Every number on this site rests on one definition. If that definition is vague, nothing built on top of it means anything. So here it is in full, including where it breaks down.
The definition
A post is a breakout when it clears five times the creator's own baseline.
Baseline is the median view count of that creator's last 90 days of posts, with the top and bottom 10% removed.
Why median, not mean
An average is wrecked by a single outlier. One 2M-view post in an account that normally does 30K drags the mean up so far that nothing afterwards ever looks like a breakout — the very thing we are trying to detect becomes the thing that hides the next one.
The median ignores the size of outliers and only counts their position, so a previous hit does not blind us to the next.
Why the top and bottom 10% come out
Trimming removes two kinds of noise: a prior viral post inflating the floor, and posts that barely circulated at all — deleted-then-reposted content, failed uploads, or posts caught by a platform issue. Neither tells you what normal looks like.
Why 90 days
Long enough to smooth out a good or bad fortnight. Short enough that an account which has genuinely grown is measured against who it is now, not who it was last year. An account that tripled six months ago should not be judged against its smaller self.
Why five times
This is the one genuinely arbitrary number, and it is worth being honest about that. Below roughly 3×, normal variance produces false positives — posting time, a slow news day, or the algorithm having a moment. Above 10×, real and repeatable wins get filtered out.
Five sits where the signal is strong enough to be worth studying and common enough to find several per account. It is a judgement call, not a discovery.
Raw view count measures how big an audience already is. The multiple measures whether the post did anything.
Where this breaks down
Four cases where the number lies, and we would rather say so than pretend otherwise:
- New accounts. Fewer than about 20 posts in the window and the baseline is too thin to trust. We do not report multiples for these.
- Accounts that post rarely. Three posts a quarter gives a baseline built on three numbers. Technically computable, practically meaningless.
- Paid promotion. A boosted post can clear 5× on spend alone. Public data does not reliably show what was promoted, so some paid posts will appear as breakouts. We flag suspected cases where we can.
- Accounts in steep growth. If a creator is doubling every month, the trailing baseline lags reality and ordinary posts start clearing 5×. The window helps, it does not solve it.
What we do not claim
A breakout is not proof of quality, and it is not a promise that copying the structure will reproduce the result. It is evidence that something in that post worked harder than that creator's usual, which makes it worth examining. The causal story is always a judgement, and we try to label ours as such.
Figures come from public post data at time of analysis. View counts keep moving; the multiple is the more durable number, which is why we lead with it.
Questions people ask
What exactly counts as a breakout?
A post that reaches at least five times the creator's baseline, where baseline is the median views of their last 90 days with the top and bottom 10% removed.
Why not just use view count?
View count measures how large an audience already is. A 2M-view post from an account that always does 2M tells you about the account, not the post.
Is the 5x threshold arbitrary?
Partly, and we say so. Below about 3x you pick up normal variance; above 10x you filter out real wins. Five is a judgement call that balances the two.
Does this work for new accounts?
No. Under roughly 20 posts in the window the baseline is too thin to trust, so we do not report multiples for those accounts.