Why Major Streaming Services Are Killing the Long Tail of Indie Artists

In 2004, a Wired editor named Chris Anderson looked at the sales data of online retailers and saw the future. Digital shelves were infinite. Freed from the physical constraints of a record store that could only stock the hits, an online catalog could carry everything—and, Anderson argued, the aggregate demand for all the obscure, niche, and forgotten titles at the far end of the curve could rival or exceed the demand for the blockbusters at the head. He called this the “long tail,” and in his 2006 book The Long Tail: Why the Future of Business Is Selling Less of More, he turned it into one of the defining business theses of the streaming age. The promise was intoxicating for musicians: in a world of infinite distribution, the middling and the marginal would finally find their audiences. A death-metal band in Jakarta, an ambient composer in Reykjavík, a bedroom folk singer in Ohio—each could reach the handful of listeners scattered across the globe who would love them, and each could earn a living doing it.

Two decades later, the infrastructure Anderson imagined exists in almost perfect form. Spotify, Apple Music, Amazon Music, YouTube Music, and Deezer host catalogs of roughly 100 million tracks, and nearly anyone can add to them for a few dollars through a distributor. The infinite shelf is real. And yet the economic promise attached to it has largely failed to materialize for the artists at the tail’s far end. The mechanics of how streaming money is calculated and distributed, combined with a set of recent platform policy changes and a tidal wave of new uploads—much of it now generated by artificial intelligence—have produced an ecosystem that systematically funnels revenue toward the head of the curve while starving the tail. This essay makes the argument that major streaming services, through both deliberate policy and structural design, are killing the long tail of independent music. But because the case is genuinely contested, and because the defenders of streaming have real evidence on their side, it also tries to present their strongest counterargument honestly before weighing the two.

What the Long Tail Promised

Anderson’s original thesis rested on a few interlocking claims. First, the cost of carrying inventory in a digital catalog approaches zero, so there is no economic reason to drop unpopular titles. Second, better search, recommendation, and filtering tools allow listeners to find niche work they never would have encountered in a physical store. Third, and most importantly, the accumulated demand for the millions of obscure items—each selling little on its own—forms a “tail” so long that it can hold as much collective value as the “head” of a few megahits. Markets, in this view, would fragment away from a monoculture of blockbusters and toward a rich diversity of niches, each economically viable.

For music, the appeal was obvious. The twentieth-century record industry was a hit-making machine that had to bet on a small number of acts because pressing vinyl, printing CDs, and buying shelf space were expensive. The long tail suggested that the internet would dismantle this gatekeeping. If distribution was free and discovery was frictionless, then the fundamental constraint on a niche artist’s career—getting heard at all—would evaporate. What remained was the belief that being heard would translate into being paid.

It is worth noting that Anderson’s thesis was contested from the start. Harvard Business School’s Anita Elberse and researchers analyzing Rhapsody and Nielsen data argued through the late 2000s that demand remained stubbornly concentrated at the head, and that the tail, while long, was extraordinarily flat—millions of titles selling almost nothing. That early skepticism turned out to be prophetic. But the romantic version of the long tail, the one that promised a living to the niche musician, became the implicit moral framework through which streaming justified itself. Understanding how that promise broke requires understanding the money.

How Streaming Royalty Math Actually Works

The single most important and least understood fact about streaming economics is that services like Spotify do not pay artists “per stream” in any fixed sense. There is no set rate of, say, $0.004 per play sitting in a contract. Instead, the dominant model is what’s called pro-rata, or the “big pool” model, and it works roughly like this.

Each month, the streaming service takes all the subscription and advertising revenue it collects, sets aside its own cut (Spotify has historically retained around 30 percent, paying out roughly 70 percent to rights holders), and pours the rest into one enormous pool. It then counts the total number of streams across the entire platform for the month and divides the pool by that total to arrive at an effective per-stream value. Each rights holder—labels, distributors, and publishers, who then pay artists—receives a share of the pool proportional to their share of total streams. The widely cited figure that Spotify pays “between $0.003 and $0.005 per stream” is not a rate; it is an after-the-fact average, and it drifts month to month depending on how much money came in and how many streams were played.

This design has a subtle but decisive consequence. Because the pool is shared globally, your streams do not just earn money—they compete with everyone else’s streams for a slice of a fixed pie. When a listener who only ever plays your obscure record pays their $11.99 subscription, that money does not follow their listening. It goes into the common pool and is redistributed according to global stream share. The megastar with billions of plays effectively draws revenue out of the same pool that your niche listener funded. In a pro-rata system, the head of the curve is not merely more popular than the tail; it is paid out of money that tail listeners generated.

