Chapter 19: Labor Management: Scheduling, Labor Cost Control, Overtime, and the Balancing Act of Staffing

"The schedule is the only document in this building that spends money before anybody walks in." — constructed; the sentence a general manager eventually says to a new assistant manager

Overview

It is Thursday morning. You are writing next week's schedule, and you have about forty minutes before the delivery lands. There is a blank grid on the screen with seven columns and about twenty rows, and every cell you fill costs money that you will not get back.

Most managers write this document by feel. They copy last week, adjust for who asked off, glance at the reservation book, and post it. It takes twenty minutes and it is, in dollar terms, the largest single decision they will make that week. Bellwether's labor line is \$500,000 a year against \$1,550,000 of sales. Divide that by the fifty-two schedules that produce it and each one is worth roughly \$9,600 — more than the restaurant's monthly rent, committed in forty minutes by somebody who is also thinking about the walk-in temperature.

Here is what this chapter is actually about. Labor is the half of prime cost you can move this week. Food cost moves over a purchasing cycle; you change a spec, you renegotiate a price, you fix a portion, and eight weeks later the number is different. Labor changes on Tuesday at nine o'clock when you decide whether to cut a server. That responsiveness is the reason labor is the most controllable line in the business and the reason it is the most frequently mismanaged: a thing you can change constantly is a thing you can change badly, constantly.

We are going to do something in this chapter that Chapter 4 did not do. Chapter 4's planning model treated labor as an assumption — \$252,000 of fixed cost plus 16.0% of sales — which is a perfectly reasonable way to build a first projection and a completely useless way to run a restaurant. Nobody has ever scheduled 16% of sales. You schedule a hearth cook from three until eleven. So we are going to throw that assumption out and build Bellwether's labor line from the bottom: position by position, shift by shift, hour by hour, at real illustrative wage rates, and then see what it costs.

Fair warning: it does not come to \$500,000. What we do about that is the most useful part of the chapter, and it is a conversation every operator has, usually about eight months too late.

In this chapter, you will learn to:

• State exactly what belongs in labor cost — wages, employer payroll taxes, workers' compensation, benefits — and compute the percentage properly, against a denominator you have defined out loud.
• Separate the salaried floor from the variable schedule, and explain why the floor is what makes slow services dangerous.
• Forecast covers and sales by day and daypart, and say honestly how wrong the forecast usually is.
• Build a staffing guide in cover bands, compute sales per labor hour and covers per labor hour, and recognize that staffing is a staircase rather than a slope.
• Write a schedule to that guide, sequence the shifts, and set a written cut order before the first table sits.
• Price an overtime hour all-in, and build the Thursday-morning report that prevents it.
• Read a weekly labor report for variance and trend — and name the four things it will never show you.

Learning Paths

🏗️ Opening — this chapter produces a section of your plan that lenders and landlords read closely. Do §19.4 and §19.8 by hand; the staffing guide is the deliverable. §19.2's salaried floor is the number that determines whether your slow nights are survivable. 📋 Managing — this is your chapter. Weight §19.5, §19.6, and §19.7. If you run someone else's restaurant, the schedule and the cut are the two levers you actually hold, and the labor report is the document your performance is judged on. 🍸 Beverage — bar labor behaves differently: one bartender covers a wide volume range, so the bar's marginal labor cost is near zero until it isn't. See §19.4's step function and §19.5's barback discussion. 🚚 Small Format — a truck or pop-up has almost no variable labor; the crew is the crew. Your version of this chapter is §19.2 (the fixed floor is nearly everything) and §19.3 (forecasting, because you cannot flex your way out of a bad day).

19.1 Labor cost percentage: what's in it, and why "labor" means more than wages

Ask an operator what their labor cost is and most of them will answer with a number derived from payroll. Ask what is in that number and the answers scatter immediately. Some include the owner's salary; many do not. Some include payroll taxes; most forget workers' compensation entirely. A few count only hourly wages, which produces a beautiful number that describes nothing.

Labor cost percentage is total labor cost divided by total sales:

$$\text{Labor cost \%} = \frac{\text{Wages} + \text{Payroll taxes} + \text{Workers' comp} + \text{Benefits}}{\text{Total sales}}$$

Every one of those four components is real money that leaves the bank account, and three of them are invisible on a schedule.

The four components, in the order people forget them

Wages. Hourly wages plus salaries, including the owner's salary if the owner works in the business. This last point produces arguments. An owner who does not pay themselves and reports a lovely 26% labor cost has not built an efficient restaurant; they have built one that requires an unpaid manager. If you ever want to sell the business, or hire someone to do your job, the buyer will add that salary back and the number will move. Put it in from day one.

Employer payroll taxes. The employer's share of Social Security and Medicare — FICA, 7.65% of wages — plus federal unemployment (FUTA) and state unemployment (SUTA) insurance. FICA is a straight percentage. FUTA and SUTA are charged on a capped amount of each employee's wages, which makes them behave strangely: they are effectively a per-head cost, so a restaurant with high turnover pays them over and over on the same job.

Workers' compensation insurance. Priced per \$100 of payroll, by job classification. Restaurant classifications are not cheap — kitchens have knives, fryers, wet floors, and hot surfaces. This is a line most first-time operators discover in month two.

Benefits and the rest. Health contributions, paid time off, meal programs, uniform allowances, required certifications. In a small independent this is modest. It is not zero.

🧮 Run the Numbers

What Bellwether's payroll taxes actually cost. (illustrative rates; verify locally)

Bellwether's bottom-up wage total, built later in this chapter, is \$496,541 for the year.

Component	Basis	Cost
FICA (employer share)	7.65% × \$496,541 | **\$37,985**
FUTA + SUTA	capped per-employee charge, ~34 W-2s issued	\$7,900
Total employer payroll taxes	≈ 9.2% of wages	\$45,885

Now notice the second line. Bellwether has 24 positions, but at the industry's roughly 75% annual turnover (Chapter 17) it will issue something closer to 34 W-2s in year one. Unemployment insurance is charged on the first several thousand dollars of each person's wages. Ten extra people cycling through the same jobs does not add ten people's worth of wages — but it does add ten people's worth of unemployment tax.

That is a real, countable, several-hundred-dollar consequence of turnover that has nothing to do with training time. It is one small piece of the number Chapter 17 made you compute, and it shows up on a line most operators never look at.

The denominator is a decision, and you must say which one you used

Here is a trap that produces more confused conversations than any other number in this book.

Bellwether's cover model — 475 dinner covers and 220 brunch covers a week, at the plan's \$46 dinner and \$24 brunch checks — produces **\$27,130 a week, or \$1,410,760 over fifty-two weeks. The plan's year-one revenue is \$1,550,000**. The difference, about **\$139,000**, is patio season, private events, and holiday-period check lift: real revenue, but lumpy, and not produced by the base week's schedule.

So a single week's labor dollars can be divided by three different denominators and produce three different, defensible, contradictory-looking percentages:

Throughout this chapter, the staffing guide is written against the base operating week and every labor percentage is computed against annual dollars, \$1,550,000, unless labeled otherwise. State your denominator the same way, every week, in writing. Half the arguments about labor cost in restaurants are two people using different denominators and neither one saying so.

⚠️ Where the Money Leaks

The three exclusions that flatter a labor number.

1. Excluding the owner's salary. A chef-owner drawing nothing shows 26% labor. Pay them \$52,000 and the same restaurant shows 29.4%. The restaurant did not change. On \$1,550,000 that is 3.4 points of self-deception.
2. Excluding payroll taxes, workers' comp, and benefits. At Bellwether these total \$73,920 — payroll taxes \$45,885, workers' compensation \$12,035, benefits and certifications \$16,000 — which is 4.8 points of sales. An operator who quotes wages-only labor is understating by nearly five points and will not understand why the P&L disagrees with them.
3. Using gross sales instead of net sales. If your point-of-sale (POS) reports gross sales before comps and discounts, and your labor is real, your percentage is optimistic by whatever your comp rate is. Chapter 31 sets this straight; use net sales, consistently.

Together these three habits can make a 37% restaurant look like a 29% restaurant. The bank account is not fooled.

The benchmark, and its limits

Full-service restaurants commonly target labor somewhere in the 30–36% range, with prime cost — labor plus cost of goods sold — at or below 60%. Those are industry rules of thumb rather than laws, and they are ranges because a diner and a tasting-menu room do not share a labor model.

The rule of thumb has a specific failure mode worth naming now: labor percentage is a ratio, and ratios move for two reasons. A restaurant whose labor percentage improved from 34% to 31% may have gotten better at scheduling. It may also have simply had a busy month. The percentage cannot tell you which, and managers are rewarded for the number rather than the cause. That is why §19.8 makes you read hours and dollars alongside the percentage, always.

19.2 Fixed vs. variable labor and the salaried floor beneath your schedule

Every restaurant's labor divides into two behaviors, and the division matters more than almost any other structural fact about the business.

Fixed labor does not move with volume. The chef is in the building whether you do 60 covers or 140. The general manager opens the office on a dead Tuesday. Salaried positions are the obvious case, but part of your hourly payroll is functionally fixed too: somebody has to receive the delivery, somebody has to break down the walk-in, and the dish station cannot go below one person.

