Hotel staffing when the occupancy forecast is wrong
· scheduling, labor cost, housekeeping
Your occupancy forecast for next Wednesday will be wrong, and no amount of care will fix that — a walk-in, a cancelled block, or a highway closure twenty miles away moves it more than your method does. What you can fix is the shape of the schedule you build against it. Most schedules are built symmetrically, as if being over-staffed and being under-staffed cost about the same. They don’t. An hour of over-staffing wastes a full hour of wage. An hour of under-staffing, if you can cover it, wastes at most the overtime premium — half that. Build for the bottom of your forecast band and hold the difference as arranged, callable hours, and the same forecast error costs you materially less.
The two errors have different prices
This is arithmetic, not opinion, and it works out the same at any wage.
Suppose housekeeping runs at $17 an hour. If you scheduled an hour you didn’t need, you pay $17 and receive nothing. The error costs the whole $17, and it is spent the moment that person clocks in — there is no recovering it later in the day.
Now suppose you were short an hour and covered it by calling someone in. You pay $17, and you get the room cleaned. The work had to happen regardless, so it was always going to cost $17. The error cost is whatever you paid on top of that: nothing at all if the person you called is under their overtime threshold for the week, or the premium — $8.50 at time-and-a-half — if they aren’t.
So the worst case of under-staffing costs half the certain case of over-staffing, and the best case costs nothing. That ratio holds at $15 an hour and at $22 an hour, because both sides scale with the wage.
Nobody schedules this way, and the reason is understandable: an unclean room at 3pm is visible and a paid idle hour is not. But the idle hour is charged every single time, and the call-in is charged only sometimes.
What a 15-point miss costs: one Wednesday at a 60-room motel
Invented property, so you can check every number: a 60-room independent, housekeeping at a placeholder $17 an hour. It has measured its own cleaning times and gets roughly 35 minutes for a checkout and 20 for a stayover. Yours will be different, and the gap between those two numbers is worth measuring at your own property rather than inherited from a post.
Wednesday’s forecast is 60% — 36 rooms occupied. Looking back over the same weekday for the last two months, the property’s forecasts have landed anywhere from about 45% to about 75%. Of occupied rooms, roughly 40% depart the following morning.
| Wednesday night | Occupied | Checkouts | Stayovers | Cleaning hours |
|---|---|---|---|---|
| Bottom of the band (45%) | 27 | 11 | 16 | 11.75 |
| The forecast (60%) | 36 | 14 | 22 | 15.50 |
| Top of the band (75%) | 45 | 18 | 27 | 19.50 |
Thursday morning’s housekeeping requirement swings across 7.75 hours — about $132 of labor a day, on one department, riding on a number nobody can pin down. Over a year that band is wider than most of the line items an owner actually spends time negotiating.
Scheduling to the middle is the expensive habit
The instinctive response to a range is to staff the middle of it. Three housekeepers on six-hour shifts is 18 scheduled hours, which sits comfortably above the 15.5 the forecast implies.
Compare it against building the base to the bottom of the band — two housekeepers, 12 hours — with the rest arranged in advance as callable.
| If Wednesday actually lands at | Middle schedule (18h) | Low schedule (12h) |
|---|---|---|
| 27 occupied | 6.25 idle hours — $106.25, certain | 0.25 idle hours — $4.25 |
| 36 occupied | 2.5 idle hours — $42.50, certain | 3.5 hours called in — $0 to $29.75 |
| 45 occupied | 1.5 hours called in — $0 to $12.75 | 7.5 hours called in — $0 to $63.75 |
Read the columns rather than the cells. The middle schedule’s costs are mostly certain — idle hours are paid whether or not anything happens. The low schedule’s costs are mostly conditional, and the condition is that the person you call is already near forty hours.
Run the three as a three-day thought experiment and treat them as equally likely — a simplification, since your own band almost certainly isn’t flat — and the middle schedule costs $149 to $162 across the three days. The low schedule costs $4 to $98. Its worst case is cheaper than the other’s best case.
Building a schedule for the band instead of the number
Five things, in order, and only the last one is software.
- Find your own band before you use anyone else’s. Pull eight weeks of the same weekday and write forecast beside actual. Most operators have never done this and are startled by the width. A band you measured is worth more than a rule of thumb, because the whole argument depends on where the bottom of yours actually sits.
- Set the base at the bottom of the band, not the middle. This is the only structural change, and it is the one that does the work.
- Name the flex hours before you need them. Who, which hours, how they are reached, agreed at the start of the week — two names per shift, not one, because the value of this plan is entirely in whether the phone gets answered.
- Put the trigger on a number. “If the board shows more than 40 rooms on the books by 3pm, call the second person for tomorrow.” A written threshold beats a judgment call made at 7am by whoever is at the desk, and it means the call goes out the night before rather than an hour into the shift.
- Check week-to-date hours before you dial, not after. Who you call is what decides whether the fix costs the premium or nothing at all — and that is the single piece of information least likely to be in the head of the person making the call.
That last step is where a system earns its place: ProfitClock’s daily labor flash email reports each property’s hours against the overtime thresholds you set, so the person choosing who to call can see who is at 31 hours and who is at 38 — start a free trial if you want to watch a real week accumulate before you rely on it. Reading a daily labor report covers what else is worth looking at on it.
Where this breaks down, and where we are not the answer
Four honest limits, because this is a money argument and money is not the only thing at stake.
A call-in that nobody answers is not cheap, it is unbounded. The entire asymmetry rests on being able to reach somebody. If the room isn’t ready when the guest arrives, the cost is a bad arrival, a moved reservation, or a review, and none of those appear on the ledger above. That risk is worst at the top of the band, where there are also more arrivals waiting.
Flexibility taken from the same person every week is not flexibility. If “callable hours” means calling one reliable employee on their day off repeatedly, you have not built a flexible schedule; you have built resentment, and the eventual cost is a vacancy. Rotate it, agree it in advance, and treat a declined call as a normal outcome rather than a problem.
In some jurisdictions a short-notice change to a posted schedule carries a premium of its own, which narrows this gap and can reverse it. The arithmetic above assumes no such rule applies. Predictive scheduling laws is the place to start, but the rules vary by city and state and they change — check what applies where you operate before leaning on any of this, and treat none of it as legal advice.
We do not forecast occupancy and this post is not about getting the forecast right. ProfitClock does not predict demand and has no plans described here to. The premise of everything above is that the forecast stays wrong and the schedule stops pretending otherwise.
Start seeing what labor actually costs
Set up your property, put staff on the clock, and read tomorrow morning’s labor number.
14-day free trial · No credit card · Cancel anytime