# Turnout Desk

> A capacity planner for meetups, hackathons, workshops and community conferences that lose some of their
> registrations. Enter the room's hard capacity, the price posture, registrations, the waitlist, the headcount lock
> date and last edition's turnout; the page works out the registration ceiling with its arithmetic, checks the room at
> the top of the show-up band, picks the waitlist rule and drain date and the day-of lane, and gives catering the
> expected-show-up number. A metered run writes the capacity plan. Built on the SkillSafe skill
> @samber/event-no-show-management (samber/dev-event-organizer-skills, MIT). Figures from the MLH hackathon organizer
> guide and Event Tech Live, each with its scope; not affiliated with either.

https://turnout-desk.skillsafe.ai/

## What is free
- The show-up figure for the price posture, named with its source: free registration - about 50% show up (MLH guide trendline across 323 events), free band 40-60% no-shows; paid - 90-97% of tickets attend; a nominal fee reads against the paid band as an upper bound. An own ratio from last edition (with its 95% interval) only for a recurring free event whose headcount locks before doors.
- The registration ceiling in five lines: capacity (plus overflow only with written venue approval), aimed attendance, the rate and its source, the ceiling (double and never past double on a free event; capacity on a paid one), the catering number with its buffer.
- The occupancy gate: what the ceiling brings at the top of the band against the limit, and the exact binomial chance of going over even at the planned rate.
- Registrations against the ceiling (close the form now / sold past the room), the waitlist rule (signup order on a named date; proximity only when someone can sort a long list), the drain date before the headcount lock, a 48-hour claim window (self-set), the day-of lane (walk-in lane, or a pre-doors release to the waitlist when nobody can walk up) and its expected volume.
- The deposit question answered as a refusal, comped seats counted separately, and an after-event debrief (show-up among registrants with its interval, promotion yield, covers left over) that becomes next edition's own ratio.

## The metered lane
One task, `plan`, on gpt-terra, metered per run. Input string fields: `task`, `event` (format, posture, dates, recurrence, walk-up population, channel, sorter, form status), `numbers` (the five-line arithmetic, registrations, the gate, catering, waitlist rule and dates, lane and volume, last edition) and `facts` (method and every finding). The reply is one JSON object: `summary`, `ceiling` (registrations, arithmetic, closes), `gate`, `waitlist` (rule, drain_date, claim_window, publish), `walk_in` (lane, volume, opens), `catering` (number, buffer, note), `deposit`, `handoffs`, `measure`, `caveats`. The page holds the ceiling, catering number, buffer, rule, drain date, lane and volume to its own and refuses numbers not in the input.

## Links
- API tutorial: https://turnout-desk.skillsafe.ai/api.html
- Source skill: https://skillsafe.ai/skill/@samber/event-no-show-management