Nothing in a restaurant P&L looks more like waste than the overtime line. It sits there every week, ticking up two hundred dollars here, four hundred there, one bad Saturday and suddenly you are looking at an extra fifteen hundred that did not have to happen. Every general manager promises to fix it. Two weeks later it is back.
I have watched this loop run in a dozen different multi-unit groups. The last time I ran the numbers hard was inside a five-unit Bay Area group where OT was running 4.1 percent of labor dollars, roughly $18K a month across the region. Six months later it was under 1 percent. Nobody was cut. Nobody was capped. What changed was the cover system.
Why overtime is a cover problem, not a wage problem
Here is the sequence, every time. A scheduled employee calls out at 2pm on Friday. The manager needs a body on the line at 5pm. The manager pulls up the schedule on their phone, scrolls the roster, and calls the first person they know is reliable. That person is reliable because they always pick up shifts. They always pick up shifts because they need the hours. Which means they are already at 36 or 38 hours for the week.
Now that person works the Friday dinner shift. They cross 40 hours somewhere around 8pm. Everything past that is time and a half. Nobody made a bad decision. The manager solved a real problem. The employee earned the money. But the OT line just went up by $120 and it did not have to.
Fig. 1 · OT compounds because the same short cover pool gets tapped repeatedly.
The reason the OT keeps coming back is that the cover pool is too narrow. Same three or four employees keep picking up shifts, keep hitting 40 hours mid-week, keep tipping into premium time for the back half. It is not a wage negotiation. It is a system that has only one answer to the "who covers this shift" question and that answer costs 50 percent more.
The real cost of an overtime hour
Operators tend to underestimate what OT actually costs, because they only count the wage premium. That is the smallest of the three costs.
The visible cost: 1.5x the wage
On a $20 base wage, an OT hour is $30 to the employee. That is what the P&L shows.
The invisible cost: payroll tax on the premium
Employer payroll taxes (FICA, SUI, workers comp) apply to the OT premium too. On that same $20 base, the fully burdened cost of an OT hour lands around $34 to $36, not $30. Over a year on a group running $150K of OT dollars, that is another $20K to $25K in true cost the operator did not budget for.
The hidden cost: burnout and turnover
The person who keeps getting called in on their day off will eventually stop answering the phone. Then they will start missing shifts of their own. Then they will quit. The turnover cost on a hourly restaurant employee lands somewhere between $1,500 and $3,000 all-in, counting recruiting, onboarding, training, and the 90-day productivity ramp. If the OT is driving even a small increase in turnover, the hidden cost dwarfs the wage premium.
The employee who picks up every shift is not saving your labor budget. They are the labor budget's biggest risk. When they burn out, you lose the labor budget and the cover pool at the same time.
The four-step system that kills the creep
None of this requires new software. Any decent scheduling platform (7shifts, HotSchedules, When I Work, R365, whatever you already have) will do everything below. What most operators are missing is not the tool. It is the standing operating rhythm.
Step 1: The mid-week 35-hour alert
Set up an alert that fires every Wednesday at 3pm for any hourly employee already past 35 hours for the week. That gives the manager 48 hours to reassign the remaining Thursday, Friday, Saturday, Sunday shifts before OT is committed.
Wednesday afternoon is the right trigger point because it is late enough that most of the mid-week callouts have already happened, and early enough that there is still runway to move shifts. Friday morning is too late. By Friday the employee is already scheduled and the shifts are already sold.
Step 2: Build a real cover roster
Every unit needs a written list of employees who are approved and willing to cover shifts across day parts, with:
- Their maximum weekly hours (some employees will only work 25, some will work 40).
- Their phone number and preferred contact method.
- The stations they are trained on.
- Their day-off restrictions.
This list gets updated every 30 days. Without it, the manager defaults to whoever they can remember at 2pm on Friday, and the person they remember is the person who always says yes, and the person who always says yes is the person already at 38 hours.
Step 3: Cross-train two positions per employee
This is the highest-impact move in the whole system. When every hourly is trained on only one station, the cover pool for each station is small. When every hourly is trained on two stations, the cover pool for each station roughly doubles.
Cross-training happens on slow shifts, Tuesday and Wednesday mornings, two hours per employee per week under supervision. Over 12 weeks each employee has a working second station. Never train during Friday or Saturday service. That is when the value of the cross-training gets spent, not built.