Variable labor moves with volume, in steps. A fourth server on Friday. A second dishwasher when the covers pass 110. A barback when the bar goes three deep.

🧮 Run the Numbers

Bellwether's salaried floor. (constructed teaching example; illustrative salaries)

Bellwether runs three salaried positions in year one:

Position	Salary	Notes
Chef-owner / executive chef	\$52,000	expedites the pass Wed–Sat
Front-of-house partner / general manager	\$50,000	runs the room five nights
Sous chef	\$48,000	expedites Tuesday, runs prep, closes
Salaried wages	\$150,000** | **9.7% of \$1,550,000

Add the costs that ride along with those three, plus the ones that ride along with everybody:

Line	Annual
Salaried wages	\$150,000
Payroll taxes on those salaries (their share of the \$45,885 bill) | \$13,860
Workers' compensation (whole restaurant)	\$12,035
Benefits, certifications, PTO accrual	\$16,000
TOTAL FIXED LABOR FLOOR	\$191,895

That is 12.4% of sales, committed before a single hourly hour is scheduled. On a Tuesday producing \$2,852 of sales, the floor's daily share alone is about **\$615** — 21.6% of that night's revenue, spent on three people who would be in the building regardless.

This is why slow services are dangerous and why the shape of your week matters more than your average. The floor does not care that it is February.

Notice what the floor is not. It is not \$252,000. Chapter 4's planning model assumed a fixed labor component of \$252,000 plus a variable component of 16.0% of sales — which, at \$1,550,000, sums to exactly \$500,000 and looks tidy on a pro forma. When we build the thing from actual positions, the fixed floor comes in at \$191,895** — about \$60,000 lower than assumed — and the variable component comes in far higher. Two errors, in opposite directions, that do not cancel. We will price that mismatch precisely in §19.4.

FIGURE 19.1 — Fixed vs. variable labor, Bellwether base week    [the Bellwether plan]

  Weekly sales by service, and the labor that sits underneath it

  SALES     Tue      Wed      Thu      Fri      Sat        Sun
            $2,852   $3,588   $4,232   $5,520   $8,298     $2,640
                                                (din+brunch)

  FIXED FLOOR (three salaried + taxes + comp + benefits) = $3,690/week
            ██████   ██████   ██████   ██████   ██████     ██████
            $615     $615     $615     $615     $615       $615      each operating day

  VARIABLE (hourly wages + taxes on them) = $7,280/week
            ███      ████     █████    ███████  ██████████ ███
            $732     $996     $1,137   $1,543   $2,207     $664

  ALL-IN LABOR AS % OF THAT DAY'S SALES
            47.2%    44.9%    41.4%    39.1%    34.0%      48.5%
            ▲                                   ▼          ▲
            worst                                best      worst

  The week averages 40.4% of base-week sales. Not one single day is at 40.4%.

Read that figure slowly, because it contains the chapter's first hard lesson.

Sunday brunch runs 48.5% labor. Tuesday runs 47.2%. Saturday runs 34.0%. The weekly average of 40.4% describes no actual day of the week. Friday and Saturday are carrying the building, and they are carrying it not because they are efficient but because they are large enough to dilute the fixed floor.

The instinctive management response to this figure is wrong, and it is wrong in an expensive way. The instinct is: Sunday brunch runs 48.5% labor, so kill Sunday brunch. Let us price that.

🧮 Run the Numbers

Should Bellwether drop Sunday brunch? (the Bellwether plan)

Sunday produces \$2,640** of sales a week — \$137,280 a year — on \$664** of all-in hourly labor a week, or \$34,537 a year. It is the worst labor percentage on the schedule.

Drop it and here is what happens:

	With Sunday	Without Sunday
Annual sales	\$1,550,000 | \$1,412,720
Annual labor (all-in)	\$570,461 | \$535,924
Labor %	36.8%	37.9%

The labor percentage gets worse by 1.1 points. You removed \$137,280 of revenue and only \$34,537 of cost, because the salaried floor, the workers' compensation premium, and the benefits do not go away on Sundays.

The correct measure is contribution, not percentage. Sunday brunch produces \$2,640 of sales, costs roughly \$734 in food and beverage at the plan's 27.8% blended cost of goods sold, and \$664 in all-in hourly labor. It contributes \$1,242 a week — \$64,584 a year — toward the fixed costs that exist anyway.

You would never drop it. A high labor percentage on a service that carries fixed cost is not a reason to close the service; it is a reason to grow it. Chapter 32 formalizes this as break-even by daypart.

👨‍🍳 On the Line

What the fixed floor feels like at 9:40 on a Tuesday.

Nine forty on a Tuesday in February. Fourteen covers left in the book, two tables still eating, nothing on the board. The hearth is banked. The sauté cook has broken down her station twice because there was nothing else to do.

In the building right now: the chef-owner, the sous, the FOH partner, one hearth cook, one sauté cook, one dishwasher, two servers, one bartender. Nine people. The last hour will produce maybe \$180 of sales and will cost about \$155 in wages, plus the fixed floor grinding away underneath it.

The move is not to send everyone home at 9:40 — you are open until ten, and the two tables still eating are the reason the business exists. The move happened eleven days ago, when somebody wrote a schedule that put two servers, three cooks, and a full FOH management presence on a night the forecast said would do 62 covers.

This is the single most important thing to understand about labor management: the decision that matters is almost never the one in front of you at 9:40. By 9:40 your options are bad, small, and personally unpleasant. The decision that mattered was made on a Thursday morning with a blank grid and a forecast, and it is the only decision in this chapter that is genuinely under your control.

19.3 Forecasting: sales by day, by daypart, by weather, by history

You cannot write a schedule against a number you do not have. Every staffing decision in this chapter depends on a forecast, and the quality of the forecast sets a ceiling on the quality of everything built on top of it.

Labor forecasting is the practice of projecting covers and sales for each future service period in enough detail to staff it. Not "next week will be busy." Thursday will do 92 covers and \$4,232, with 61 of those covers between 6:00 and 8:00.

What a forecast is made of

History first. Same day, last week. Same day, four weeks ago. Same day, last year, if you have it. Restaurant demand is powerfully weekly — a Tuesday looks like other Tuesdays far more than it looks like the Monday before it — and the single best predictor of next Thursday is the last four Thursdays.

Then the book. Reservations on the books at the time you write, adjusted by your historical conversion: at Bellwether, a Friday with 84 covers reserved on the Thursday before has historically finished around 120, so the reservation book at that point in the cycle carries about 70% of the eventual night. That ratio is yours to measure, and it is different for every restaurant and every night of the week.

Then the calendar. Paydays. The first Friday of the month. School calendars. Local sports. A holiday. The Rivermill District's summer street market on the third Saturday. A concert two blocks away is worth 20 covers of early business and a dead 8:30.

Then the weather. Rain suppresses walk-in traffic and helps reservations hold. Cold suppresses everything. The first genuinely warm evening in April empties every dining room in a Midwestern city because everyone is on a patio somewhere, and the first genuinely warm evening in April fills Bellwether's patio, which is why the patio is worth having.

Then judgment, applied last and sparingly. A manager's gut is a legitimate input. It is not a legitimate substitute.

