The Invisible Replacement: How Spotify's Ghost Artists Are Disappearing Real Musicians

As AI music generators improve, the "Perfect Fit Content" program that already displaced independent artists from billions of streams will become an extinction-level threat to human creativity

I wanted to understand a simple puzzle: why do some tracks with millions of streams sound so... forgettable? Not bad, exactly. Just generic. Anonymous. Designed to fill space rather than demand attention.

So I spent a few hours scraping data on 25,000 Spotify playlist curators, and what I found confirms something music journalist Liz Pelly spent a decade documenting: Spotify isn’t just hosting music anymore. It’s replacing musicians with cheaper alternatives, and most listeners have no idea it’s happening. The mechanism is elegant, systematic, and about to get much worse.

The Ghost in the Playlist

Here’s how it works. You open Spotify, click on “Peaceful Piano” or “Jazz Vibes” or “Ambient Chill” — playlists with millions of followers. You assume you’re hearing real pianists, real jazz musicians, real ambient artists. But increasingly, you’re hearing something else entirely: tracks commissioned by production music companies, released under fabricated artist names, designed to sound just good enough that you won’t skip them.

Spotify has a name for this content internally: “Perfect Fit Content,” or PFC. The definition, according to leaked internal communications, is music “commissioned to fit a certain playlist/mood with improved margins.” Improved margins. Not improved quality. Not better artistry. Just cheaper.

The Swedish newspaper Dagens Nyheter discovered one composer, Johan Röhr, operating behind 650+ invented identities. His ghost artists have accumulated over 15 billion streams — placing him among the top 100 most-streamed “artists” globally, ahead of Michael Jackson, the Red Hot Chili Peppers, and ABBA. One person. 650 names. 15 billion streams. His company reported royalty earnings of roughly $30 million in 2022 alone.

But Röhr isn’t a scammer exploiting Spotify’s system. He’s working with the system. Production music companies like Firefly Entertainment and Epidemic Sound license these tracks to Spotify at reduced rates. In exchange, Spotify’s internal “Strategic Programming” team ensures they appear on high-follower mood playlists. Internal communications reviewed by Pelly show that by 2023, over 100 playlists were composed of 90%+ ghost artists, generating €61.4 million in annual gross profit for the platform.

Every ghost artist stream is a stream that doesn’t go to a real musician. And because Spotify pays out royalties based on each artist’s share of the total stream pool — a “pro-rata” system — every billion streams captured by fabricated identities mathematically reduces what real artists earn, even if their absolute stream counts stay constant.

The Structural Problem

This is why everything else in this article happens. The royalty pool is fixed — 52% of Spotify’s revenue, divided among rights holders based on their share of total streams. When ghost artists or AI tracks capture 10% of streams, they take 10% of the pool. When they capture 50%, they take 50%. Human musicians’ share shrinks proportionally, even if their absolute stream counts don’t change.

The “1,000-stream threshold” introduced in 2024 weaponizes this dynamic. Any track generating fewer than 1,000 streams annually now earns zero royalties. An estimated 86% of tracks on Spotify fall below this threshold. The revenue those millions of tracks would have generated doesn’t disappear — it gets redistributed to the top. To major label superstars. To ghost artists with guaranteed playlist placement. To AI-generated content that can flood the platform with infinite variations.

Spotify frames this as removing “noise” from the system. But noise to whom? For a DIY artist with 50 dedicated fans streaming their album 20 times each, that’s 1,000 streams. That’s a connection between musician and listener. Spotify’s system defines that connection as worthless — as content to be demonetized so the revenue can flow to fabricated identities and algorithmic outputs.

The Deception Runs Deep

The ghost artists don’t appear as corporate content. They have artist pages. They have fabricated biographies. “Ekfat,” for instance, was presented as a classically trained Icelandic beatmaker who graduated from the Reykjavik Music Conservatory and joined the “Lo-Fi Rockers crew” before releasing limited-edition cassettes. Dagens Nyheter discovered the entire story was fiction — a marketing narrative created by Firefly Entertainment to make stock music feel authentic.

When Lance Allen, an instrumental guitarist, was featured by Spotify in 2022 as their model independent artist — someone building a career through playlists — he was earning enough from “Acoustic Concentration” and “Peaceful Guitar” to pay his mortgage. But by December 2023, he tweeted that after 30 releases and consistent promotional efforts, Spotify had stopped supporting him editorially. His playlist spots had been replaced by tracks from Epidemic Sound and Firefly Entertainment.

