Tough life for emerging bands and singers. Not only does music streaming data confirm that every year dozens of millions of audio tracks go unheard by anyone, but now on Spotify artificial intelligence is increasingly used to create thematic playlists. And the expert editor’s nose is increasingly snubbed.
The nerve center as usual is in the data, especially considering that the business is now closely linked with streaming platforms: Luminate’s latest 2023 report indicates that of the 184 million songs on the platforms, approximately 86.2% have received 1000 or fewer plays. Even more striking is that 45.6 million recorded zero plays, causing damage to those who deal above all with distribution, because archives have a cost. There is now talk of 120 thousand new tracks uploaded every day on companies such as Spotify, while in the 80s they were published a few thousand cassettes or CD per year.
Another critical issue is linked to the fact that we are moving from a royalty distribution model called pro-rata to that user-centric. The first (so far most widespread) considers all the platform’s revenues (advertising and subscriptions), subtracts a compensation between 30% and 40% and distributes the rest to the rights holders: distribution takes place on the basis of the percentages of streams relating to the tracks and the large ones are favored over the small ones. The second model, which Spotify will introduce starting this quarter, requires that a track, in order to monetize, reach a minimum threshold of listeners (yet to be defined) and reproductions (at least 1000).
Curiosity Taylor Swift is Time’s Person of the Year. But on YouTube, TikTok, Spotify and Tumblr she was already there since 2014 by Emanuele Capone 06 December 2023
AI takes the work out of playlist curators
And here is the reason why Spotify has increasingly open to artificial intelligence. If playlists once enjoyed the expertise and sensitivity of human curators to discover new phenomena, now the goal seems to be to reduce any unexpected events, play it safe by maximizing the number of reproductions and make the most of statistics and data analysis. As Bloomberg explains, if starting from 2015 the RapCaviar hip hop playlist on Spotify made it possible to debut in global panorama (then) totally unknown artists like Cardi B, later TikTok’s music selection algorithms stimulated a new reflection. Automated recommendation systems are generally popular with the general public and work without major risks.
And former Spotify employee he defined this transition as “meritocratic”, given that it is based on data, and less linked to the tastes of the curators. Over the years, many have been fired and human teams have become smaller and smaller. Some playlists are fully automated, others only partially. The company encourages the few remaining to use data better and tag songs to help the AI in its tasks. The first side effect, according to the employees of the majors, is that they have encountered a decline in reproductions: with the historic RapCaviar, the contraction was between 30% and 50%. As if some value had been lost.
“Everyone likes being on playlists, but it’s still like everything else in the music industry, there’s no one thing that works,” he said. Jody Whelan, who runs Oh Boy Records – Everyone does it because they hope it will lead to the next success.” Sulinna Ong, global editorial manager of Spotify, has a different opinion, confirming that there will always be editorial playlists and that their audience is constantly growing, only that tastes are increasingly diversified.
And so what What can an emerging artist do? Maybe try Spotify’s new Discovery Mode that lets artists enjoy an algorithm nudge (while listening to radio or autoplay) in exchange for a lower royalty rate. More visibility but less revenue than usual, in practice.