🧾 Read the Numbers

```text FIGURE 19.2 — "The forecast worksheet, week 31" [the Bellwether plan] THE ARTIFACT The one-page forecast a manager fills in on Thursday morning before writing the following week's schedule. Covers and sales by service, with the four inputs that produced each number. THE CONTEXT Bellwether, late July, week 31 of year one. Patio open. Two of the four original line cooks have left; one replacement is in week two of training.

               SERVICE   4-WK AVG   ON BOOK   CALENDAR/WX      FORECAST   ACTUAL
               Tue         64         18      —                    64        58
               Wed         79         24      —                    80        77
               Thu         94         41      street market Sat    96       101
               Fri        124         86      warm, clear         128       126
               Sat        126         92      street market       134       139
               Sat brunch 112         14      street market       120       118
               Sun brunch 110         11      rain forecast       104        82
               ─────────────────────────────────────────────────────────────────
               WEEK       709        286                          726       701
               SALES                                          $30,433   $29,940

WHAT IT SHOWS A competent forecast. Five of seven services landed within 5%. The week came in 3.4% under on covers and 1.6% under on sales — meaning the covers that went missing were the cheap ones. Total error is small enough to schedule against; the manager staffed 461 hours and the week needed roughly 458. WHAT IT DOESN'T It does not show that the entire week's miss is ONE service. Sunday brunch forecast 104 and did 82 — a 21% miss on the day the forecaster hedged for rain and then hedged too little. Every other service was within 5%. An aggregate error of 3.4% is hiding a service-level error of 21%, and the schedule was written service by service, not in aggregate. It also does not show WHO the covers were. Thursday beat forecast by five covers on the strength of one six-top that ordered two bottles of wine — revenue with almost no incremental labor in it. THE DECISION Stop forecasting Sunday brunch off the four-week average. Sunday is the most weather-elastic service Bellwether runs, and the four-week average washes that out. Build a Sunday-specific rule: forecast the four-week average, then apply a documented weather adjustment, and schedule the last server and the second brunch cook as a 10:00 a.m. call rather than a 9:00 a.m. shift. THE LESSON Forecast accuracy at the week level is a vanity metric. You do not schedule a week; you schedule fourteen separate services. Measure your error where you make the decision. ```

How wrong a forecast is, honestly

A well-run independent restaurant with a year of history and a reservation book forecasts individual services within about 5–10% most of the time, with a long tail of much larger misses driven by weather, events, and the occasional inexplicable Tuesday. A brand-new restaurant has no history at all and forecasts by guessing for roughly the first quarter — which is one of several reasons the opening period runs an ugly labor number no matter how good the manager is.

This has a direct scheduling consequence. If your forecast is ±8%, do not build a schedule that only works if the forecast is exact. Build one with a designed release valve: shifts that can be cut cheaply, a late server who starts at 6:00 rather than 4:30, a second dishwasher scheduled as an on-call. §19.5 turns this into a technique.

🔍 Check Your Understanding

1. A manager forecasts a week at 726 covers and the week does 701. Is that a good forecast? What would you need to see before answering?
2. Why is last Thursday a better predictor of next Thursday than yesterday is?
3. Bellwether's reservation book carries about 70% of an eventual Friday when the schedule is written. Fifty-six covers are on the books. What is your Friday forecast, and what would make you override it?

(1: Not yet answerable. A 3.4% aggregate error is good; but if it is composed of one service missing by 21% and six landing within 5%, you have a Sunday problem, not a forecasting problem. 2: Restaurant demand is weekly-periodic — day-of-week is a stronger signal than recency, because the guest population, occasion mix, and competitive set differ more between weekdays than between the same weekday in consecutive weeks. 3: 56 ÷ 0.70 = 80 covers, well below the 120 baseline — which should immediately make you check whether the book is genuinely soft or whether a large party cancelled, and whether a private event, holiday, or weather event is in play.)

19.4 The staffing guide: covers per labor hour and sales per labor hour

This is the section that produces a document. Everything before it was preparation.

A staffing guide — also called a labor matrix — states, for each service and each band of forecast volume, exactly which positions are scheduled and for how many hours. It converts a forecast into a schedule mechanically, so that the same Tuesday produces the same staffing every time, and so that a manager who is tired, new, or in a hurry cannot quietly overstaff a Wednesday because they like the crew.

The two metrics that build it:

$$\text{Sales per labor hour (SPLH)} = \frac{\text{Net sales}}{\text{Labor hours}} \qquad \text{Covers per labor hour (CPLH)} = \frac{\text{Covers}}{\text{Labor hours}}$$

SPLH is the productivity number you manage to; it moves with your check average, so it is comparable across your own history but not across restaurants with different price points. CPLH is the production number; it is comparable across similar concepts and it is what tells you whether the kitchen is actually capable of the night you booked. Use both. An SPLH that improves while CPLH falls means you sold more per guest, not that you got more efficient.

Building it position by position

Bellwether's peak kitchen line is four: hearth, sauté/pasta, garde manger, and the chef expediting at the pass. The hearth caps the kitchen at roughly 144 covers a night. Those two facts, established in Chapter 7 and lived in Chapter 14, define the top of the guide — there is no band above 144, because there is no kitchen above 144.

Working down from there, here is the back-of-house build for a base week. These are station hours, not people; §19.5 assigns them to bodies.

FIGURE 19.3 — Bellwether BOH station hours, base operating week      [the Bellwether plan]

  STATION            TUE    WED    THU    FRI    SAT   |  WEEK
  Covers              62     78     92    120    123   |   475
  ─────────────────────────────────────────────────────┼────────
  Hearth             7.0    7.5    8.0    8.5    8.5   |  39.5
  Sauté / pasta      7.0    7.5    8.0    8.5    8.5   |  39.5
  Garde manger        —     5.0    5.5    7.0    7.0   |  24.5
  AM prep            6.0    6.0    6.5    7.0    7.0   |  32.5
  Dish / porter      6.0    6.5    7.0   12.0   12.0   |  43.5
  ─────────────────────────────────────────────────────┼────────
  BOH HOURLY HOURS  26.0   32.5   35.0   43.0   43.0   | 179.5
  (chef or sous expedites — salaried, not counted here)

  BRUNCH             SAT    SUN   |  WEEK
  Covers             110    110   |   220
  ────────────────────────────────┼────────
  Brunch line (lead) 6.5    6.5   |  13.0
  Brunch line (2nd)  5.5    5.5   |  11.0
  Brunch prep/dish   5.0    5.0   |  10.0
  ────────────────────────────────┼────────
  BOH BRUNCH HOURS  17.0   17.0   |  34.0

  TOTAL BOH HOURLY HOURS PER WEEK = 179.5 + 34.0 = 213.5

On Tuesday there is no garde manger. Sixty-two covers do not need four stations, so the pantry work folds onto the sauté cook's station and the sous chef expedites rather than the chef-owner, who takes the night. This is not a corner being cut; it is what a 62-cover night is.

The front of house, built the same way:

FIGURE 19.4 — Bellwether FOH station hours, base operating week       [the Bellwether plan]

  POSITION           TUE    WED    THU    FRI    SAT   |  WEEK
  Covers              62     78     92    120    123   |   475
  ─────────────────────────────────────────────────────┼────────
  Servers           11.0   17.0   18.5   26.0   26.0   |  98.5
    (bodies)          2      3      3      4      4    |
  Bartender          6.5    6.5    7.0    7.5    7.5   |  35.0
  Barback             —      —      —     4.0    4.0   |   8.0
  Host                —     4.0    4.5    5.5    5.5   |  19.5
  Busser / runner     —      —     4.0   10.5   10.5   |  25.0
  ─────────────────────────────────────────────────────┼────────
  FOH HOURLY HOURS  17.5   27.5   34.0   53.5   53.5   | 186.0

  BRUNCH             SAT    SUN   |  WEEK
  ────────────────────────────────┼────────
  Servers (3)       15.0   15.0   |  30.0
  Bartender          4.5    4.5   |   9.0
  Host               3.5    3.5   |   7.0
  Busser / runner    4.0    4.0   |   8.0
  ────────────────────────────────┼────────
  FOH BRUNCH HOURS  27.0   27.0   |  54.0

  TOTAL FOH HOURLY HOURS PER WEEK = 186.0 + 54.0 = 240.0

  TOTAL HOURLY HOURS, ALL SERVICES = 213.5 + 240.0 = 453.5

On Tuesday the FOH partner works the host stand personally. That is a defensible use of a salaried manager on a 62-cover night and an indefensible one on a 120-cover Friday, when the manager needs to be in the room rather than behind a podium.

What it costs

Now attach illustrative wage rates. These are constructed teaching figures for a mid-size Midwestern metro. Minimum wages and tipped-wage rules vary enormously by state, county, and city; several states have no tip credit at all, and the cash wage a tipped employee must receive is a question of local law that Chapter 20 takes up in detail. Verify locally before you build a real model on any of these numbers.

Total hourly: 453.5 hours, \$6,664.25 a week. Twenty-one hourly positions plus three salaried — 24 people on the roster.

The server cash wage of \$9.50 assumes a jurisdiction that permits a tip credit against the full minimum wage. It is a cash wage, not a compensation figure: Bellwether's servers are earning meaningfully more than \$9.50 an hour once tips are counted, and on a Friday considerably more than the hearth cook. That gap — a tipped front of house and an untipped kitchen inside the same four-point business — is one of the genuinely hard problems in American restaurants, and it drives tip-pooling and service-charge decisions that Chapter 20 owns and Chapter 21 has to live with.

The annual build, against the plan

(Percentage column rounds to the nearest tenth; components may not re-sum exactly.)

There it is. The schedule this restaurant actually needs costs \$70,461 more than the plan says labor will cost — four and a half points of sales.

Do not adjust the plan's \$500,000 to make this go away. That number is load-bearing: it is what the 60% prime cost target and every projection built on it depend on. What you do instead is state the gap out loud, understand exactly where it came from, and decide — deliberately, in writing — what gives. §19.4's remaining pages do the first two. The chapter's Business Plan checkpoint does the third.

Where the gap came from

Chapter 4 modeled labor as **\$252,000 fixed + 16.0% of sales variable**. At \$1,550,000 that is \$252,000 + \$248,000 = \$500,000, exactly. It is an elegant model and both of its terms are wrong.

The fixed assumption was too high because a 68-seat restaurant genuinely runs on three salaried people, not the four or five a planner instinctively pencils in. The variable assumption was far too low because 16% of sales, at Bellwether's blended hourly rate of \$14.70, funds about 369 hourly hours a week — and the four-station line, a dish pit that handles 123 covers, two brunches, and a room with bussers needs 453.5.

The plan is short 84 hours a week. That is not a rounding error. That is two full-time positions and a Saturday busser.

Sit with that for a moment, because it generalizes. "Labor will run 32%" is not a model. It is a wish with a decimal point. It works backward from the answer you want, and it stays invisible until the first month's payroll register lands and the percentage is 38.

The tell is simple, and you can apply it to anyone's plan including your own. Ask: how many hours a week does that fund? If the answer does not arrive in under ten seconds, the number is not connected to a schedule and it will not survive contact with a Friday.

Build the schedule first. Let the percentage be the output. If the output is unacceptable, you now have a real conversation about hours, stations, service standards, and price — instead of an argument about a number nobody can trace.

The step function nobody warns you about

Now for the most useful thing in this section, and the reason a staffing guide is written in bands rather than as a formula.

FIGURE 19.5 — Bellwether's dinner labor staircase                    [the Bellwether plan]

  HOURS
  110 ┤                                                    ┌───── 106.0
      │                                                    │   131–144: prep stays late,
  100 ┤                                    ┌─────────────── ┘   2nd dish extended, mid server
      │                                    │  96.5
   90 ┤                                    │   111–130: 4th server, busser, barback,
      │                                    │   2nd dishwasher, sous expedites
   80 ┤                                    │
      │                                    │
   70 ┤                    ┌───────────────┘
      │                    │  69.0    86–110: 3rd server, runner, prep flexes to 6:30
   60 ┤          ┌─────────┘
      │          │  60.0   66–85: garde manger on, host on
   50 ┤          │
      │          │
   40 ┤──────────┘  43.5   ≤65: three stations, two servers, no host, no busser
      │
      └──┬────┬────┬────┬────┬────┬────┬────┬────┬────┬────┬──
        50   60   70   80   90  100  110  120  130  140  150  COVERS
                                                          ▲
                                                    hearth ceiling: 144

  MARGINAL HOURS PER ADDITIONAL COVER, BY STEP
    65 → 85 covers   (+20)   +16.5 hrs   =  0.83 hrs/cover   expensive
    85 → 110 covers  (+25)    +9.0 hrs   =  0.36 hrs/cover   cheap
   110 → 130 covers  (+20)   +27.5 hrs   =  1.38 hrs/cover   most expensive step
   130 → 144 covers  (+14)    +9.5 hrs   =  0.68 hrs/cover   moderate

Labor is not a slope. It is a staircase, and the risers are not evenly spaced.

Between 85 and 110 covers, each additional cover costs about 0.36 labor hours — roughly \$5.30 of wages against \$46 of sales. Those are the best covers in the building. You have already paid for the step; you are simply filling it.

Between 110 and 130 covers, each additional cover costs about 1.38 labor hours — about \$20 of wages against \$46 of sales, because crossing 110 means adding a fourth server, a busser, a barback, and a second dishwasher all at once. That step is why a 112-cover Friday can be less profitable than a 108-cover Friday, and it is a genuinely counterintuitive fact that most managers discover by accident.

