We cross-check creator behavior, audience signals and content context: engagement consistency over the last 30 days, follower growth patterns, median/min/avg/max views, semantic match of titles/descriptions/transcripts, audience geography/gender/language splits, and reachability scores — all flagged by our fraud-detection heuristics to highlight bots, sudden spikes, or low-quality reach.