ($1.7 million), goes to the completed songs’ royalty pool. That in turn increases the completed songs’ pool from 45 million pounds ($58 million) to 46.3 million pounds ($59.6 million). On a per-stream basis, a typical 0.0048-pound ($0.0062) pro-rata royalty becomes either a 0.0035-pound ($0.0045) incomplete royalty or a 0.005-pound ($0.0064) complete royalty.    

Importantly, Page believes this completion-based scheme complements the current royalty accounting system, whether it’s pro-rata, user-centric or artist-centric. “If we are going to depart from the pro rata model, which has served us since Rhapsody got its license in December 2001 — which is 23-plus years ago — then we need a baby step that doesn’t mess with royalty accounting,” says Page. Tracking duration would add too much stress to a royalty accounting system that encompasses trillions of streams annually, accounting experts told Page. In contrast, setting a threshold that creates a binary outcome — either a song was completed, or it wasn’t — is more feasible, he argues.  

The proposal may run into naysayers who believe skipping is a critical aspect to streaming. On-demand services with hundreds of millions of songs charge for the right to skip through playlists and algorithmically created radio stations. In contrast, free, non-interactive streaming services such as Pandora don’t allow unlimited skipping. What’s more, decidedly unskippable formats such as terrestrial radio are losing listening time to platforms that give the listener greater freedom. Whether TikTok has reduced attention spans or listeners are impatient in a world of unlimited choice, skipping is simply a way of life in 2024.    

But skipping, however prized by today’s music listeners, isn’t necessarily rampant. As Page explains in an interview with Billboard, he gained confidence in completion-based royalty accounting after learning that completion rates surpass 90% once a person has been listening longer than three minutes. To Page, this means shorter attention spans select shorter songs and people willing to listen longer will do so. “Sprinters enter sprints; marathon runners enter marathons,” says Page. “For the most part, people who want longer songs go for longer songs and stay the journey. Jazz and classical have got the highest completion rates from all the genres.” 

Paying based on completing a song makes sense intuitively, because in streaming the business goal is listener engagement, and one sign a listener is engaged is how much a song gets heard. From that perspective, a stream that ends halfway through a song is less valuable to both the streaming platform and the rights holders than a song that somebody listens to all the way through. So, rewarding completion makes sense from this business point of view. 

It does. And I think a key strength of the proposal, and I’ve road tested it with the great and good in music and tech — I’m very open on strengths and weaknesses and anomalies. I’m putting all my cards on the table here for this to be accepted and be a model to give people even more assurances. But the strength is it’s asymmetrical. I am not promoting completion. If Glenn Peoples does nothing with this listening experience, I do nothing with these royalty calculations. I must be absolutely clear here. I am only punishing incompletion. I take action when you show intent. If you do nothing, I do nothing. If you step in there and say, “I’m done with this song, move me on to the next one,” I’m going to do something with the royalty structure. That’s crucial in terms of the argument. It’s got a strong common-sense property, as you alluded to, but it’s asymmetric. And to be absolutely clear, streaming services don’t pay a penny more or a penny less. We simply reallocate away from the incomplete pool to the complete group.  

The deterrence against fraud or gaming the system, whatever you want to call it, seems to be a strong argument. If some artists are making music based on this 30-second threshold, I don’t see how that’s good for anybody. The royalty model shouldn’t be influencing how music is created and released.  

Drake had an album where there were like eight songs which lasted between 40 and 50 seconds — skits — and they’re going to get paid the same as a seven-minute jazz composition with McCoy Tyner? These are questions of fairness. The current model has unfair properties in it as well. We have to remember [that] nobody thought about jazz and classical when they invented the 30-second rule. [An on-demand stream earns a royalty if it is streamed for 30 seconds or longer.] Nobody argued for duration.   

Now let me allow me to play Devil’s Advocate. As a user of a subscription service, I pay for the ability to skip songs. And if I skip a song 45 seconds in, it doesn’t necessarily mean that song is less valuable. It means that I enjoy that ability to skip songs. If I don’t want to skip songs, I’ll listen to SiriusXM. And the ability to skip songs is one of the best things about an on-demand service. So why should skipping be punished if it has so much value to me? 