The Mechanisms That Disadvantage the Tail

Pro-Rata Dilution

The first and most structural mechanism is what critics call pro-rata dilution. Because payouts are a function of your share of total platform streams, anything that increases the denominator—the total number of streams—reduces the value of every individual stream, unless the pool grows just as fast. As catalogs balloon and as streaming volume is padded by passive listening, background playlists, and low-attention “functional” audio (sleep noise, white noise, lo-fi loops), the tail artist’s fractional slice shrinks. A dedicated fan base of 500 devoted listeners is worth less in a pool of 700 billion monthly streams than it would be in a pool of 100 billion. The niche artist is, in effect, taxed by the sheer scale and churn of the platform they inhabit.

Minimum-Stream Thresholds

The second mechanism is explicit policy. Beginning in early 2024, Spotify implemented a rule that a track must reach a minimum of 1,000 streams within a rolling twelve-month period before it earns any recorded-music royalties at all. Tracks below that threshold are demonetized; the money they would have generated does not go to their creators. Spotify’s own framing, laid out in its “Modernizing Our Royalty System” announcement, is that these are tracks earning on average “less than five cents per month,” that the tiny sums often never reach artists because distributors impose minimum withdrawal amounts, and that redirecting roughly $40 million a year away from this population and toward artists above the threshold puts money into the hands of “emerging and professional” musicians rather than letting it evaporate in micro-payments.

Critics see the same policy very differently. Believe CEO Denis Ladegaillerie asked the obvious question: “Why would you not pay such an artist? It doesn’t make any sense.” The threshold does not save the money for the platform, but it does redistribute it upward—from the vast population of tracks below 1,000 annual streams, which is precisely the long tail, to the tracks above it, which are by definition more popular. A track with 900 genuine streams from 900 real people earns exactly nothing while contributing its listeners’ subscription dollars to the pool. In December 2025, Spotify publicly defended the threshold again after a critical report challenged it, underscoring that the policy remains both live and contentious. Whatever one makes of the rationale, the mechanical effect is unambiguous: the policy is a rule that pays the head and zeroes out the tail.

Playlist Gatekeeping and “Discovery Mode”

The third mechanism concerns discovery itself—the very thing the long tail promised to democratize. In practice, listening on major platforms is heavily shaped by editorial and algorithmic playlists, and placement on influential playlists can make or break a track’s numbers. This recreates, in software, exactly the gatekeeping the internet was supposed to abolish. The gatekeepers are now playlist editors and recommendation algorithms rather than radio programmers and record-store buyers, but the bottleneck is real.

Spotify’s Discovery Mode, launched in 2020, sharpened these concerns. The program lets artists and labels flag tracks for algorithmic prioritization in radio, autoplay, and certain mixes. There is no upfront cash payment; instead, artists accept a roughly 30 percent reduction in royalties on streams generated through those promoted placements. Spotify markets it as a tool, reporting that participating tracks see average first-month lifts on the order of 50 percent more saves and 44 percent more playlist adds. Critics, including members of Congress and the Recording Academy, have argued it functions as a form of payola—pay-for-play—that pressures cash-strapped independent artists into accepting lower per-stream rates simply to be surfaced. In November 2025, Spotify was hit with a proposed class-action lawsuit alleging Discovery Mode is a deceptive “pay-for-play scheme,” with the plaintiff claiming the feature pushed major-label acts like Drake that bore little resemblance to her actual listening. Spotify responded that the suit was “riddled with misunderstandings” and that Discovery Mode “does not allow artists to buy plays.” The legal question is unresolved, but the structural point stands: a discovery system in which visibility can be traded for a royalty discount tilts the field toward those who can most afford to give up income, which is rarely the artist at the tail’s edge.

AI and Spam Flooding the Pool

The fourth mechanism is the newest and perhaps the most corrosive. The infinite shelf has been discovered by people who have no interest in music at all, only in extracting money from the pool. The scale is staggering. In 2026, Deezer reported that fully AI-generated tracks had crossed more than 50 percent of all new uploads to its platform, with the daily figure climbing from around 75,000 in April to roughly 90,000 by mid-summer. Critically, Deezer also estimated that up to 85 percent of the streams generated by fully AI tracks in 2025 were fraudulent—the product of bots and stream-manipulation farms designed to siphon royalties. TechCrunch, Music Business Worldwide, and Deezer’s own newsroom all documented the trend, and a 2026 Forbes investigation characterized AI-generated music as the engine of a multibillion-dollar fraud operation.

There is a partial silver lining in the same data: despite dominating uploads, fully AI-generated tracks accounted for only 1 to 3 percent of actual streams on Deezer, suggesting listeners still overwhelmingly choose human-made music. Deezer also excludes streams it identifies as manipulated from its royalty calculations. But the flood still harms the long tail in two ways. First, in a pro-rata system, every fraudulent or spam stream that is not caught dilutes the pool and drags down the per-stream value for everyone. Second, and more insidiously, the deluge of low-effort content makes genuine niche work harder to find, burying the human artist deeper in an undifferentiated ocean of algorithmic filler. The infinite shelf, it turns out, is only a blessing when the things on it are worth finding.

