The morning paragraph, the ETA, and the demand
What the queue predicts for today, which arrivals are at risk, the move that helps — and what the store will need this week.
The morning paragraph (*Housekeeping dashboard → Morning*): "84 departures and 61 arrivals today. 6 attendants on the floor (1 away). Current manpower predicts 71 of 84 rooms ready by 14:00; 13 are at risk. Moving 2 rooms cuts delayed rooms from 13 to 4." Every number in it is arithmetic you can check: how long a clean takes is the median of what *that attendant* took on *that kind of visit* in *that kind of room* over the last ninety days (falling back to the room type, the kind, then the target — the screen says which); the ETA of every room is that attendant's queue run one visit after another from now; a room is at risk when its ETA is after the guest's expected arrival — the waiting guest's now, a booking's check-in time; a recommended move is the same run again with the room given to someone else, kept when it cuts the delay. *Apply* makes the moves through the ordinary reassignment; the story says what was expected.
Every prediction is kept the hour it was made, so *Forecast vs actual* is a plain comparison: the median error in minutes, the share within 15 and 30 minutes, and how many predictions rested on the attendant's own history. A learned model replaces the medians only when it beats them here.
Demand (*Demand*): what the store issues per room served over the last thirty days, times the bookings ahead (315's forecast), is each item's need by day; on hand against it is days of cover; *reorder by* is the day cover runs out less the lead time; the quantity is cover to the horizon (both under *Settings*). Linen sets, towel sets and guest laundry for the week sit beside it.