Author: Bas Grasmayer

Could virtual music stars reach millions of people without doing so through communal experiences? It’s something we’re bound to find out in the next decade.

Virtual stars are not a new thing. Vocaloid avatar Hatsune Miku has been around for over a decade and opened for Lady Gaga in 2014. Another type of avatars are virtual YouTubers who have been around for about half a decade. Arguably the most well-known avatar of that group, Kazuna AI, recently performed at Porter Robinson’s online Secret Sky Festival. Chinese streaming giant iQiyi’s research into virtual idols has shown that 64% of people aged 14 to 24 in the country follow one or more avatar stars.

Thus far, the music stars in this area of entertainment have followed a similar strategy to real-life pop stars: getting on a (virtual) stage and singing to many fans. An AI-powered chatbot named Xiaoice shows the potential for another approach.

Xiaoice is a virtual assistant designed to keep users engaged by forming deep emotional connections to them. The result? Over 600 million people have tried the bot out over the past years with many of them becoming hyper-engaged. The company, spun off from Microsoft, estimates that “half the interactions with AI software that have taken place worldwide have been with Xiaoice.” The AI’s fans are 75% male and reading about some of the ways they describe interacting with her may remind one of the movie Her.

Xiaoice is a type of mass media to which each user feels a personal connection, because instead of one-to-many communication (like pop stars), the virtual friend is able to have one-to-one communication at scale, much like Siri and Alexa. The communication model it uses is hard to apply for a human being who can only do one thing in one place at one time, but with improvements in AI and the increasing virtualization of music the one-to-one model is easier to utilize by the music business.

I’m not aware of a one-to-one virtual music idol at scale (do leave a comment if you know of something), but I have no doubt that something like this will emerge over the next years, no doubt using channels like TikTok as part of a user acquisition funnel and becoming part of mainstream consciousness that way.

Interestingly, I think this discussion mimics one of the cultural discussions we are having around music right now as exhaustion sets in trying to attain higher streaming numbers for unsatisfactory incomes. So as musicians explore models beyond extracting small amounts of money from giant numbers of listeners — such as membership models like Patreon — we may see virtual idols mimic the cultural shift in music, but apply the dynamics at a scale that non-AI creators can’t employ.

As long as that doesn’t lead to a situation where AI idols start cannibalising the membership revenues of non-AI performers, that scale may not be an issue beyond the fact that the bargaining power may distort other markets that musicians rely on for income. While the next year’s theme will be a return to normal, the next decade will stretch and upend the meaning of normal.

The year is 2023. After intensive lobbying, the governments of the US, the EU, and Britain have decreed changes to the way music streaming services pay out artists. Instead of pooling all of users’ revenue together and distributing it pro-rata based on the number of streams, the money will now be divided based on listening-share per user. That means, if a listener’s playback is composed for 50% of one artist’s music, that artist will get 50% of the revenue that specific user generated.

With new economics, the game changes – and so does the way we measure success.

Why user-centric is a paradigm shift beyond the economics

A common argument against the user-centric model is that it doesn’t necessarily bring more money in, would be costly to transition to, and might not end up in radically different sums at the bottom of royalty statements. The problem with the studies that these arguments are based on is that the studies took data generated by the current status quo of streaming and ran it through a new system. A different system means different incentives and would have led to a different data set.

More importantly, it changes the overall game as Spotify’s then-chief economist Will Page pointed out in 2018:

“Under ‘pro rata,’ an artist is less concerned with diversity and will simply prefer the platform whose users stream their music most. Under ‘user-centric,’ artists prefer streaming platforms where their listeners exhibit less diversity in taste.”

The game in user-centric streaming is to make sure your listeners spend as much time on you as possible. As an artist, you want to make sure they’re a) paying for a subscription, and b) listening to you primarily. What you don’t care about is their exact number of listens.

A familiar metric will stay just as important: your number of monthly listeners. What will become more important, is a new metric: average listening-share per user (ALPU).

ALPU: average listening-share per user

Fast-forward to imaginary 2023 again. Music marketeers are obsessed with ALPU, which they can monitor in the dashboards of popular streaming services. Some of the tactics they use to increase ALPU are:

• Getting fans to turn disable Autoplay features on streaming services, since it’s better if playback stops rather than moves on to another artist.
• Prioritizing profile-focused platforms over catalogue & discovery-focused platforms. While getting discovered is still important, the money is made when you get people to stream on high-ALPU platforms.
• Move fast on new platforms, since being first on a new service is a massive advantage for ALPU due to the fact that there’s not much else to stream yet.

It would slowly lead to a landscape that looks more like that of newsletters and Patreon communities in the sense that they’re isolated and there’s not much discovery between them. As a matter of fact, Patreon stated they’re not a discovery platform up until a recent funding round. Their now-deleted blog post was cached by the Wayback Machine and can still be read. In it, their then-SVP Product Wyatt Jenkins said:

“Because we’re not focused on discovery, we’re not constantly trying to recommend new creators to your fans. It takes time to build an audience, for a fan to become passionate enough about you as a creator to become a paying patron. We respect that relationship. We’re not here to market other creators to your patrons.”

While a user-centric payments landscape doesn’t exist yet for music steaming, it’s useful to think about what it would look like. While I think something healthy can flourish from such a landscape, it’s not a cure-all for today’s most important problems. For example, a shift to user-centric is among the demands in the Justice at Spotify campaign by the Union of Musicians and Allied Workers, because they claim that pro-rata “puts artists in competition with each other.” I suspect a user-centric economy would make that competition much more fierce.

Finally, ALPU already makes for a useful thought experiment at present. Before you get to revenue, you have to garner attention. What’s your attention share of your most dedicated fans? How do you stay top of mind and have fans checking in on you, so you can get your latest music, merch, or livestream tickets under the eyes and into their shopping carts? Shifting to user-centric would mean the streaming landscape is more closely aligned with the same metrics used to build dedicated fan bases.