The Evidence of Payout Concentration

If these mechanisms were merely theoretical, we would expect the distribution of streaming income to look roughly like the distribution of listening. Instead, the data—much of it published by the platforms themselves—shows extreme concentration at the head.

Spotify’s own “Loud & Clear” transparency site, updated in 2026 with 2025 figures, reports that the service paid out more than $11 billion to the music industry in 2025, bringing lifetime payouts to roughly $70 billion. Within that, the company highlights that about 13,800 artists earned at least $100,000, more than 1,500 earned over $1 million, and 80-plus top artists each pulled in more than $10 million. Spotify presents these numbers as evidence of a broadening middle class of musicians, and there is something to that (more on the counterargument below). But turn the same numbers over and the concentration is stark. The platform hosts on the order of 10 to 12 million uploading artists. If fewer than 14,000 clear the $100,000 mark, that is well under a fifth of one percent. The overwhelming majority of the catalog—millions of acts—shares the thin remainder. Spotify itself notes that the “100,000th-ranked artist” earned around $7,300 in 2025; the artist ranked one-millionth or five-millionth earns a rounding error. The curve does not flatten gently into a viable tail. It collapses.

Independent analyses have long reached similar conclusions: a small fraction of tracks generate the overwhelming majority of streams, and therefore of revenue, while the tail—comprising the vast bulk of the catalog by count—captures a sliver. This is precisely the pattern that Elberse and other early skeptics predicted, and it is the empirical heart of the case against the long-tail promise. The tail exists. It is longer than anyone imagined. But as a place to earn a living, it is close to barren.

The Counterargument: Streaming Democratized Access

Intellectual honesty requires taking the other side seriously, and the defenders of streaming are not making things up. Their strongest case runs as follows.

Before streaming, the barrier facing a niche artist was not a low per-stream rate—it was total exclusion. To reach listeners at all, you needed a label, a distributor, physical manufacturing, shelf space, and radio play, and the gatekeepers said no to almost everyone. The long tail, in the pre-streaming world, did not earn little; it earned nothing, because it could not get on the shelf. Streaming eliminated that barrier almost entirely. For roughly the price of a coffee, any musician on earth can place their work beside Taylor Swift’s in the same catalog, reachable by 600-plus million listeners in nearly every country. That is a genuine and historic democratization of access, and it is not nothing.

The defenders point to real distributional gains. Spotify’s data indicates that the number of artists generating meaningful money has grown substantially: the count earning $100,000 or more rose by roughly 1,400 year over year, and the artist ranked 100,000th saw their annual take grow more than twentyfold over a decade, from around $350 to over $7,300. They emphasize that roughly half of all Spotify royalties now flow to independent artists and labels rather than the majors, that more than a third of artists crossing $10,000 started as DIY acts, and that 85 percent of newer artists breaking $100,000 are based outside the United States—evidence, they argue, of a genuinely global and decentralizing opportunity that the old system could never have produced. A Nigerian or Korean or Brazilian artist reaching a worldwide audience without a Western label is a real phenomenon streaming enabled.

On the specific policies, defenders make coherent arguments too. The 1,000-stream threshold, they note, does not enrich Spotify—the money is redistributed to other artists, and the demonetized sums were often so small they never actually reached anyone, dying in distributor minimum-payout requirements. On AI and fraud, they can fairly say that the platforms are the ones detecting and excluding manipulated streams, that human music still commands 97-plus percent of actual listening, and that the flood is a problem the services are actively fighting rather than causing. And on the deepest point, they argue that the concentration of income is not a streaming pathology but a feature of every cultural market in history: hits have always dominated, superstars have always captured the lion’s share, and blaming Spotify’s math for the fact that most music finds few listeners confuses the messenger with the message. People, in the end, mostly want to listen to what other people are listening to.

This is a serious position. The strongest version of it is that streaming did exactly what the long tail promised on the dimension of access—it made the shelf infinite and open to all—and that the failure to convert access into income reflects the immovable reality of concentrated demand, not a fixable flaw in royalty design.

Weighing the Two

Where does this leave the argument? The counterposition is right that access has been democratized and wrong to conclude that this settles the question. The long-tail promise was never only about access; it was about viability—the claim that the accumulated demand for niche work could sustain the artists who made it. On that promise, the evidence is damning, and crucially, several of the reasons are not natural laws of taste but choices baked into how the system pays.

