199 AI films, 188 million views: 10% of them carried 90% of it
Reach follows a power law, not a bell curve — in our data, the top 10% of films holds 89.6% of the 188M views, so growth comes from more attempts, not a better median video.
What did 188 million views actually look like when we broke it down?
We pulled the numbers on everything we've published: 199 films, 188 million views total. Split them into tenths by view count and you get a very different picture than "average performance."
The top 10% — about 20 films — holds 89.6% of all 188 million views. The other 179 films split what's left. The median film, the one sitting right in the middle of the pack, got 36,000 views. The single biggest film got 61.6 million on its own — more than a third of everything we've ever put out, in one video.
- 199 films published, 188M total views
- Top 10% of films = 89.6% of all views
- Median film: 36,000 views
- Best single film: 61.6M views
Why is "average views per video" a useless number here?
Divide 188M by 199 films and you get roughly 945,000 views per video — a number that describes almost none of the 199. It's dragged up entirely by the outliers.
The median (36k) is a much better description of what a typical film does, and it's more than 25x smaller than the average. When your average and your median are that far apart, you're not looking at a bell curve. You're looking at a power law: a few results dominate, most cluster low, and the gap between "usual" and "best" is enormous.
What actually separates the top 10% from everything else?
We don't have a clean formula, and anyone who tells you they do is selling something. What we can say: the films that broke out tend to share a hook that reads in the first two seconds with the sound off, a character viewers already recognize from a prior video, and a beat that still plays out of context on a feed.
None of that guarantees a hit — plenty of films with all three still landed at the median. We don't run the platform's distribution algorithm, and we don't know exactly why one clip catches a wave and another doesn't. That's the honest answer, and it's also the point: if the winners were fully predictable, this wouldn't be a power law.
What should you actually do differently because of this?
Stop asking "how do I make the median video better." Turning a 36k-view video into a 50k-view video doesn't change your outcome much — you're still nowhere near the results that matter.
Ask instead: how many attempts can I afford, and how do I make a hit compound instead of getting lost. That means budgeting for a run of films rather than one hero video, keeping a consistent, recognizable character across the run so a breakout pulls attention back to the others, and judging most individual films as break-even assets rather than bets that each have to win on their own.
- Budget for volume, not a single "best" video
- Keep a recognizable character across the run — a hit should lift the others, not just itself
- Expect most films to land near the median — that's normal, not underperformance
- Judge success at the run level, not per video
When is this the wrong lesson for your brand?
If you only have budget for one video and it's tied to one date — a launch, an event, a single campaign — betting on volume doesn't help you, because you don't get to average out over a run.
Plan for the median outcome, something closer to 36k views, and make sure that single video earns its cost as an asset even if it never breaks out — as an ad, a website piece, something you can point people to. Don't buy one video expecting it to be the 61.6M-view outlier. Statistically it won't be, and a studio that tells you otherwise isn't being straight with you.
Common questions
Does a 36,000-view median mean most of our videos are failing?
No. That's the normal, expected result in a power-law distribution — most attempts land near the middle and a small number carry the reach. Judging each film against the outliers is the actual mistake.
How many videos do we need before we're likely to get a breakout?
We don't have a magic number, and anyone who gives you one is guessing. More attempts improve your odds, but no volume guarantees a hit — treat each film as a chance, not a formula.
Should every script be engineered to try to go viral?
No — keep a quality floor, a clear hook, a consistent character, a story that holds up, but don't over-engineer every film for virality. Most won't hit regardless, and forcing it usually reads as try-hard.
If most videos only get the median, is it worth paying for a single video?
Depends what it's for. As a standalone brand or ad asset, yes — a 36k-view video that runs as a website piece or paid creative still earns its cost. As a solo bet on going viral, no.