🧮 Run the Numbers

The most expensive covers Bellwether sells.

Compare two Fridays, both real possibilities against a 120-cover baseline:

	Friday A	Friday B
Covers	108	116
Sales at \$46 | \$4,968	\$5,336
Staffing band	86–110	111–130
Hourly hours scheduled	69.0	96.5
Hourly wages (at \$14.70 blended) | \$1,014	\$1,418
All-in hourly labor (×1.092)	\$1,107 | \$1,549
Labor as % of that night's sales	22.3%	29.0%

Friday B sells \$368 more** and spends **\$442 more on labor. Eight additional covers made the night worse.

This does not mean turn away the eight covers. It means three specific things:

1. Fill the step you are standing on. The covers between 116 and 130 are nearly free — the team is already scheduled. Chapter 24's revenue-management tools exist precisely to move a 116-cover Friday to a 128-cover Friday, and that move is worth far more than trimming hours.
2. Know where your risers are, in writing. The manager writing Friday's schedule must know that 111 is a threshold. If the forecast is 112, that is the moment to ask whether the reservation book can be shaped to 108 or pushed to 125 — because 112 is the worst number on the board.
3. Design a soft riser where you can. A 5:30 p.m. busser who can be released at 8:00, or a barback scheduled as a call-in, converts a hard step into a gentle one. §19.5 does this.

SPLH and CPLH, reconciled

Now check the build against the frozen anchors. Bellwether's base week: 475 dinner covers, 220 brunch covers, 695 total, and 36,140 covers a year. Annual sales \$1,550,000.

(695 covers × 52 = 36,140 ✓. 453.5 hours × 52 = 23,582 ✓. 36,140 ÷ 23,582 = 1.53 ✓.)

And by service:

Two things in that table deserve attention.

Brunch's CPLH is nearly double dinner's — 2.50 against 1.30 — because brunch turns fast, runs a two-station kitchen, and serves a menu built for throughput. On production efficiency, brunch is the best service Bellwether runs.

And brunch's SPLH is identical to dinner's, at \$60. Twice the covers per hour at half the check lands in exactly the same place. That is a coincidence of arithmetic, not a law, but it makes the point that SPLH and CPLH answer different questions and you need both. If you managed Bellwether on SPLH alone you would conclude the two services are equivalent. They are not: brunch is a production win and a revenue problem, and the fix is a higher brunch check (Chapter 24), not more brunch hours.

Always state which hours are in the denominator. Bellwether's SPLH is \$65.73 on hourly hours and \$50.65 on all hours including the salaried three. Both are true. Industry rules of thumb for full-service SPLH generally sit in the \$50–\$70 range, and which end of that range you are comparing against depends entirely on a convention nobody states.

🔍 Check Your Understanding

1. Bellwether's forecast for a Thursday is 108 covers. The staffing guide's band is 86–110. Two more reservations come in. What should the manager do, and why is this a harder question than it looks?
2. A restaurant's SPLH improves from \$58 to \$63 while CPLH falls from 1.6 to 1.4. What happened?
3. Why is the fixed floor excluded from Bellwether's headline SPLH of \$65.73, and what does including it change?

(1: Nothing to the schedule — 110 is still inside the band, and those two covers are close to free. The hard part is what to do at 111: crossing the riser costs 27.5 hours, so the manager should either hold the book at 110 or work to fill toward 125. 2: The check average rose and volume fell — more sales per guest, fewer guests. Productivity did not improve; pricing or mix did. 3: Because SPLH is a scheduling tool and salaried hours are not schedulable. Including them drops SPLH to \$50.65 and makes the number comparable across restaurants with different management structures, which is the right convention for benchmarking and the wrong one for writing Tuesday.)

19.5 Writing the schedule: sequencing, shift lengths, overlaps, and the cut order

The staffing guide says how many hours. The schedule says which human being, starting when.

The sequence that produces a schedule in forty minutes

1. Post the forecast on the schedule itself. Every schedule Bellwether posts carries the forecast covers for each service in the header. The staff should see the number they are staffed to. It turns "why are we so slammed" into "we booked 138 and staffed for 134," which is a conversation rather than a grievance.
2. Place the fixed positions first. Chef, sous, FOH partner, the AM prep cook, the opening dishwasher. These do not flex; putting them down first stops you from "finding" hours later.
3. Place the anchor shifts. The hearth cook and sauté cook Friday and Saturday. Your two strongest servers on the two biggest nights. Build the weekend, then work backward into the week — because the weekend is where a mistake is unrecoverable.
4. Fill to the guide, band by band. Read the forecast, read the band, place the positions.
5. Stagger the starts. Not everyone at 4:30.
6. Write the cut order. Before you post.
7. Check hours by person against 40. This is the step people skip, and §19.7 is what it costs.

Shift lengths, overlaps, and the split shift

A split shift is a single employee working two separated blocks in one day — a prep shift from nine to two and a service shift from five to eleven, with the middle unpaid. It is a real tool and it is one you should use sparingly and honestly. It solves a genuine problem: prep demand and service demand peak four hours apart. It creates a genuine one: an employee whose workday is fourteen hours long and whose paid day is eleven, who cannot go home, and who did not choose it.

Bellwether does not schedule split shifts. Its AM prep cook works nine to four, and its line cooks come in at two-thirty or three and work through. That is a deliberate cost — a straight-through schedule costs more hours than a split one — and it is a retention decision, not a generosity one. Chapter 21 makes the case with the turnover math from Chapter 17 behind it.

Overlaps are where a schedule's quality actually lives. Consider Bellwether's Friday:

FIGURE 19.6 — Bellwether Friday dinner, 120-cover forecast            [the Bellwether plan]

           2p   3p   4p   5p   6p   7p   8p   9p  10p  11p  12a
           |    |    |    |    |    |    |    |    |    |    |
  BOH
  AM prep  ████████████████████                              9:00a–5:00p (7.0 of it here)
  Hearth        ▓████████████████████████████████████        2:30p–11:00p   8.5
  Sauté         ▓████████████████████████████████████        2:30p–11:00p   8.5
  Garde              ▓█████████████████████████████          3:30p–10:30p   7.0
  Dish A             ▓█████████████████████████████████      3:30p–12:00a   8.0*
  Dish B                       ▓████████████████             6:00p–10:00p   4.0
  Sous     ▓█████████████████████████████████████████        1:00p–11:00p   (salaried)
  Chef          ▓██████████████████████████████████████      2:30p–11:30p   (salaried)

  FOH
  Server 1            ▓████████████████████████████          4:00p–11:00p   7.0
  Server 2            ▓████████████████████████████          4:00p–11:00p   7.0
  Server 3                 ▓██████████████████████           4:30p–11:00p   6.5
  Server 4                      ▓██████████████████          5:30p–11:00p   5.5   ← first cut
  Bartender           ▓█████████████████████████████         4:00p–11:30p   7.5
  Barback                       ▓█████████████               6:00p–10:00p   4.0   ← second cut
  Host                     ▓████████████████████             4:30p–10:00p   5.5
  Runner                   ▓█████████████████████            4:30p–11:00p   6.5
  Busser                        ▓████████████████            5:30p–10:30p   5.0   ← third cut

  ▓ = pre-service overlap    █ = on the floor
  * Dish A's Friday station total is 12.0 hours, covered by A (8.0) and B (4.0).

  DEMAND CURVE (covers seated per half hour)
   5:00 ▂▂  5:30 ▃▃▃  6:00 ▅▅▅▅▅  6:30 ███████  7:00 ████████  7:30 ██████
   8:00 ▅▅▅▅▅  8:30 ▃▃▃  9:00 ▂▂  9:30 ▁

Look at what the staggered starts are doing. Servers 1 and 2 are on at 4:00 to set the room and take the early business. Server 4 arrives at 5:30, forty minutes before the push starts, and is the first person cut. The busser arrives at 5:30 and leaves at 10:30, sitting exactly over the peak. The barback covers 6:00 to 10:00 — the four hours the bar cannot keep up alone.

Nobody in this schedule is scheduled for the whole night just because the night is busy. Each position is scheduled for the window in which it is worth its wage, and the shape of the schedule matches the shape of the demand curve underneath it.

That is what a good schedule looks like: not a flat block of people from four to eleven, but a staircase that goes up before the guests arrive and comes down after they leave.

👨‍🍳 On the Line

Pre-service overlap is the cheapest thirty minutes you will ever buy.

Every shift on that grid starts with a shaded block — thirty minutes to an hour before service. A cook uses it to finish mise en place, taste, check pars, and set the station. A server uses it to fold napkins, run through 86s, taste the special, and set their section.

Managers cut this first, because it is invisible on a labor report — nobody sees a table waiting at 4:30. It is also the single most expensive cut in the building.

A cook who starts at 3:00 instead of 2:30 on a 120-cover night is not thirty minutes behind. They are thirty minutes behind at the moment the first ticket lands, and they do not catch up, because a station that starts behind stays behind for the entire service. Ticket times run four minutes long from six-thirty onward. Tables that would have turned at 8:20 turn at 8:40. You lose the second seating on two four-tops — \$368 of sales — to save \$10.50 of a cook's time.

Cut the end of a shift, never the beginning. The end of a shift is a cook cleaning a station that will still be there tomorrow. The beginning of a shift is the whole night.

The cut order, written before service

Cutting the floor is releasing staff early as volume declines. It is the primary real-time labor lever and it is where a schedule either holds or falls apart.

A cut order is a written sequence, posted before service, stating who goes first, who goes second, who goes third, and — critically — the condition that triggers each cut.

FRIDAY CUT ORDER — posted at 4:00 p.m.                            [the Bellwether plan]

  #  POSITION      CUT WHEN                                       SAVES
  1  Server 4      no tables on a wait AND ≤ 6 tables seated       ~1.5 hr
                   in Server 4's section, not before 9:00 p.m.
  2  Barback       bar under 8 seated AND back bar restocked       ~1.0 hr
  3  Busser        last seating done AND dining room reset ≥ 60%   ~1.0 hr
  4  Garde manger  no apps on the board AND pantry restocked       ~0.5 hr
  5  Dish B        pot sink clear AND ≤ 3 tables remaining         ~0.5 hr

  DO NOT CUT: the hearth cook (closes the fire), the bartender (closes the drawer),
  Dish A (closes the pit). These three own a closing task nobody else can do.

  MAXIMUM AVAILABLE SAVING: 4.5 hours ≈ $60 in wages ≈ $66 all-in.

Two things about that document.

First, it is small. The entire cut order on a Friday is worth about \$66. The reason to write it is not the \$66; it is that \$66 a service, across three hundred and twelve services a year, is roughly \$20,600 — and it is the difference between a manager who makes the same decision every time and one who makes it based on how tired they are.

Second, the conditions are objective. "Cut when it slows down" is not a cut order; it is a mood. "No tables on a wait and six or fewer seated in that section, not before nine" is a rule a twenty-three-year-old shift lead can execute at 9:15 without calling you.

🤝 Hospitality

The cut is a service decision before it is a cost decision.

There is a version of labor management that treats the cut as pure savings — get them out, protect the number. It is expensive and it is the mark of a manager who has never worked a floor that got cut too early.

When you cut Server 4 at 9:00 and a walk-in six-top arrives at 9:20, three things happen. The remaining servers' sections grow by 40%. Their existing tables get less attention at exactly the moment those tables are deciding whether to order dessert and a second round — the highest-margin part of the check. And the six-top gets a compressed, harried version of your restaurant.

The saving was \$14.25. The dessert-and-digestif round you did not sell was \$60 at a very high contribution margin. The six-top's opinion of Bellwether was formed permanently in about eleven minutes.

This is why the cut order has conditions rather than times, and why "no tables on a wait" is the first clause of every one of them. You are not cutting to a number. You are cutting to a floor that can still deliver the thing you sell. Chapter 23 prices the second visit properly; for now, hold this: a cut that costs you one table's return has cost more than the cut saved.

⚖️ Code and Compliance

Scheduling rules that exist before your judgment does. (preview — Chapter 20 owns this)

The schedule is a legal document as well as a financial one, and several rules constrain it before any management judgment applies:

• Overtime. Federal law under the Fair Labor Standards Act (FLSA) generally requires time-and-a-half above 40 hours in a workweek for non-exempt employees, and some states add daily overtime thresholds.
• Predictive scheduling. A growing number of cities and a few states require advance posting of schedules (often 7–14 days), with premium pay owed for late changes — including the kind of call-in and early cut this section just described.
• Reporting-time pay. Some jurisdictions require minimum pay when an employee reports for a shift and is sent home.
• Meal and rest breaks. Requirements vary enormously by state, and a schedule that leaves no room for a required break is not a schedule.
• Minor labor rules. If you employ anyone under 18, hour limits and prohibited-equipment rules apply and are strictly enforced.

All of this varies by state, county, and city. Verify locally before you post a schedule, and use an employment attorney for anything consequential. Chapter 20 treats each of these properly.

19.6 Managing to it in real time: cutting the floor, calling in, and the sunk-cost trap

A schedule is a plan. Service is what happens to it.

The second Friday in October

Chapter 14 put you in the middle of it. Here it is again, with the money attached.

142 covers on the books, including a 40-top private party at 6:30. At 3:40 p.m. the grill cook does not arrive and does not answer. The sous chef, who came in at eight that morning to prep the party, covers the hearth. The night finishes at 138 covers and \$6,828 in sales. The sous leaves at eleven, having worked fifteen hours.

Chapter 14 walked you through the decisions. Now read the receipt.

🧾 Read the Numbers

