<b>Search as a hidden retention variable</b>

<b>Search as a hidden retention variable</b>

We rarely list "findability" as a retention factor. The platform comparison suggests it's quietly load-bearing.

<b>What the data shows</b>
In surveys of why members stop opening a community, "I couldn't find the thing I came back for" recurs more than its reputation suggests — often cited alongside, not below, "too much noise." Communities organized around searchable, persistent knowledge (pinned indexes, wikis, well-tagged threads) show lower long-tail churn than pure-chat servers of similar size.

<b>Why it happens</b>
Much community value is <i>reference</i> value, not live-conversation value. A member who returns to re-find an answer and fails learns the place is unreliable. Discord's search is notoriously weak relative to its message volume; Telegram's in-chat search is faster but its threading is weaker, so the knowledge fragments differently — Discord buries answers in channels, Telegram buries them in scroll.

<b>The caveat</b>
Self-reported reasons for leaving are unreliable — people rationalize. And "organized" servers correlate with competent owners who do many things right, so search may be a proxy, not the cause.

Open question: if you instrumented failed in-server searches, would the failure rate predict 90-day churn — and would fixing search move the number?