I respect that view. I would say that argument is weak because the majority of people are paying for the concierge service. In the vast majority of instances, the act of skipping is a negative signal by the consumer. And for a lot of people, the engagement they have with their music platform is approximately this: in the pocket it goes and that’s it for the day. I’m not paying so I have to skip songs. I’m not paying so I have to select songs. I’m paying to enjoy the music. If you can serve it up for me, I’ll pay, I’ll stay even longer. So I quote [intellectual property expert] David Safir in a piece where there was a heated debate at the NY:LON conference in London. David calmed the debate down by saying, “Hold on, we haven’t even decided who we’re defining fairness for. Is it the creator, the platform, or the consumer?” As the consumer pays for convenience, the act of skipping, or the act of even leaning in, could be a sign of inconvenience. That is negative for the consumer’s experience in terms of willingness to pay and willingness to stay.  

When I skip, it’s to sample the big catalog of music. It’s one way to listen to more music — not all of which I’m going to go back and listen to again. But at least I hear it. Again, whether it’s an editorial playlist, or just bouncing around the app, skipping allows me to sample the catalog. And not skipping would really get in the way, I think.  

I remember with [Spotify’s] Discover Weekly, we began to wonder whether the reason it was successful is you used to spend a bit of your time searching for music that could involve a lot of skipping, and a bit of your time consuming music. And as time became more precious, you didn’t have any time to search. Nobody went to record shops anymore, and therefore there was even less time to consume. And what Discover Weekly did was internalize the search cost, the experimental costs, the skipping costs, and it gave you exactly what you needed. In terms of what pays everyone’s bills in this business, it might be the skipping — I doubt it. It might be the searching — I doubt it. I think what drives it is I just pull out my phone and it delivers me music and I stay the course. I think it’s that.  

The [U.K. Competition and Markets Authority] asked the four streaming platforms in the U.K. to reveal a source of streams and just how much is human editorial: not a lot, 5% back then, probably two and a half percent now. How much is algorithmic? Not a lot. The vast majority of listening is people-owned playlists. That was a bombshell. That shook the industry out of a rut because, wait a second, 85% of listening might not be platform directed.  

So, you know, it’s interesting to just think about that context as well. If you’re skipping, and you look at that table, you look at all the evidence, I think that the evidence weighs towards skipping as a negative signal in terms of the attribution, the value, utility that person’s gained from their platform, as opposed to a positive one. People want to stay in the saddle of music. They want to complete. 

Reading the paper, I sensed some undercurrents, perhaps, of criticism of how people, especially young people, listen to music these days. You quoted somebody saying that wedding bands only play two minutes of a song because TikTok has ruined its users’ attention spans. Is part of this about trying to get people to listen to an entire song, and get their attention spans back? 

I really owe a long-time mentor of mine, Fred Goldring, for that quote. He told the story about a wedding band that played a two-and-a-half-hour medley because people don’t have the attention spans for full songs anymore. I was like, “Oh, my goodness! What has TikTok has done? Is that what the 30-second rule has done to our music? Is that where we’re at?” If I can expand on that, Arctic Monkeys are a very successful band. They played the Emirates Stadium [in London] twice last summer. The first night was predominantly die-hard fans in their 40s and 50s. The second night was teenage girls who had discovered them on TikTok, and they only knew 34 seconds of all its songs. If you stick around after the chorus, we’re going to sing another verse. It’s called a composition, people; we’ve had these things for a long time. Yeah, there is a concern there.  

Now, the concern could just be misplaced. I think the concern is actually very real. Songs are getting shorter. Choruses have been moved to the front, and Swedish artists were doing this in 2013. Many artists are doing it now. But in an attention economy, any alteration to pro rata [royalty calculations] that helps music win attention, that creates incentives that compete for attention, has to be good. Because music is in competition with so many other distractions. Now, completion has a different agenda, but it’s going to help this industry think about, how does it compete for attention? 

You noted in the paper that complexity could be the opponent of a successful royalty system. I’m wondering to what extent people, and mainly creators, will need to understand how this royalty system would work. You’ll understand it. Attorneys will understand it, as they must. But ostensibly, these new royalty schemes are to create more fairness for creators. Do you think creators would understand this well enough? 

Is the consumer aware that under pro rata, that if I’m a light user, and Glenn Peoples is a heavy user, my money is being used to compensate Glenn’s consumption? Probably not. If they were, would they change your habits? Maybe. Maybe that user-centric property is interesting. But I’m not sure how interested the consumer is in the actual royalty model. If you surveyed them and said, “How many people know it takes 30 seconds before you get paid?” Less than 1%.  

On the industry side, something as simple as a completion index, a third threshold, I feel fits the curve. Even drummers will understand this. That’s really important. Now, where it could get complex in that proposal is that Glenn’s completion of a two-minute pop song would be worth more than my incompletion after listening to six and a half minutes of a seven-minute song. Curb the concern, though, because I did go on to show that genre is not necessarily a driver of completion; neither is song length. That’s a reassurance.