```text FIGURE 19.7 — "The Friday the grill cook didn't come" [the Bellwether plan] THE ARTIFACT The posted schedule for the second Friday in October, the punch record that actually happened, and the hard costs the night generated. THE CONTEXT 142 covers booked, including a 40-top at 6:30 on a set menu. Forecast written eight days earlier. Hearth cook no-shows at 3:40 p.m.

               AS POSTED (written to a 120-cover band, 30% labor target)
                 Hearth cook      3:00p – 11:00p    8.0 hrs @ $21.00   $168.00
                 Sauté cook       2:30p – 11:30p    9.0 hrs @ $20.00   $180.00
                 Garde manger     3:00p – 10:30p    7.5 hrs @ $18.00   $135.00
                 AM prep          9:00a –  5:00p    8.0 hrs @ $17.50   $140.00
                 Dish A           3:30p – 12:00a    8.5 hrs @ $16.00   $136.00
                 Dish B           6:00p – 11:00p    5.0 hrs @ $16.00    $80.00
                 Sous (hourly)    8:00a –  8:00p   12.0 hrs @ $23.00   $276.00
                 FOH (4 svr, bar, barback, host, runner, busser)      $312.75
                 ────────────────────────────────────────────────────────────
                 SCHEDULED WAGES                                     $1,427.75

               WHAT ACTUALLY HAPPENED
                 Hearth cook      did not report                     -$168.00
                 Sous             8:00a – 11:00p (15.0 hrs; 3 at OT)  +$103.50
                 Sat. call-in     prep cook in 3 hrs early             +$52.50
                 ────────────────────────────────────────────────────────────
                 NET WAGE EFFECT                                       -$12.00

