Investigation · Economics

Who answers the money questions in music

If you ask the internet how artists make money from streaming, the first two answers come from companies that sell t-shirts.

We ran four high-volume music-business queries and classified every returned publisher by what that publisher sells rather than by what the article says. The result is not that the answers are wrong. It is that on the questions where money is at stake, almost everyone answering has a product riding on the answer — and one question in the set proves this is specific to money rather than true of music search generally.

The method, before the finding

Four queries were run on 2026-08-12: three about money and rights, and one about credits included deliberately as a control. Each returned publisher was classified by its commercial position — distributor, merchandise, promotion, creative services, platform, journalism, tool — based on what the company sells, not on the quality of the page.

The classification is the variable. A distributor answering a question about distribution has a commercial position on the answer whether or not the article is accurate. We are not measuring accuracy. We are measuring who is structurally free to say the discouraging thing.

What came back

“How do artists make money on Spotify”

The top two results are merchandise companies. Their answer to a question about streaming income is, structurally, that streaming income is not where the money is — one of them notes that a single $35 t-shirt out-earns roughly 7,000 to 10,000 streams. That may well be true. It is also precisely what a t-shirt company would say either way, and that is the whole problem: the reader has no way to separate the accurate part from the interested part.

“How much does Spotify pay per stream”

The first two results are distributors — Ditto Music and TuneCore — followed by calculators, a promotion service, and a company that sells cover artwork. The question is *what does the pipe pay*, and the answer is being given by the pipe.

“Black box royalties”

Here Billboard ranks first — real reporting, no product in the answer. Behind it: two distributors, a course seller, and a playlist-promotion company. The money at stake is not small; the MLC received $424,384,787 in accrued historical black-box mechanicals from digital services in 2021.

The control, which is the part that matters

If vendor dominance showed up everywhere, this would be a finding about search, not about music money. So the fourth query asked a credits question: who writes for rappers.

Six of seven results were editorial outlets with nothing to sell — Forbes, XXL, HotNewHipHop, Vocal, Bleu, and an independent list project. Only one had a product, and it sells ghostwriting.

What this actually means

The easy conclusion is that editorial cannot compete on money queries. The data says the opposite. Where independent journalism does show up on a money question, it places well: The Pudding sits in the top three on the Spotify-income query with a piece from 2022, and Billboard takes first on black box.

So the gap is not that this content cannot rank. The gap is that almost nobody is publishing it. The questions with the most money attached to them are the ones with the fewest disinterested answers, and that vacuum is filled by whoever has a marketing budget and a reason to be in the conversation.

Which is a strange thing to be able to say out loud, so we will: this publication exists because of the hole this audit measures. We sell no distribution, no promotion, no merchandise and no position in any ranking, which is not a virtue — it is simply the only reason we are able to publish a number that discourages a purchase.

What would prove us wrong

• A larger sample flipping the ratio. Four queries is indicative, not conclusive. Run twenty and publish it — the dataset is open and the classification rule is stated so it can be extended.
• The control failing to hold. If credits queries turn out to be vendor-dominated too, the finding is about search generally and this framing is wrong.
• Vendor content proving to be the most accurate. We measured commercial position, not correctness. If a systematic accuracy audit found vendor pages more reliable than editorial ones, the conflict would matter far less than we are claiming.

Limits, plainly. n=4 queries, collected on a single day, unweighted and unpersonalised, with no click or impression data because we have none. Search results are volatile and regional; a different day returns a different set. Publisher classification is a judgement call, recorded per row so it can be disputed. No claim is made about the accuracy of any individual article listed.

The data

Every observation, the classification scheme and the stated limits ship as an open dataset. Take it, extend it, or use it to contradict us.

Download data.json How we measure

Collected 2026-08-12 · CC-BY 4.0, commercial reuse permitted, attribution and a link back required.
Referenced: Billboard on unclaimed black-box royalties · The Pudding, How Artists Get Paid From Streaming · The MLC · Spotify for Artists royalties guide
Related: Black box royalties · Per-stream rate · Pro-rata vs user-centric · Economics