Every owner suspects their community runs on a handful of people. Very few know how small that handful is, and almost nobody knows that it is the same size in every community they have ever been in.
How few people write most of the posts?
Across 399 Skool communities with a complete public post history, a median of 2.15% of members write three quarters of all posts.
In a 200-member community that is about four people.
Four. Not forty. If those four went quiet next month, three quarters of everything anyone reads in your community would stop.
The spread is narrower than you would expect too: a quarter of communities concentrate harder still, at 0.92%, and a quarter are more distributed, at 4.05%. There is no band of communities where participation is broadly spread. The concentrated case is the normal case.
Does it stay that concentrated as a community grows?
This is the part that makes it a law rather than an observation. The number of people scales with the community. The share does not.
Members grow 62-fold across those bands. The engine grows about 50-fold. The proportion barely moves.
The obvious objection is that bigger communities are also older, with more posts, and more posts mean more chances for someone to appear in the author list. That was tested directly. Holding post volume constant between 100 and 600 posts, the engine share is 2.12% in small communities (median 248 members) against 1.73% in large ones (median 1,332) — five times the size, and the share moves by four tenths of a point instead of by a factor of five.
The formal version: the log-log slope of engine size against member count is 0.896 (95% confidence interval 0.819 to 0.972, R-squared 0.57). Near-proportional, and member count explains engine size about twice as well as post volume does.
The limit worth stating: the figure is computed only in communities where at least twenty distinct people have posted and the crawl returned a complete history. In a community with eight posters, the top ten is everyone, and the statistic would be meaningless.
Is 2.15% the real number, or just the one that was chosen?
It is threshold-dependent by construction, so it publishes as a ladder rather than as a constant of nature:
Half the content comes from the loudest 0.59% of members. Three quarters from 2.15%. Nine tenths from 5.16%.
The 75% figure is the middle of a smooth ladder, not a cherry-picked cut. Whichever rung you prefer, the story is the same: a very small group carries a very large share, and the curve is steep at every point along it.
Is the biggest poster usually the owner?
Yes, and this is the finding most owners take hardest.
In 77.4% of communities the single biggest poster is the owner, whose median rank among all posters is 1. The median owner writes 25.8% of everything posted in their own community.
It gets sharper by size:
In a small community the owner writes over half of everything. That is not a community yet — it is a broadcast with a comment section, and calling it something else does not change how it behaves when the owner takes a week off.
At 5,000-plus the owner writes about a tenth and the members genuinely carry it. The gradient between those two states is, more or less, the whole job of building a community.
Does the law survive if you remove the owner? It does, restated. Setting the owner's posts aside entirely, a median 3.58% of members still carry three quarters of what members write. So this is not merely an artefact of counting the owner. The concentration is real among members too — the owner just sits on top of it.
What should you do with a 2% engine?
Three things follow, and the first two are uncomfortable.
Know who they are, by name. Four to nine people in most communities. That is a list you could write out. Most owners cannot, because the tooling surfaces recent activity rather than sustained contribution — a person who posts thoughtfully once a fortnight is invisible in a feed and enormous in the data. How to find your most engaged members is the practical version of that job.
Treat losing one of them as a real event. If four people carry three quarters of your community, a single departure is a measurable content loss, not a routine bit of churn. That is a very different risk profile from the one most owners carry in their heads, and it argues for knowing when one of them goes quiet rather than finding out afterwards.
If the owner is more than half the output, that is the number to move. Not total post count — the owner's share. Growing that community means other people writing more, which is a different activity from you writing more. The measurement that tells you whether it is working is your own share falling while total volume holds.
And a fourth, which is really a warning: do not try to convert the 90%. The engine law holds across a 62-fold range of community sizes and every threshold tested. A community where 30% of members post regularly does not appear in this data because it does not appear anywhere. Aim at moving the second tier — the people who have posted once and stopped, which is what the one-and-done rate measures — and let the majority read in peace.
These figures come from The State of Skool 2026, an independent measurement of 18,412 Skool communities, including 399 with a complete public post history. See the full report for the full concentration analysis and 21 other findings.