How the Outlier Score works — the math, the limits, the wins
Every score is views ÷ creator median × 100. We explain why this beats raw view counts, the edge cases, and how to read it.
The math
Outlier Score = (video views / creator's 30-day median views) × 100.
That's it. A score of 100 means the video performed exactly at the creator's recent average. 500 means 5× over. 1000+ is a genuine breakout.
Why this beats raw view counts
Raw view counts measure audience size, not signal. A creator with 2M followers averaging 800K views per post who hits 1M with a new format isn't doing anything special — that's noise. A creator with 8K followers averaging 2K views who lands a 60K-view post is doing 30× better by signal-to-baseline.
The Outlier Score asks: did this format work, given the audience this creator has built? That's the question content strategists actually care about.
The limits
Three things to know:
- Cold-start creators (< 5 videos in 30d) get score = 0. We don't have a stable baseline yet.
- Median, not mean — one viral hit doesn't poison the baseline. The 6th-best video in 30 days is the reference point.
- Posted-at sensitivity — a video that's only been live 2 hours is still rolling up. We re-score on every scrape cycle.
How to read it in practice
A 200+ score on a creator over 100K followers is a real format signal.
A 600+ score on a creator under 10K followers is a great hook with limited reach.
A 1000+ score on a verified account is a "stop everything, study this" moment.
Stack the score next to follower count and 7d engagement split (likes/comments/saves) and you've got a much sharper read than what the native discover tab gives you.