The Numbers Are Lying: What a Label Should Trust in a Post-Trust Streaming Era
In March, a North Carolina man pleaded guilty to using AI-generated songs and a farm of bots to extract more than eight million dollars from streaming platforms. The Department of Justice framed it plainly: fraud. Around the same time, SoundCloud confirmed that it blocks millions of inauthentic actions every week, and Spotify disclosed that it has swept tens of millions of spammy tracks out of its catalogue. The number that should stop every label owner cold is this: the industry estimates that fraudulent streams drain roughly two billion dollars a year from the royalty pool that pays real artists.
The cleanup is underway — detection vendors like Beatdapp now scan streaming data on behalf of major labels and platforms, and the IFPI has organised more than two dozen companies into a streaming integrity initiative. But for a label, the damage is not only financial. It is epistemic. The numbers we use to make decisions have become unreliable, and almost nobody has adjusted their decision-making to account for that.
A Pool Everyone Drinks From
Streaming royalties are paid from shared pools. Every fake stream dilutes the payout for every legitimate artist, which makes this an economic tax on the entire independent sector — not a platform nuisance. Smaller artists are the most exposed, because they rarely have the resources to monitor their own profiles, where AI-generated songs have been known to appear under real names, quietly collecting money that was never theirs. When platforms identify fraud, clawbacks typically land on distributors first, and distributors pursue the accounts. The chain of damage is long, and it starts nowhere near the people who caused it.
The A&R Problem Nobody Wants to Admit
Labels sign on signals: monthly listeners, playlist placements, chart velocity, that vague word everyone uses — momentum. Every one of those signals can now be manufactured more cheaply than ever. Bots imitate human listening patterns; VPNs hide their origins; AI generates catalogue at industrial scale. If the input data is corrupted, then every pipeline built on top of it is compromised — including the pipelines of labels that would never dream of buying a stream.
This is the uncomfortable part. You can no longer separate a hit from a hologram by looking at the numbers alone. The chart is not dead, but it is no longer a truth-teller; it is a rumour with good production values.
What to Trust Instead
None of this means abandoning data. It means ranking your signals by how expensive they are to fake:
- Paid intent. Beatport and Bandcamp purchases, vinyl pre-orders, ticket scans. Bots do not have wallets — not real ones. Money is the hardest signal to counterfeit.
- Quality of attention. Skip rates, completion rates, repeat listening from verified cohorts. Ten thousand listeners finishing a track beats a million skipping it in eight seconds.
- Scene evidence. DJ support that shows up in recorded sets. Records that move through second-hand channels. The murmur in a room when a certain track comes on.
- First-party relationships. Newsletter open rates, community activity, direct sales. Your own list is the only audience you can verify end to end.
- Verification. Where possible, insist on third-party streaming audits in distribution and licensing deals. Make clean data a contract term, not a hope.
The Label as a Trust Layer
There is an opportunity buried in this mess: verification is becoming a curatorial act. Rights vetting, consent and provenance for AI-assisted material, transparent campaign reporting — the rigour that once felt like bureaucracy is turning into brand equity. When anyone can manufacture a number, the labels that can vouch for what is real become more valuable, not less.
When counts are cheap, context is the only luxury.
The post-trust era rewards the patient. Sign the artists whose demand shows up in places a bot cannot reach, and let the inflated numbers belong to whoever needs them.