This is what happened to Lance Allen — and it’s the pattern emerging across ambient, jazz, classical, and lofi: real artists with real catalogs, quietly replaced by anonymous stock music that costs Spotify less to license. The platform presents this as organic discovery — the algorithm surfacing music listeners want — but internal documents show it’s deliberate strategy, managed by dedicated teams with profit targets.

Why Mood Playlists Are the Target

Mood playlists are the target because they’re full of fungible content — music that works as utility, not art. Spotify’s 2012 market research revealed that passive “lean-back” listening — music as background for work, sleep, study — represented a larger market than active discovery. The company pivoted from being “the Google of music” (search-focused) to “music for every moment” (mood-focused).

This shift created categories where musical complexity is irrelevant. If you’re listening to “Deep Focus” while writing emails, you’re not evaluating artistic vision. You’re tolerating sound that doesn’t distract you. The music becomes utility — like wallpaper or air conditioning. And if it’s utility, why pay a real jazz ensemble when three people in a studio can generate “peaceful jazz” in one-take sessions for a flat buyout fee?

Pelly interviewed musicians who make this content. They describe being sent reference playlists of existing Spotify mood music and told to create tracks that “stream well alongside” them. The most common feedback: “Play simpler.” The goal, one musician explained, is “to be as milquetoast as possible.” The musical process: “I just write out charts while lying on my back on the couch. We knock out 15 tracks in an hour or two.”

The musicians making this content don’t share in those streams. Epidemic Sound pays a flat $1,700 buyout per track, takes ownership of the master, and requires composers to formally resign from performance rights organizations — surrendering any future royalty claim. The company is valued at $1.4 billion and claims 40 million streams per day. Its investor pitch: “This is at the end of the day a data business.”

The Racial Mathematics

The profit targeting has a demographic shape. When Ambient Chill removed tracks by Brian Eno, Bibio, and Jon Hopkins — pioneers of the genre — their slots weren’t filled by emerging experimentalists. They were filled by anonymous tracks from Swedish production companies. As one source told Pelly, “Spots for Black and brown artists making this music started getting cut down to make room for a few white Swedish guys in a studio.”

This is genre capture. Ambient music, jazz, lofi hip-hop — these have been spaces where diverse, global artists built international followings outside major label systems. But when Spotify optimizes for “improved margins” in mood categories, it systematically favors Northern European production houses with direct licensing deals. The platform isn’t just reducing costs. It’s homogenizing culture, replacing the messy diversity of actual music scenes with corporate-controlled content optimized for passive consumption.

The AI Endgame

Ghost artists still require human musicians, even if they’re working in assembly-line sessions creating forgettable tracks for flat fees. But generative AI removes even that constraint.

Boomi, an AI music generator, has released 14.5 million songs — roughly 14% of the world’s recorded music. Users select preset styles (”lofi,” “chill electronic,” “relaxing meditation”), and the app generates tracks in minutes. Boomi keeps 20% of streaming revenue and owns the copyright. When Spotify temporarily banned Boomi in 2023, it wasn’t for using AI — the company made clear it had “no objection to generative AI content.” The ban was for artificial streaming (bot farms inflating play counts). Actual AI music? Perfectly fine.

By the end of 2023, Boomi had secured a partnership with Warner Music Group, ensuring continued access to streaming platforms. Universal Music Group and Warner have both partnered with Endel, another AI music generator, to create “personalized functional soundscapes” using their back catalogs. Roberta Flack’s “Killing Me Softly” has been remixed into AI-generated sleep music. Sia’s Christmas album exists as algorithmic chill-out versions.

The infrastructure Spotify built for ghost artists — the Strategic Programming team, the “improved margins” mandate, the mood playlist dominance — becomes the perfect delivery system for AI-generated content. Why hire even three people when an algorithm can generate infinite variations for near-zero marginal cost?

Daniel Ek, Spotify’s CEO, has called AI-generated music a “great cultural opportunity.” For the platform, perhaps. For musicians, it’s the logical endpoint of a system designed to minimize what they’re paid while maximizing how their labor is extracted.

What Gets Lost

When I analyzed the 25,000 playlist curators, I found that the top 1% control roughly 50% of total reach. Corporate operations like Filtr (owned by Sony) sit at the top with 9.19 million followers. The system presents itself as democratized — anyone can make a playlist, any artist can get discovered — but the data shows extreme concentration. Power law distribution. Winner-take-all dynamics. And at the very top, major labels and production music companies optimizing for “improved margins.”

