The Boyan Cluster: A 21-Artist Spotify Manipulation Network, Found From One URL

1. The Starting Point

The investigation began as a routine BADS-Audio scan on one artist — Boyan, a Dutch act with 119 monthly listeners and 54 followers — requested by a client evaluating the artist for sync licensing. Boyan's public profile showed no bio, one external link, six singles released in a single 2022–2023 window, and no activity since.

The full 8-signal BADS-Audio scan produced a Mark Score of 37.5 (WATCH band), driven almost entirely by one extreme signal: a 932× stream-to-listener ratio. Boyan's top track had 51,348 plays on a profile with a follower base of 54. Combined with zero releases in two and a half years and a geographically incoherent listener base (Amsterdam, Johannesburg, and three other cities at near-equal weight), the pattern read as a classic deployed-and-abandoned streaming farm — not a random inactive artist.

That single profile, on its own, would already have been enough to flag Boyan as unsuitable for licensing without further explanation from the rights-holder. The client authorized a network expansion to see whether the pattern extended further.

2. First-Degree Expansion: The Pattern Locks In

Pulling Boyan's Spotify-generated "Fans also like" related-artist list and scanning the five most densely-connected candidates produced a unanimous result: all six artists exceeded the 150× stream/listener ratio that BADS-Audio treats as the confirmed-manipulation threshold, several by an order of magnitude.

Every one of the six showed the same underlying signature: an active release cadence in a 2019–2023 window, followed by total silence, with historical tracks stranded at 30,000–336,000 plays against current follower counts in the dozens-to-hundreds.

3. Second-Degree Expansion: 15 More Artists

Batch-scanning 15 additional related artists — prioritizing those appearing in multiple members' "Fans also like" lists — surfaced two hub nodes: Jae Fiend (6 overlapping network connections) and Ruth McKnight (5 overlapping connections), both confirmed via the same DK/DK2 distributor fingerprint and extreme stream-ratio anomalies. Three more artists — Tommy, Lord Joey, and 99HeartBreak — carried the distributor fingerprint but showed zero related-artist graph overlap, consistent with a farm operator deliberately spreading accounts across genres to avoid Spotify's recommendation algorithm linking them together.

Two artists in the expansion — Troli Bear and Elevenex — scored high on the stream-ratio signal but showed verified badges, ongoing 2025–2026 releases, and credible independent press coverage. BADS-Audio flagged both as likely false positives: a documented edge case (a single legitimately viral track) that the methodology is built to catch and exclude rather than paper over.

4. Attribution: The Humans Behind the Network

Beyond fraud detection, ten of the twenty-one profiles disclosed enough identity information in their public bios and linked social accounts to constitute real attribution leads — among them a full name and date of birth, a stated hometown and prior stage alias, and direct contact information in a Spotify artist bio. None of these were independently verified in this cycle; they are documented as the highest-value next step for an upstream investigation, not as confirmed identities.

5. Confidence Assessment

Evidence for coordination: mutual related-artist density well above base rate at the two hub nodes; a shared distributor-pipeline fingerprint across roughly one-third of the network; a uniform 2019–2023 deployment-then-silence pattern; 20 of 21 artists scoring in the confirmed-manipulation band on stream/listener ratio alone.

Evidence against blanket inclusion: several network members disclose detailed, checkable personal narratives inconsistent with a pure bot-farm shell; a subset of distributor-fingerprint carriers show zero graph connectivity, which argues either for deliberate genre-spreading by the operator or a weaker-than-assumed fingerprint signal; two members are actively releasing music into 2026 and do not fit the dormant pattern.

Verdict: moderate-to-strong confidence that a coordinated manipulation network exists, with a core of 10–14 tightly-connected artists rather than a single monolithic 21-artist ring. The methodology's own discrimination of 2–3 likely false positives inside the discovery graph is, if anything, evidence the scoring is working as designed rather than over-flagging.

6. What This Demonstrates

• Starting from one URL produces an ecosystem map. Spotify's own trust-and-safety tooling operates per-account; it does not publicly demonstrate networked-operator identification the way this scan did.
• Every claim is source-cited. Every ratio, follower count, distributor imprint, and bio quote in the underlying investigation traces to a specific, timestamped data pull. Reports are built to survive audit.
• Attribution is a separate deliverable layer. Beyond flagging manipulation, the methodology surfaces the humans plausibly behind it — the operational intelligence layer platform-native tools don't produce.
• The unit economics are favorable. Twenty-one artists scanned, a network graph built, and real-name attribution leads collected — for roughly $0.60 in data costs and under 30 minutes of active investigation time.

Who This Is For

• Distributors and aggregators — batch-scan a roster to catch manipulation networks operating through your pipeline before Spotify enforcement escalates to distributor-level fines.
• Labels and artist management — run this exact workflow before signing, to know whether a prospective artist clusters with a known farm network.
• Rights holders, PROs, and legal teams — evidence packages suitable for royalty-dispute proceedings and distributor escalation.
• Investigative journalists and IFPI-adjacent researchers — case-study-grade material on how streaming manipulation actually looks in the wild, with real numbers.