Almost every community owner eventually discovers that the overwhelming majority of their members never say anything, and concludes something has gone wrong.
The 90-9-1 rule is the reason to stop and check that assumption before acting on it.
Where does the 90-9-1 rule come from?
Jakob Nielsen described it in 2006 in an article on participation inequality for Nielsen Norman Group. It was an observation drawn across many online communities, not a prescription.
The split he described:
| Group | Share | What they do |
|---|---|---|
| Lurkers | ~90% | Read or observe. Never contribute. |
| Intermittent contributors | ~9% | Contribute occasionally, when something prompts them. |
| Heavy contributors | ~1% | Account for most of the content in the community. |
The important part of Nielsen's original piece is often dropped when the rule gets quoted: he also gave different splits for different platforms. He put blogs closer to 95-5-0.1, and Wikipedia at roughly 99.8-0.2-0.003, where over 99% of users never edit anything.
So even in the source, the numbers move. What holds is the shape — a small minority produces most of what everyone else consumes.
Is 90-9-1 still accurate?
Treat it as a shape rather than a measurement. It was published in 2006, before most platforms people now use existed, and it was never derived from a controlled sample in the first place.
Three things are worth holding in mind:
It varies enormously by platform design. A place that makes contributing easy and low-stakes gets more contributors than one where posting feels like publishing. That is a design fact, not a fact about human nature.
It varies by size. Smaller, more intimate communities generally see a larger share of members contribute than a forum with fifty thousand members. If you have eighty people who all know each other, expecting 90% silence is probably too pessimistic.
It varies by purpose. A support community where people arrive with a specific problem produces a different curve from a community people join to learn from an expert. In the second, quiet consumption is often exactly what the member came for.
None of which makes the rule useless. It is a useful prior — a reasonable expectation to hold before you have your own data, and a reason not to panic.
What does 90-9-1 actually mean for my community?
It means that widespread silence is the normal condition of an online community, and on its own it is not evidence that anything is broken.
That reframing matters, because the alternative interpretation — nobody cares, this isn't working — leads to bad decisions. Owners who believe silence is failure tend to post more frequently, ask for engagement more insistently, and eventually message everyone at once. All three usually make things worse.
What the rule does not mean:
- It is not a target. There is no version of this where you should aim for 90% lurkers.
- It is not a ceiling. Communities routinely beat it, particularly small ones.
- It is not a reason to ignore your quiet members. Lurkers are not worthless; many are paying, and many are getting exactly what they came for.
- It is not a benchmark you can be judged against. Your community, your topic, your size, your design.
How should I use it, then?
Use it as a sanity check, once, and then stop using it. The useful comparison after that is against your own community over time.
Here is the practical sequence.
- Work out your own split. What share of your members have ever posted or commented? What share do it regularly? You do not need a tool for this — a rough count is enough for the purpose.
- Compare it to the shape, not the number. Is a small minority producing most of the activity? Almost certainly yes. That is normal.
- Then ignore the rule entirely and watch your own number over time. Whether your contributing share is rising or falling tells you far more than whether it currently sits at 8% or 14%.
- Ask what would change it. Usually the answer is about how easy you make contributing, not about how much you exhort people to contribute.
That third step is where the value is. A cross-community average cannot tell you whether your community got better or worse last month. Your own trend can.
The number that matters is not the gap between you and 90-9-1. It is the direction your own contributing share has moved since you last looked.
Why does measuring this get confusing?
Because "engaged" means different things depending on what is being counted, and the flattering definition is the one most tools reach for.
If a member is counted as engaged for loading the page, your participation figure will look strong and will tell you almost nothing. If a member counts only when they comment, reply or vote in a poll, the number is lower and considerably more useful, because it measures people who actually did something.
This is worth pinning down before comparing yourself to any benchmark, including this one. Nielsen's 90% lurkers are defined as people who read but do not contribute — so a visit-based metric is measuring a different population entirely. Participation rate versus engagement rate covers the distinction, and what counts as a good engagement rate covers what to do with the answer.
So should I try to convert my lurkers?
Some of them, deliberately, and not all of them.
A member who reads every post, renews every month, and never comments is not a problem to be solved. They are a satisfied customer with a different usage pattern from yours. Pushing them to perform engagement they do not want serves your metrics, not them.
The lurkers worth attention are the ones where silence signals something you can act on: people who joined recently and have not found a way in, people who used to contribute and stopped, and people who are visibly active in other communities while silent in yours. That last group is the most reachable of the three, because the habit already exists.
How to activate Skool lurkers covers the approach, and the members you are ignoring covers finding the third group specifically.
What do the other versions of this rule mean?
You will run into several near-relatives of 90-9-1, and they are mostly the same observation rescaled.
The 1% rule is the same idea with the middle group dropped: roughly 1% of an online population creates, the rest consume. It is the version most often quoted in journalism, and it is blunter than useful for a community owner, because the 9% is the group you can actually move.
The Pareto principle — 80% of effects from 20% of causes — gets applied to communities constantly and is a poorer fit. Participation curves in online communities are considerably steeper than 80-20. If you use Pareto as your mental model you will expect a far broader contributing base than you will ever see, and read the gap as failure.
Zipf distributions are the underlying mathematical shape. Nielsen's original piece notes that plotting activity per user produces a curve of this family — a small number of very high contributors, a long tail of near-zero. This is the part that genuinely transfers, because it is the same shape across platforms even when the specific percentages are not.
The practical upshot: expect a steep curve, expect most members to be near the bottom of it, and do not treat any particular set of round numbers as the standard you are failing to meet.
What is the honest summary?
The 90-9-1 rule is a twenty-year-old observation about a pattern that keeps recurring, and it is genuinely useful for exactly one thing: talking yourself out of panic when you notice most of your community is silent.
It is not a benchmark, it does not describe your community in particular, and the exact percentages should not be treated as meaningful to one decimal place. Nielsen himself gave three different splits in the original article.
Use it once, to calibrate your expectations. Then measure your own community and watch which way it moves.