Step 4: Review the OT line every Monday
In the weekly P&L review, OT dollars go up on the whiteboard first, before anything else. The general manager names last week's number and the specific root cause: "Maria hit 43 hours because Jose called out Saturday and I did not have a Sunday cover." Not "we had OT again." Root cause every time.
If the same root cause appears twice in a row, it escalates: the cover roster is broken, the cross-training is behind, or the same employee is being over-tapped. The escalation forces a fix instead of a promise.
The math on a real unit
Take a full-service restaurant doing $200K a week in sales with labor running at 32 percent of sales. That is $64K a week in labor dollars. If OT is running at 4 percent of labor, that is $2,560 a week, or roughly $133K a year on that unit alone.
Killing the OT creep down to 1 percent (which is realistic, not aspirational) recovers $1,920 a week, or roughly $100K a year on that unit. On a five-unit group, that is $500K a year.
Framed as a percent of sales: reducing OT from 4 percent to 1 percent of labor on a 32 percent labor line pulls labor cost from 32.0 percent to about 31.0 percent. One point of labor on a $200K weekly unit is $2K a week, $104K a year. Across five units, that is over $500K a year, recovered without cutting a single scheduled hour.
Sample math, one unit
Weekly sales: $200,000 Labor at 32%: $64,000 OT at 4% of labor: $2,560/wk OT at 1% of labor: $640/wk Weekly recovery: $1,920 Annual recovery per unit: ~$100,000 Five-unit group: ~$500,000
The 90-day rollout, week by week
None of this happens in a memo. The system installs over about 12 weeks and gets fully absorbed in the general manager's operating habits by week 16. Faster rollouts do not stick because the cover roster and cross-training pieces need real time to build.
Weeks 1 to 2: instrument
Turn on the mid-week 35-hour alert. Print the current cover roster (which will be incomplete and probably out of date). Pull the last 12 weeks of OT by day, by employee, by unit. This is the diagnostic baseline. Nothing changes yet.
Weeks 3 to 4: rebuild the roster
Sit with every hourly for 10 minutes and update their availability, phone number, maximum hours per week, day-off restrictions, and stations they are trained on. This meeting is also where you signal that the roster is a real tool, not a piece of paper on the wall. Once complete, publish it on the manager's phone and in the break room.
Weeks 5 to 12: cross-train
Two hours a week per hourly on their second station, Tuesday or Wednesday morning under supervision. Not during service. By week 12 every hourly has a working second station and the cover pool has effectively doubled.
Weeks 13 to 16: measure and reinforce
Weekly OT number goes up on the whiteboard first at every Monday P&L review. General manager names last week's number and the root cause. If the same root cause appears twice, escalate. By week 16 the OT line is quiet and staying quiet.
The salaried manager side of the same problem
Salaried management overtime does not show up as OT on the P&L (salaried is fixed cost), but the burnout side of the equation is worse. A salaried general manager working 65 hours a week for 12 weeks in a row will not be at the unit in six months. Their replacement cost is $8K to $15K plus a 90-day productivity ramp.
The same fix applies at the salaried level: build the second-in-command bench, cross-train assistant managers on tasks the general manager currently owns alone (deposit reconciliation, ordering, scheduling), and put an alert on any salaried manager whose punch record shows more than 55 hours in a week. Not because 55 hours is a violation. Because 55 hours a week for 12 weeks in a row is the pattern that predicts a resignation.
Two mistakes I see managers make
Capping employee hours
The instinct is to write a policy: "no employee shall exceed 38 hours in a workweek." This punishes the employee for a system failure. Worse, it teaches the employee to stop trusting the schedule, and within a quarter your best hourly is job hunting because they cannot count on the hours. The cap treats the symptom and creates a new problem.
Trying to fix it on the schedule alone
Rebuilding the schedule tighter looks like the answer. It is not. A tighter schedule with a narrow cover pool just means more callouts get answered with OT, because the cover pool cannot absorb them. Fix the cover pool first, then tighten the schedule.
The point
Overtime is a system output, not a decision. The general manager is not choosing to spend the money. They are being handed a shift-cover problem at 2pm on Friday and reaching for the only tool in the drawer. Widen the drawer.
Set the mid-week alert at 35 hours. Build the written cover roster and refresh it every 30 days. Cross-train every hourly to a second station on slow mornings. Put the OT line first on the Monday P&L review with root cause every time. Do those four things and OT stops being a running argument and starts being a number that decays quietly toward zero.
The savings are real. The employees are better off. The manager stops running the same crisis every Friday. That is the whole return.