What ready actually means, and who says it
The final state of every room in this system is set by a person looking at it, and resisting the temptation to infer it is what makes the whole thing trustworthy to reception.
Key takeaways
- Cleaned and inspected are different states with different meanings.
- Neither is ever set automatically, by a timer or by anything else.
- A maintenance issue found mid-clean is a different workflow and needs one tap.
- Reception sees the state and the time it was set, not a prediction.
- Inspection rates can fall over time; measure them.
Two states
| State | Set by | Means |
|---|---|---|
| Dirty | Checkout, or the day rolling over | Needs attention |
| In progress | The housekeeper starting | Somebody is in there |
| Cleaned | The housekeeper finishing | The work is done, in their judgement |
| Inspected | A supervisor, or a sampled check | Verified, and sellable |
| Out of order | Anybody who finds a fault | Not sellable until fixed |
The distinction between cleaned and inspected matters most at the moment reception is under pressure. A guest is waiting, a room shows cleaned but not inspected, and somebody has to decide whether to give it out. That is a judgement, and it is only possible if the two states are separate.
Collapsing them into one loses that option in both directions: either every room waits for an inspection that may be twenty minutes away, or no room is ever verified.
Never inferred
A room does not become cleaned because a timer expired, because the housekeeper’s list moved on, or because the average clean time has passed. It becomes cleaned when somebody says it is.
This sounds obvious and the pressure to infer it is real, because housekeepers forget to mark rooms and reception wants current information. The correct fix is making the marking take one tap, not guessing.
The maintenance issue
- Compute
- App integration
- Machine learning
- Security & identity
- Analytics
- Front-end & mobile
A job, not a note
Maintenance issues reported as free text in a housekeeping system go nowhere. Creating an actual job, in whatever the maintenance process is, is what makes the report worth making, and a housekeeper who reports three things that are never fixed stops reporting.
This is the same principle as the deferral record in Day 118: the reporting mechanism only survives if the reports visibly lead somewhere.
What reception sees
- Compute
- Security & identity
- Management
- Analytics
- People
The absence of a prediction is the same principle as the delivery exception handler in Day 116: an estimated time will be passed on to a guest as a promise, and a promise based on an average is broken often enough to be worse than the honest answer.
Measuring inspection
Inspection rates fall over time, quietly, because inspecting is the thing that gets dropped on a busy day. A hotel that believes it inspects every room and actually inspects sixty per cent of them has a quality process that exists mainly in a policy document.
Counting is free once the states are separate: what proportion of rooms went from cleaned to inspected, and how long the gap was. Both numbers are worth putting in front of somebody monthly.
What this system does not do
It does not set states automatically, it does not predict ready times, it does not override do-not-disturb, and it does not rank people by speed. The last one is worth stating: the timing data exists to allocate work fairly, and turning it into a productivity ranking would make everybody stop marking rooms accurately, which destroys the data the fairness depends on.
Next: what all of this costs to run.
All posts