How to audit your server's onboarding funnel in one afternoon

How to audit your server's onboarding funnel in one afternoon

Most server owners measure joins. Almost none measure the path from join to first message — which is where the leak actually is. Here is a reproducible audit.

What to instrument
— Pull the last 500 joiners with timestamps
— Tag each: never posted, posted once, posted 3+ times in week one
— Record time-to-first-message for the active cohort

What the data tends to show
In join cohorts I've examined, roughly 60-70% never send a single message, and of those who do, the median first message lands inside 48 hours. If your median is days, your gating or first-touch is too slow.

The funnel stages to score
— Landing (do they see something actionable, not a wall of rules?)
— Verification (each extra click sheds people)
— First prompt (is there an obvious place to say hi?)
— First reply received (did anyone respond?)

That last stage is the strongest predictor. A new member who gets a reply within an hour retains far better than one who doesn't — though this is correlational, since engaged members may simply post in livelier moments.

The caveat
A single cohort is noisy. Run the same audit on three monthly cohorts before changing anything, or you'll chase seasonal noise.

Open question: if first-reply latency is the lever, should onboarding be a bot's job at all, or a roster of humans on shift?