How do platforms detect multiple accounts?

Platforms use five layers of detection, each progressively harder to circumvent: Layer 1 — Account data: matching email patterns, phone numbers, payment methods, names, and addresses across accounts. Defense: use unique data per account. Layer 2 — IP address: accounts sharing the same IP are flagged. Residential IPs sharing between 2-3 accounts is normal (shared household). 10+ accounts on one datacenter IP is suspicious. Defense: residential proxies, one IP per account or small group. Layer 3 — Browser fingerprint: canvas, WebGL, audio, fonts, screen resolution, and dozens more attributes create a unique device fingerprint. Two accounts with identical fingerprints are linked. Defense: anti-detect browsers with unique fingerprints per profile. Layer 4 — Behavioral analysis: login timing patterns, navigation speed, click patterns, typing cadence, and mouse movement characteristics. ML models detect when behavior across accounts is too similar. Defense: vary behavior patterns, use automation that mimics natural variance. Layer 5 — Network graph: if Account A and Account B interact with the same third accounts, follow the same pages, or share content, they are linked. Defense: isolate account activity. Do not cross-interact between your own accounts. Most platform detection combines layers 1-4. Layer 5 (network analysis) is used by Facebook, LinkedIn, and Amazon but is harder to implement at scale.