But the numbers don’t capture what’s actually being erased. When Lance Allen loses playlist placements to anonymous tracks, it’s not just his income that disappears. It’s the possibility that a listener discovers his specific sound, follows his work, comes to a show, buys a record, forms a connection. Ghost artists can’t tour. They can’t respond to emails. They can’t grow with an audience because they don’t exist.

The ambient artist Anoni discovered her deeply despairing track “Why Am I Alive Now” had been added to a “Chill Vibes” playlist. “That song is so despairing,” she told Pelly. “If we’re just going to accept the economy of music as a paradigm, there used to be complex conversations happening within that. Now it’s a very politically astute maneuver that favors a narcotic relationship to music over a complex meditative relationship to music.”

This is what Spotify’s ghost artist program accomplishes: it replaces complex relationships with narcotic ones, artistic intent with algorithmic optimization, human creativity with corporate content designed to be as inoffensive and forgettable as possible.

Where This Goes

The trajectory is clear. What required human musicians working for flat fees in 2016 now requires algorithms generating tracks in minutes. What displaced artists from 100 playlists in 2023 will displace them from thousands in 2026. What earned production companies millions will earn them billions as AI removes the last cost constraint.

This isn’t just Spotify. Apple Music, Amazon Music, YouTube Music — every platform operating on the same economic model faces the same incentive: replace expensive (real musicians) with cheap (ghost artists) with free (AI-generated). The major labels aren’t resisting. They’re partnering with AI companies, generating infinite mood remixes of their back catalogs, ensuring they control the automated future.

For independent musicians, this creates a catastrophic double bind. To compete for the playlist placements that drive discovery, they must optimize their sound for algorithmic compatibility — simpler structures, shorter intros, consistent emotional registers. The music that succeeds is music designed to blend into mood playlists alongside ghost artists. But succeeding in that system means becoming indistinguishable from the corporate content replacing them.

The alternative — making challenging, complex, culturally specific music — means invisibility. The algorithms won’t surface it. The mood playlists won’t include it. The listeners trained to expect frictionless background utility won’t tolerate it.

This is the choice Spotify’s ghost artist economy imposes: optimize for algorithmic survival or accept economic extinction.

The Only Real Exit

The only real exit is infrastructure built by musicians, not platforms.

Journalists have documented this. Data analysis can quantify it. But exposure alone won’t stop it. The system is working exactly as designed. The Strategic Programming team has profit targets. The PFC providers have guaranteed playlist placement. The AI companies have major label partnerships. Every financial incentive points toward more ghost artists, more algorithmic content, more displacement of human musicians.

What’s needed isn’t better transparency within Spotify. It’s exit from Spotify’s logic entirely.

Some artists are building alternatives. Cooperatives like Catalytic Sound (30 improvised music artists, equal revenue distribution, rotating catalog) prove a different ownership structure is viable. Libraries in several cities have launched local streaming programs that pay artists $200–300 upfront licenses for two-year terms, bypassing per-stream calculations entirely. Public funding models in Ireland, France, and Norway — basic income for artists, performance rights organizations that actually function, cultural policy treating music as public good rather than market commodity — demonstrate that streaming doesn’t have to optimize for “improved margins” at musicians’ expense.

These alternatives are small. Catalytic Sound serves 30 artists. Library streaming reaches tens of thousands, not hundreds of millions. But they prove the model. They show what infrastructure looks like when it’s built by and for the people making music, not the platforms extracting value from their labor.

Musicians, listeners, and policymakers have a shrinking window to recognize what’s being replaced — and to decide whether to act before the infrastructure that could support an alternative is gone entirely.

Because once the ghosts outnumber the living, once AI content dominates the stream pool, once the infrastructure is optimized entirely for corporate content generation — it becomes nearly impossible to rebuild what was lost. The playlists won’t remember real artists. The algorithms won’t know how to surface them. The listeners won’t remember what they’re missing.

The invisible replacement will be permanent.

---

If this research is useful to you, the playlist intelligence work — including the curator database and fraud detection methodology — lives at musinique.substack.com. The data is open. The methodology is not a secret.

---

Tags: Spotify ghost artists, Perfect Fit Content streaming fraud, AI music displacement musicians, pro-rata royalty system exploitation, playlist intelligence independent artists

#MusiqueAI #HumansAndAI #AIMusic #IndieMusician #MusicResearch #GhostArtists #AIforHumans #OpenSourceAI