Concentrated demand may be immovable, but pro-rata pooling actively amplifies it by letting the head draw revenue from money the tail’s listeners generated. That is a design decision, not a fact of nature. Minimum-stream thresholds are a policy, chosen and defensible perhaps, but a policy that by construction zeroes out the tail. Discovery Mode’s royalty-for-visibility trade is a designed feature. The vulnerability of a pro-rata pool to fraud and AI dilution is a consequence of the model. Each could be otherwise. The defenders’ best move—”demand has always been concentrated”—is true but incomplete, because the specific mechanisms of streaming take concentrated demand and make the resulting income more concentrated than the listening itself. When 1 to 3 percent of streams can be AI spam, when sub-1,000-stream tracks earn zero, and when a fixed pool means your fan’s subscription funds someone else’s stream, the system is not neutrally reflecting taste. It is bending the payout curve further toward the head than the attention curve already bends.

The Alternatives

If the pro-rata pool is the problem, the natural question is what could replace it. Several alternatives exist, in ascending order of radicalism.

User-centric (or “fan-centric”) royalties would abolish the common pool. Instead of dividing all revenue by all streams, each individual subscriber’s fee would be divided only among the artists that subscriber actually listened to. If you pay $11.99 and play nothing but one obscure noise artist all month, that artist receives your subscription (minus the platform cut) rather than a globally averaged fraction of it. In principle this directly rewards devoted niche fandom and starves stream-farm fraud, since bots cannot manufacture real paying subscribers. In practice, the results of trials and studies have been more muted than advocates hoped: analyses commissioned in France and elsewhere, and SoundCloud’s “fan-powered royalties” experiment, suggested the aggregate reshuffling is smaller than expected, because most listening—even individual listening—still skews toward popular artists. User-centric helps the tail at the margin and closes the fraud loophole, but it is not a panacea. Deezer and others have advocated versions of it, and it remains the most-discussed structural reform.

Direct-to-fan platforms attack the problem from the other end by largely abandoning the streaming model. Bandcamp is the canonical example: rather than fractions of a cent per stream, it sells music and merchandise directly, passing roughly 80–85 percent of revenue to the artist after its cut on digital sales (its share on physical and merch differs). A single $10 album sale on Bandcamp can equal what thousands of streams would pay, which flips the economics for a niche artist with a small but committed audience—exactly the long tail’s population. The catch is reach: Bandcamp requires fans who already want to buy, and does not deliver the passive, algorithmic discovery that streaming does. It rewards existing relationships rather than creating new ones.

Self-hosting and owned channels—an artist’s own website, mailing list, Patreon or membership platform, direct downloads—push this logic further. They cut out the platform almost entirely, letting artists keep the large majority of what fans pay and, critically, own the relationship with their audience rather than renting access to it from an algorithm. The tradeoff is that the artist must now do the work of discovery, marketing, and retention themselves, with none of the frictionless reach that made streaming appealing in the first place.

The pattern across all three alternatives is the same tradeoff: models that pay the tail fairly tend to sacrifice reach, and models that maximize reach tend to pay the tail poorly. The long-tail dream imagined a world where an artist could have both—infinite reach and viable income. No current model delivers both at once, and that gap is precisely where the tail’s artists are stranded.

Conclusion

The infinite shelf was built. That much of Chris Anderson’s vision came true with a completeness he could hardly have predicted: a hundred million tracks, open to anyone, reachable everywhere. But the shelf was only half the promise. The other half—that the accumulated demand for niche work would sustain the people who made it—has not been kept, and the reasons are not merely the stubborn concentration of human taste. They are, in large part, the specific design of how streaming counts and distributes money: a shared pool that lets hits feed on the tail’s revenue, thresholds that pay the popular and zero out the obscure, discovery systems that can be bought with royalty discounts, and a fraud-and-AI flood that dilutes the pool from below.

The honest verdict is a divided one. Streaming genuinely democratized access and genuinely built a larger middle class of professional musicians than the old gatekept industry ever allowed—there really are more artists earning real money, in more countries, than before, and the defenders are entitled to that victory. But it did not democratize income at the tail, and on several fronts it actively concentrated income more sharply than attention alone would dictate. The long tail is not a myth; it is longer and more populous than anyone imagined. What is a myth is that living there could pay. Whether that changes depends less on the size of the shelf, which is already infinite, than on the arithmetic of the pool—arithmetic that, so far, the major services have every incentive to leave exactly as it is.

How Independent Programmers Build a Personal Spotify on

Why Bandcamp Pays Artists Dozens of Times More

Navidrome vs Jellyfin: Which Open-Source Server Is Better

Funkwhale and Decentralized Streaming: Music Without Algorithm Censorship

Contact Info

Some Populer Post

Cantilever, Lissen and Other Niche Services: Streaming Without Algorithms

Open-Source Music Players in 2026: From Feishin to Spotube