<b>Why DAU/MAU lies about community health</b>

<b>Why DAU/MAU lies about community health</b>

The stickiness ratio (daily over monthly actives) is the default health metric. For communities specifically, it can point the wrong way.

<b>What the data shows</b>
Product-led servers chase a high DAU/MAU — 50%+ signals a daily habit. But across a sample of 600 hobby and creator Discords, the healthiest-by-survival servers clustered at 15–25% stickiness, not 50%+. The high-stickiness outliers were disproportionately support and trading servers — high daily need, high churn once the need resolved.

<b>Why it happens</b>
A topic community isn't a productivity app. Members orbit it weekly, not daily. Demanding daily returns means competing with the member's job, which a hobby loses. The 50%+ servers were often <i>extractive</i> use cases: people show up because they have a problem, not because they belong.

<b>The caveat</b>
Survival isn't the only goal, and the sample skewed toward English-speaking servers under 50k. A 25% ratio in a 200-member server and a 200k-member server are not comparable phenomena. DAU/MAU also can't distinguish ten power-users from a thousand drifters.

Telegram makes this worse to measure: "online" is noisy and channel views aren't the same as engagement.

Open question: should community health be measured in <i>weekly cohort return rate</i> instead of daily stickiness — and if so, what's the equivalent of a "good" number?