               THE HARD COSTS OF THE NIGHT
                 Wages, net (saved shift less OT and call-in)          -$12.00
                 Comped food at plate cost (2 apps, 1 entrée, 2 des.)  +$33.40
                 Comped beverage at pour cost (4 glasses, 2 cocktails) +$22.80
                 Product lost to re-fires and the 86'd pasta           +$23.90
                 Dish-shift overtime premium (1.5 hrs)                 +$12.00
                 Saturday expedite delivery on the unplaced order      +$35.00
                 Card-processing and tip-out on comped checks           +$1.60
                 ────────────────────────────────────────────────────────────
                 NET HARD COST                                        $116.70
                 As a percentage of the night's $6,828 in sales           1.7%

WHAT IT SHOWS A catastrophe that cost 1.7% of one night's revenue. The restaurant came out AHEAD on wages: the absent cook's unworked $168 shift exceeded the sous's $103.50 of overtime plus the $52.50 Saturday call-in by twelve dollars. If a labor report is all you read, this was a good night. The week's labor percentage improved. WHAT IT DOESN'T Everything that mattered. · The sous worked FIFTEEN HOURS. That appears as three hours of overtime, $103.50. It does not appear as a person who will be worse at their job Saturday, Tuesday, and Wednesday, or as one more entry in a ledger that eventually ends with a resignation. · ELEVEN TABLES went past the service standard. That appears as $33.40 of comps. It does not appear as the second visits that will not happen — the covers that are simply never booked. · The sauté cook worked ALONE ON TWO STATIONS FOR FOUR HOURS. That appears NOWHERE. On the labor report it is a $168 saving. · The 40-top's contract funded no additional labor. · Nobody placed Saturday's protein order, because the person who places it was on the hearth. That is the $35 expedite, and it is luck that it was only $35. THE DECISION Three things, in order. (1) Rewrite the forecast rule: a night with 142 on the books is a 131–144 band night and gets 106 hours, not the 96.5 the 120-cover band funds. (2) Put a written weekend on-call in place with a call-in bonus, and cross-train the garde manger and sauté cook onto the hearth so the on-call is a backfill rather than a rescue. (3) Any party over 20 covers gets its own labor line in the event contract (Chapter 29). THE LESSON A labor report measures dollars paid, not work done or damage sustained. The night this restaurant "came out ahead" was the worst service it ran all year. If your only instrument for labor is a percentage, you will reward the shifts that hurt you most. ```

The forecast was the failure, not the no-show

Read the schedule header again: it was written to a 120-cover band on a night with 142 on the books. The staffing guide's 131–144 band calls for 106.0 hours. The posted schedule carried 96.5.

The night was 9.5 hours short before anybody failed to show up.

Price the missing hours. In the 131–144 band, the additions over the 111–130 band are the AM prep cook staying two hours later, the second dishwasher extended two hours, and a mid-shift server — 9.5 hours in total:

All-in, with payroll taxes and workers' comp, that is about \$130.

\$130 of scheduled hours, committed eight days earlier, against \$116.70 of hard cost, eleven unhappy tables, a fifteen-hour day, and a Saturday morning spent chasing a protein delivery.

Be honest about what that comparison does and does not prove. On hard cost alone it is roughly a wash — the insurance cost about what the accident cost. That is the entire argument for buying it: the accident also cost eleven tables and a sous chef, and the insurance does not. The \$130 is priced in dollars. The rest of the damage is not priced in anything, which is exactly why operators underbuy it.

Note also what the extra prep hour would have done for the no-show itself. With the prep cook on the line until seven, the sous covers the hearth for the first push with support rather than alone, and the sauté cook is never abandoned on two stations. The schedule that was correct for 142 covers was also the schedule that would have absorbed the no-show. Slack is not waste. Slack is the thing that converts a disaster into an inconvenience.

The schedule that would have made that Friday survivable

Four changes, priced.

Against \$1,550,000 of sales, that package is 0.24% of revenue — a quarter of a point of labor — and it removes most of the failure modes that produced the October Friday. It is also, and this matters, the single easiest thing in this chapter to cut when the labor number runs hot in February, because it is a cost with no visible output. Protecting it is a management decision that has to be made once and defended repeatedly.

⚠️ Where the Money Leaks

The sunk-cost trap: "they're already here."

It is 8:45. The room is emptying. You have a server, a busser, and a garde manger who have each worked four of a six-hour shift.

The trap is the sentence "they're already here, might as well keep them." It is a sunk-cost error in its purest form. The four hours are spent and unrecoverable. The only live question is whether the next hour of each person's time produces more than it costs.

The busser at \$13.50, all-in about \$14.75: at 8:45 with the room 30% seated and resetting itself, almost certainly not. The garde manger at \$18.00, all-in \$19.66, with no appetizers on the board: almost certainly not. The server at \$9.50 plus tips, with four tables who are about to order dessert: absolutely yes, and cutting them is the error in the other direction.

The trap runs both ways, and the same manager usually falls into both in the same night. The countermeasure is the written cut order, which forces the question what does the next hour produce for each position separately, before the moment arrives and everyone is tired.

Calling in, and when it is worth it

The reverse decision — a night running hotter than forecast, and the question of whether to call someone in.

The arithmetic is straightforward and almost always favors the call. Bringing a server in for four hours costs about \$38 of wages, \$42 all-in. Four hours of an under-staffed floor at 130 covers costs you ticket times, table turns you do not get, comps, and second visits. A single four-top that turns at 8:15 instead of 8:45 is a \$184 second seating you would otherwise have lost.

The reasons managers do not call in are not financial. They are that it is awkward, that the person on the other end of the phone is having a day off, and that the manager will personally have to work harder for the twenty minutes before help arrives either way. Recognize those as what they are, and make the call — and note that in some jurisdictions, calling someone in on short notice or changing a posted schedule carries premium-pay obligations (Chapter 20). Build that into what a call-in costs where you operate.

👨‍🍳 On the Line

A sauté cook alone on two stations for four hours.

This is the line from Figure 19.7 that ought to bother you most, because it is the one that costs nothing on any report Bellwether produces.

Here is what it is. Two stations, four feet apart. Pasta water, six sauté pans, a salamander, and the pantry: cold apps, salads, the crudo, dressings, garnishes for everything the hearth sends. On a 138-cover night that is somewhere north of 300 individual plates through one person's hands, at a pace that leaves no seconds anywhere.

What actually happens is that the cook stops doing the parts nobody checks. Garnishes get simpler. The plate wipe gets skipped. A sauce that should be re-emulsified goes out broken because the alternative is a ticket dying on the rail. Nothing here is a failure of character; it is the only available response to a physically impossible workload.

Then service ends, and the labor report shows a \$168 saving.

Every hour that station ran short degraded the product, and not one cent of that degradation appears anywhere in the accounting. It shows up eleven weeks later in a review that mentions the pasta, and in a cook who starts checking job postings on her day off. This is what people mean when they say a restaurant is only as good as its worst night, and it is why the number this chapter teaches you to manage is not the only thing you are managing.

Chapter 21 owns what this does to people. This chapter owns the hours and the money, and the hours and the money say the same thing: that station should not have run alone, and the schedule knew it eight days before service.

19.7 Overtime: how it happens, what it really costs, and how to see it coming

Overtime is hours worked beyond a defined threshold — under the federal Fair Labor Standards Act, generally more than 40 in a workweek for non-exempt employees — paid at a premium rate, typically one and a half times the regular rate. Some states impose daily thresholds as well. Chapter 20 owns the law, including the genuinely tricky question of which salaried people are actually exempt from it. This section owns the management of it.

What an overtime hour actually costs

More than you think, for three reasons that compound.

🧮 Run the Numbers

The true cost of one overtime hour. (illustrative rates)

Bellwether's hearth cook earns \$21.00. An overtime hour is billed at 1.5×, or \$31.50.

	Straight hour	Overtime hour
Wage	\$21.00 | \$31.50
Payroll taxes (9.2%)	\$1.93 | \$2.90
Workers' compensation (≈2.4% of payroll)	\$0.50 | \$0.76
All-in	\$23.43** | **\$35.16

The premium is not \$10.50. It is **\$11.73**, because taxes and workers' compensation ride on the premium too.

Now the comparison that matters. That same hour of coverage could have been scheduled to the garde manger at \$18.00 — \$20.08 all-in. The overtime hour costs \$15.08 more than the scheduled hour that would have prevented it, a 75% penalty, and it buys you a cook in hour forty-one who is slower than the cook in hour four.

Bellwether's guide, written correctly, contains zero scheduled overtime: the hearth station is 39.5 hours, sauté 39.5, dish split between two people. A restaurant that leaks even 1.5% of hourly wages to overtime premium — a modest, entirely ordinary rate — spends about \$5,200 a year, or 0.34 points of labor, on nothing.

