Most restaurant schedules are written from a template and adjusted by feel. That produces two failures at once — overstaffed hours you pay for and understaffed hours that cost you covers — and both are visible in data you already have.
Schedule to covers, not to days
The template says four servers on a Friday. Your actual Friday is quiet at five, heavy from seven to nine, and tailing off by ten.
Four servers across that whole span means you are overstaffed for three hours and correctly staffed for two. The fix is not fewer people, it is different start times — staggered starts so the floor builds and thins with the room.
Pull covers by hour, by day of week, over eight weeks. The shape will be strikingly consistent, and it will not match your template.
The number to schedule against
Pick a ratio and schedule against it rather than against a headcount.
Front of house: covers per server per hour, at the service level you want. A high-touch room supports fewer covers per server than a bistro. Work out what yours actually is on a night that went well, and use that.
Kitchen: covers per hour against station capacity, which is usually a hard ceiling rather than a ratio — the pass can do what it can do.
Then schedule so that expected covers divided by your ratio gives your headcount for each hour, and set starts and finishes accordingly.
The mechanics that save the most
Staggered starts. The single biggest saving available. Not everyone needs to be there at five.
A cut order decided in advance. Who goes first when it is slow, decided before service, applied without a discussion at nine o'clock every night.
One person on call, where your rules allow it — arrangements around on-call and reporting time vary by jurisdiction and some places regulate them tightly, so check yours rather than assuming.
Cross-training. Somebody who can move between stations is worth more than a specialist on a quiet night.
Where scheduling rules bite
This is an area with real legal weight and genuine local variation.
Overtime thresholds and rates are set by federal and state law, and scheduling someone into overtime is a cost decision you should be making deliberately rather than discovering in payroll.
Predictive scheduling laws exist in a number of cities and states — requiring schedules to be posted a set number of days ahead, with penalties for late changes. If you operate somewhere with one, it constrains exactly the last-minute adjustment most restaurants rely on.
Breaks and rest periods are regulated, and the rules differ substantially between states.
Minors have their own hour restrictions, which are stricter and enforced.
None of that is something to work out from a blog post. Find out which of these apply where you operate, and build the schedule inside them rather than discovering the boundary afterwards. What is labour cost and what counts as part of it covers the cost side.
The trade against food cost
Scheduling is one half of prime cost, and the halves move against each other.
Cutting labour by buying more prepped product raises food cost. Doing more in-house lowers food cost and needs more skilled hours. Neither number alone tells you whether you improved anything, which is the argument for watching them together. Prime cost has the calculation.
What the schedule cannot fix
A restaurant that is overstaffed because it is quiet does not have a scheduling problem. It has a demand problem, and cutting hours treats the symptom while making the service worse.
Two things worth separating before you cut:
Are you quiet because fewer people want to come, or because they could not get through? Reservation calls that ring out during service look exactly like weak demand in every report you run, because the call leaves no trace anywhere. A restaurant cutting staff in response to covers it never received is making the problem worse. What missed calls cost has the arithmetic.
Are your quiet hours genuinely quiet, or badly promoted? A dead Tuesday is a scheduling problem in the short run and a marketing one in the long run.
The weekly habit
Pull covers by hour for the week just gone. Compare to what you scheduled. Look at the two or three hours where the gap was largest, in either direction.
Adjust next week's starts by thirty minutes where the gap was consistent. Small, weekly, and after a couple of months the schedule fits the restaurant rather than the template.