Methodology
This page documents how Vyrex computes viral scores, builds formulas, and publishes the Trend Index — so journalists, creators, and search engines can cite us with confidence.
Viral score (0–100)
Each analysed video receives a composite score from hook strength, engagement velocity (likes, comments, shares relative to views), share ratio, retention proxies where available, and platform-specific view floors. TikTok viral threshold: roughly 500k views with ~5% engagement; Instagram Reels ~100k / 4%; Facebook video ~200k / 3%. Scores are normalised so cross-platform comparison is directional, not absolute.
Follower-adjusted scoring down-weights outliers when a creator's audience size makes raw views misleading. Full breakdown appears in the platform after URL analysis.
Formula clustering
Videos marked viral enter hook-type and emotional-trigger clustering. Patterns with sufficient sample size become formula archetypes. Only admin-approved patterns surface publicly (`public_visible`). Each formula page includes sample counts, average scores, and related hooks — all aggregate.
Trend Index cadence
- Live index — recomputed continuously from recent viral records; web ISR hourly.
- Weekly snapshot — frozen every Sunday 23:00 UTC for permanent archive URLs.
- Monthly report — compiled State of Viral report from weekly snapshots.
Editorial living data
Pillar guides embed live components (`LiveTopHook`, `Stat`, `TrendChart`) that refresh on ISR without rewriting entire articles. A monthly drift check flags when narrative sections need a light refresh; humans approve before publish.
What we exclude
No private URLs in public outputs. No individual creator call-outs in aggregate editorial. No fabricated performance claims. Health or bodily-effect language is excluded from all Vyrex content by policy.