The four ways overtime actually happens

Almost no overtime is deliberate. It arrives through four doors:

1. The 42-hour schedule. A manager writes five 8.5-hour shifts because 8.5 is the shift length, and never adds the column. Five times 8.5 is 42.5, and the employee is 2.5 hours into premium pay before the week starts. This is the most common cause and the easiest to fix: add the row.
2. Coverage. Someone calls out, and the person who covers is already at 36 hours. Every hour of the covered shift is a premium hour. The October Friday is this case exactly.
3. Creep at the edges. A cook who clocks in eight minutes early and out twelve minutes late every shift adds 1.7 hours a week. Five shifts, four cooks, fifty-two weeks: 347 hours a year, much of it at premium because it pushes people over 40. Off-the-clock work is a serious legal problem in the other direction — Chapter 20 — so the answer is never "don't pay it." The answer is a clock-in policy, enforced, with managers who notice.
4. The heroic employee. The person who says yes to everything, picks up every shift, and is at 48 hours by Saturday. They are your best employee and they are the most expensive person on the schedule, and in about nine months they will leave.

Seeing it coming: the Thursday-morning hours report

Overtime is entirely preventable and almost never prevented, because managers look at hours after the week closes. By then the premium is owed.

The countermeasure is one report, run mid-week, showing hours-to-date against scheduled hours-remaining for every hourly employee.

FIGURE 19.8 — Hours-to-date report, Thursday 9:00 a.m.               [the Bellwether plan]

  Workweek runs Monday–Sunday. Report shows actual worked Mon–Wed
  plus hours still scheduled Thu–Sun.

  EMPLOYEE (role)        WORKED   SCHEDULED   PROJECTED   OT       ACTION
                          MON-WED    THU-SUN      WEEK   RISK
  ────────────────────────────────────────────────────────────────────────────
  Hearth cook              14.5       25.0        39.5    —        ok
  Sauté cook               15.5       26.0        41.5    1.5 hr   ⚠ trim Sun close
  Garde manger              5.0       19.5        24.5    —        ok
  AM prep cook             12.0       20.5        32.5    —        ok
  Dishwasher A             12.5       23.0        35.5    —        ok
  Dishwasher B              0.0       18.0        18.0    —        ok
  Brunch cook (lead)        0.0       13.0        13.0    —        ok
  Server 1                 13.0       20.5        33.5    —        ok
  Server 2                 10.5       20.5        31.0    —        ok
  Server 3                  4.5       28.5        33.0    —        ok
  Bartender A              13.0       22.0        35.0    —        ok
  Host A                    4.0       15.5        19.5    —        ok
  ────────────────────────────────────────────────────────────────────────────
  PROJECTED OVERTIME THIS WEEK: 1.5 hours ≈ $32 premium, all-in

  The sauté cook picked up 1.5 hours covering Wednesday's close. Moving Sunday's
  brunch-line shift to the second brunch cook removes the entire exposure and
  costs $0, because the hours were already budgeted to that station.

That report takes a scheduling system ninety seconds to produce and a manager four minutes to read. It is the difference between knowing about overtime on Thursday, when it is free to fix, and knowing about it on Monday, when it is owed.

One structural note. Notice the report defines the workweek as Monday–Sunday while Bellwether's operating week runs Tuesday–Sunday. That is deliberate: it puts the closed day at the start of the payroll week, so a Saturday call-in lands in the middle of a workweek rather than pushing a Sunday-Monday employee over 40 across a boundary nobody was watching. Your defined workweek is a choice you make once, in writing, and it has real consequences for how overtime accrues. Chapter 20 covers the rules for setting and changing it.

⚠️ Where the Money Leaks

The overtime you pay to avoid hiring.

Here is a decision that feels prudent and is usually wrong. A cook leaves. Rather than replace them, you spread the hours across the remaining three cooks — five extra hours each.

	Replace the cook	Spread the hours
15 hours a week at	\$18.00 straight | \$29.25 average premium rate
Weekly wages	\$270.00 | \$438.75
All-in weekly	\$294.95 | \$479.29
Over 12 weeks	\$3,539 | **\$5,752**

Twelve weeks of "we'll manage" costs \$2,213 more than hiring — and Chapter 17 already made you compute what replacing a burned-out cook costs on top of that.

The reason operators do it anyway is that hiring is a project and overtime is a decision you can avoid making. It also looks like restraint on a headcount report. It is not restraint. It is a more expensive version of the same labor, paid to fewer people, who are getting tired.

19.8 Reading the labor report: variance, trend, and where the hours actually went

Every week, one page. Not a payroll register — a report you can read in six minutes and act on the same morning.

