Analytics for Community Sites and Forums

Community health is famously vibes-based — the forum 'feels' alive or dying long before anyone quantifies it. But the vibes have measurable correlates, and a community site (forum, Discourse instance, member hub) can track them without violating the trust that makes communities work. The line to hold: measure the room, never surveil the members.

The four health signals

1. Contributor ratio, trended: visitors who post vs visitors who lurk. Track posting as an event (thread_created, reply_posted — no content in properties) and watch the ratio monthly. Healthy communities run 1–10% contributors; a falling ratio with stable traffic is the silent-death pattern, visible quarters early.
2. New-contributor flow: first-post events per week. Communities die from the entrance, not the middle — veterans keep posting while newcomers stop starting. The first_post event cohorted by join month (cohort basics) is your immigration statistic.
3. Thread engagement depth: replies per thread, read depth on long discussions — whether conversations actually conclude or just stop.
4. Return rhythm: the gap between member visits, stretching or holding — retention logic applied to belonging.

Where growth comes from

Communities grow by being cited: a thread answers someone's Google query (organic landing on deep threads), a member shares a discussion into a group chat (dark traffic patterns), a partner links you. Referrer data per landing thread tells you which content earns the community its newcomers — and which threads deserve curation into permanent resources.

The trust constraint

Members extend communities a higher privacy expectation than they extend brands. Cookieless, aggregate measurement matches it: no persistent identifiers, no behavior profiles, no consent wall at the door — and the no-content-in-properties rule keeps even your event stream clean of what people actually wrote. You are measuring whether the campfire is warm, not recording the conversations around it.