The Virality Math

Before you build anything, you need the one number the whole app is built around. Get this right and the tracker is trivial — it's a scraper plus a division.

Why not likes

A post with 10k likes from an account with 5M followers is a flop. A post with 2k likes from an account with 8k followers broke through. Likes are relative to account size — they're the weakest signal on every platform in 2026 because a like costs half a second and zero intent.

You don't want the biggest posts. You want the posts that outperformed what their own account normally does. That's an outlier, and outliers are where the breakthrough topics and formats live — the stuff that reaches new audiences instead of just the followers you already have.

The formula

Every serious creator-analytics tool converges on the same definition — Shortimize, OverseerOS, and Apify all land on it independently:

Outlier score = a post's views ÷ that account's median views (in the same time window).

If an account's median is 20,000 views and a video gets 140,000: 140,000 ÷ 20,000 = 7x. The post did seven times that creator's usual. Rank a whole niche by this and the breakouts float to the top, regardless of follower count — a 10x on a small account is a stronger signal than a popular post on a huge one.

Median, not mean

Use the median, not the average. This is the nuance that trips people up.

One huge post distorts the average. On a 30-post window with a single 110k breakout, the mean gets dragged up to something like 40k when the account's true normal is ~16k. Divide by that inflated mean and every real outlier looks weaker than it is. The median ignores the one freak post and gives you the honest baseline.

Rule of thumb: if the average is much higher than the median, the account already has extreme outliers polluting the baseline. Trust the median.

The tiers

Bake these thresholds into the scorer. They're consistent across OverseerOS, Apify, and OutlierKit:

• 2x — slightly above normal. Worth noticing, log it.
• 3–5x — a strong performer. Study why it worked.
• 10x+ — a major outlier, a "monster." Deep analysis: reverse-engineer the hook, the format, the topic.
• 25x — a breakout hit. Drop everything and dissect it frame by frame.

Below 2x is just normal content for that account — noise, not signal.

Velocity — the early signal

The outlier score is a lagging measure: the post has to mature before its final view count means anything. Velocity catches winners while they're still climbing.

Compute views_per_hour = views ÷ hours_since_posted. And watch the first 24 hours — views gained in the first day (and even the first hour) are one of the strongest predictors of organic lift across every platform. A post that hits a high fraction of its account's usual first-hour engagement tends to get picked up by the algorithm and pushed to Explore or the For You feed.

So the tracker ranks on two axes: outlier multiplier (what already broke through) and velocity (what's breaking through right now).

Key takeaway

Rank by outlier multiplier = views ÷ that account's median (median, never mean). Tiers: 2x notice · 3–5x strong · 10x monster · 25x breakout. Add velocity (views in the first 24h) to catch winners early. That single division is the whole engine — the app around it is just plumbing.