🧾 Read the Numbers

```text FIGURE 19.9 — "Weekly labor report, week 31" [the Bellwether plan] THE ARTIFACT The one-page weekly labor report a manager reads Monday morning: budget versus actual for dollars, hours, and productivity, by department. THE CONTEXT Bellwether, late July, week 31. The forecast from Figure 19.2. Two of the four original line cooks have left; one replacement is in week two.

                                 FORECAST      ACTUAL     VARIANCE
               Net sales          $30,433     $29,940        -$493
               Covers                 726         701          -25

               HOURLY WAGES
                 Back of house     $3,995      $4,268        +$273
                 Front of house    $2,781      $2,802         +$21
                 Overtime premium      $0        $187        +$187
                 Hourly subtotal   $6,776      $7,257        +$481
               Salaries             $2,885      $2,885          $0
               Payroll taxes 9.2%     $889        $933         +$44
               Workers' comp          $231        $231          $0
               Benefits               $308        $308          $0
               ───────────────────────────────────────────────────
               TOTAL LABOR         $11,089     $11,614        +$525
               LABOR % OF SALES      36.4%       38.8%      +2.4 pt

               HOURS
                 BOH                  216.5       228.0        +11.5
                 FOH                  244.5       247.5         +3.0
                 Total hourly         461.0       475.5        +14.5
               SPLH (hourly hrs)     $66.02      $62.97       -$3.05
               CPLH (hourly hrs)       1.58        1.47         -0.11

WHAT IT SHOWS A 2.4-point miss, and the report is specific about where. Sales came in $493 light while hourly hours came in 14.5 heavy — the two moved in opposite directions, which is the worst combination and the one that turns a small miss into a visible one. Eighty percent of the hour overage is in the back of house. SPLH fell $3.05, which says the hours were genuinely unproductive rather than merely mistimed. And $187 of overtime premium appeared in a schedule that was written with zero. WHAT IT DOESN'T It does not say WHY. Eleven and a half extra BOH hours could be a cook staying late three nights, a prep list that did not get finished, a new hire working slow, a walk-in that had to be reorganized, or a manager who forgot to send someone home. Those are five different problems with five different fixes, and this page cannot distinguish them. It does not show the day. The week's sales miss is one service — Sunday brunch, forecast 104 and did 82 — and the hour overage is a different set of days entirely. An aggregate report hides the fact that these are two unrelated problems. It does not show WHO. Nothing here tells you the overtime premium is one person, or that the person is the sauté cook, who has now covered three of the last four short-staffed closes. THE DECISION Pull the daily detail before doing anything else, and ask the two questions the report generates rather than answers: which days carry the 11.5 BOH hours, and who worked the overtime. If the answer is "the new cook is slow," that is a training cost with a known end date and you leave it alone (Chapter 18). If the answer is "prep is not finishing by four," the fix is the prep list, not the schedule (Chapter 14). Fix Sunday's forecast rule separately. THE LESSON A labor report is a question generator, not an answer. Its whole job is to tell you which four minutes of the week to investigate. An operator who reads it and reacts to the percentage will cut hours on the wrong days. ```

The three readings, in order

Variance is this period against its own budget. Read it in hours first, then dollars, then percentage — in that order, always. Hours are the physical fact; dollars add the rate mix; the percentage adds the sales denominator and is therefore the last thing you should look at, because it is the only one of the three that can move without anyone doing anything.

Week 31's percentage moved 2.4 points. Only about 1.1 of those points came from hours; the rest came from a sales miss on a rainy Sunday. A manager who reads the percentage first responds by cutting Tuesday, which had nothing to do with it.

Trend is this period against the last several. One bad week is weather. Four weeks of BOH hours running 5% over is a system. Keep a rolling four-week line for hours, SPLH, and labor percentage, and treat the trend as the real signal.

Cause is where the hours went, and it always requires the daily detail: hours by day, by department, by position, against that day's forecast and actual covers. The weekly report tells you to look. The daily detail tells you where.

Splitting week 31's miss takes about ninety seconds and is worth doing every time. Hold one variable at a time:

Of the 2.4-point miss, 1.8 points is a labor problem and 0.6 points is a sales problem. They live in different departments and have different owners. Reporting them as a single number guarantees that the wrong person gets asked about it.

The trend that matters most: the ramp and the run rate

Now the hardest arithmetic in Bellwether's plan, and the reason this chapter exists.

Chapter 9 established that Bellwether's first quarter runs at roughly 66.6% prime cost on \$363,100 of ramp revenue. That is not a failure — it is what an opening quarter is. You are staffed for the restaurant you intend to be, serving the volume of the restaurant you currently are, with a crew that has never worked together and a kitchen still finding its yields.

But the year has a target, and the target is 60.0% prime cost on \$1,550,000. Work out what that obligates the remaining thirty-nine weeks to do.

🧮 Run the Numbers

What weeks 14–52 are actually required to deliver.

Step	Arithmetic	Result
Annual prime cost dollars	COGS \$430,280 + labor \$500,000	\$930,280 (60.0%)
Q1 prime cost dollars	Chapter 9's quarterly build	**\$241,714** (66.6% of \$363,100)
Q1 cost of goods sold, at 27.8%	27.8% × \$363,100 | \$100,942
Q1 labor, therefore	\$241,714 − \$100,942	\$140,772 = 38.8% of Q1 sales
Weeks 14–52 sales	\$1,550,000 − \$363,100	\$1,186,900
Weeks 14–52 prime allowed	\$930,280 − \$241,714	\$688,566 = 58.0%
Weeks 14–52 COGS, at 27.8%	27.8% × \$1,186,900 | \$329,958
Weeks 14–52 labor allowed	\$688,566 − \$329,958	\$358,608
As a percentage	\$358,608 ÷ \$1,186,900	30.2%
Per week, over 39 weeks	\$358,608 ÷ 39 | **\$9,195**

Against a headline labor target of 32.3%, weeks 14 through 52 must average about 30.2% — two full points tighter than the number on the front of the plan — because the ramp already spent the cushion.

⚠️ Notice where the top two rows came from, because it matters more than the answer. Both are the plan's dollar figures, not its percentages rebuilt. Reconstruct COGS from the displayed 27.8% and you get \$430,900 — \$620 of money that does not exist. Reconstruct Q1 prime from the displayed 66.6% and you get \$241,825 instead of Chapter 9's \$241,714. Do both, and this table reports that weeks 14–52 must run at 58.1%, when the figure Chapter 9 and Chapter 31 both publish is 58.0%. One tenth of a point on this line is about \$1,187 a year, and nothing produced it except reading percentages back into dollars. Dollars are canonical. Percentages are rounded displays.

And now put the two numbers side by side:

	Weekly labor dollars	Weekly sales	Labor %
Q1 (weeks 1–13), per the plan	\$10,837 | \$27,931	38.8%
The schedule this chapter built	\$10,970	—	—
Weeks 14–52, per the plan	**\$9,208** | \$30,433	30.3%

Read that carefully, because it is the single most important line in this chapter.

Q1's budgeted labor dollars are almost exactly the full schedule. The plan funds \$10,837 a week during the ramp; the real schedule costs \$10,970. Close enough that Q1 is nearly honest.

Then, in week 14, the plan removes \$1,762 a week — and never gives it back. Over thirty-nine weeks that is \$68,729.

Bellwether's plan asks the restaurant to spend the most labor when it has the fewest guests and the least labor when it has the most. That is exactly backwards, and it is not a mistake anyone made on purpose. It is what happens automatically when labor is modeled as fixed plus a percentage of sales instead of built from a schedule: the percentage term shrinks the labor line in the quarters where sales are low and expands it where sales are high, which is the opposite of how a staffing guide behaves.

Can a staffing guide deliver 30.2%? Show the work.

\$9,208 a week, all-in. Unwind it to hours:

The plan funds 344 hourly hours a week in weeks 14–52. The staffing guide needs 453.5.

The gap is 110 hours a week.

Here is what 110 hours actually is, laid against the guide:

That is the schedule the plan funds: a three-station line on a 123-cover Saturday with the chef working the pantry while expediting, no bussers, no barback, one dishwasher, three servers on Friday, and one cook doing brunch alone.

We have seen that restaurant. It is the second Friday in October, made permanent. That night produced eleven tables past standard, a fifteen-hour day, a station running alone for four hours, and a Saturday morning spent chasing a delivery — and it produced all of that on a night that was only 9.5 hours short.

FIGURE 19.10 — Labor as a percentage of sales: the plan vs. the build   [the Bellwether plan]

  45% ┤
      │  ▓▓▓▓▓▓▓▓  38.8%  Q1 as planned — and the real schedule costs about this
  40% ┤  ▓▓▓▓▓▓▓▓         ┌────────────────────────────────────────────┐
      │  ▓▓▓▓▓▓▓▓         │ ████ 36.0%  the staffing guide, run rate,  │
  35% ┤  ▓▓▓▓▓▓▓▓         │ ████        on planned weeks 14–52 sales   │
      │  ▓▓▓▓▓▓▓▓         │ ████                                        │
  30% ┤  ▓▓▓▓▓▓▓▓         │ ▒▒▒▒ 30.3%  what the plan allows            │
      │  ▓▓▓▓▓▓▓▓         └────────────────────────────────────────────┘
  25% ┤  ▓▓▓▓▓▓▓▓            ▲
      └──┴────────┴──────────┴──────────────────────────────────────────
         WEEKS 1-13          WEEKS 14-52
                             ▲
                    THE GAP: 5.8 points = $68,729 = 110 hours a week

  Note the shape. The plan's labor line FALLS as volume RISES, because it was
  modeled as a percentage of sales rather than built from a schedule.

What would give — the four honest levers, priced

Nothing on this list is free. Each one buys points with something.

Levers 1 through 4 are real, achievable, and worth doing regardless. They get Bellwether from 36.0% to about 32.6% in weeks 14–52 — which, worth noting, roughly hits the headline 32.3% the plan advertises. They do not get to 30.3%.

The last 2.3 points has to come from one of exactly three places, and an operator should choose consciously rather than discover the choice in February:

A. Change the kitchen's production model. Trade labor for cost of goods sold. Buying stocks, portioned proteins, and pre-cut produce would remove roughly 41 hours a week of prep and butchery — essentially the entire AM prep position plus a shift — and raise COGS by about 2.4 points, from 27.8% to roughly 30.2%. Prime cost does not move. This is Chapter 1's trade, exactly: you can put the cost in either half of prime, and the only honest question is which half your concept can afford. For a wood-fired, scratch, chef-driven restaurant whose entire proposition is that the food is made here, that trade is not a scheduling decision. It is a decision about what Bellwether is, and it belongs to the chef-owner and to Chapter 2's concept work, not to a manager with a blank grid.

B. Reduce what the owners pay themselves. Dropping the two partners from \$52,000 and \$50,000 to \$38,000 each removes \$26,000 of wages, \$28,392 all-in — about 1.7 points in weeks 14–52. This is by far the most common way the gap actually closes in American independents, and it is worth naming plainly rather than pretending otherwise: a large share of independent restaurants balance their labor line on the unpaid or underpaid hours of the people who own them. It works. It is not free, it is not sustainable indefinitely, and it makes the business unsellable and unstaffable at the top — because the day you hire a general manager to do the FOH partner's job, the cost reappears at market rate.

C. Sell more than the plan says. To reach 30.3% on the post-lever labor of \$417,894, weeks 14–52 would need about \$1,383,700** of sales rather than \$1,186,900 — roughly 17% above plan**. On a 68-seat room with a hearth that caps the kitchen at 144 covers a night, that is not impossible, but it requires Friday and Saturday running at the ceiling and Tuesday through Thursday up by a third. Chapters 22, 23, and 24 are where that work lives, and it is the only lever on this list that makes the business better rather than smaller.

⚠️ Where the Money Leaks

The fifth lever, which is not a lever.

There is a fifth way the gap closes, and it closes the gap on paper in almost every restaurant that eventually fails: you write the schedule you can afford, run it, and let the difference come out of the guests and the staff.

Nobody decides to do this. It happens by accumulation. A busser is not replaced. The garde manger's Wednesday shift quietly disappears. The host starts at 5:30 instead of 4:30. The prep cook stops coming in on Saturday because Saturday is expensive. Each individual decision is defensible, saves a real \$40, and moves the labor percentage in the right direction.

Six months later the restaurant is running a three-station line, the sous is on a fifteen-hour day most Fridays, the review average has slid from 4.5 to 4.1, and the labor line reads 31.5% — which looks like excellent management.

This is the single most dangerous number in this chapter, because it is the one that looks like success. A labor percentage that improves while ticket times, review scores, turnover, and comps all worsen is not an improvement. It is a transfer — from the guest experience and from your staff, into the labor line, at a very poor exchange rate. Chapter 21 shows you what it costs on the people side. Chapter 23 shows you what it costs on the revenue side. Both are larger than what it saves.

🔍 Check Your Understanding

1. Week 31's labor percentage missed budget by 2.4 points. How much of that is a labor problem, and how would you demonstrate the split to a skeptical owner?
2. Why does Bellwether's plan require 30.3% labor in weeks 14–52 when the front page of the plan says 32.3%?
3. An operator closes the labor gap by buying pre-portioned proteins. What happens to prime cost, and what has actually been decided?

(1: 1.8 points is hours; 0.6 points is the sales miss. Demonstrate it by holding one variable at a time — budgeted labor over actual sales isolates the denominator effect, then actual labor over actual sales adds the hours effect. 2: Because Q1 ran 66.6% prime on \$363,100 and consumed more than its share of the annual prime-cost budget; the remaining 39 weeks must absorb the overage on \$1,186,900 of sales. 3: Prime cost is unchanged — cost moved from the labor half to the COGS half. What has been decided is not a scheduling matter but a concept matter: what the kitchen makes and what it buys.)

🍽️ The Business Plan

Checkpoint 19 of 40 — the Labor Model.

This chapter contributes the Labor Model section: the staffing guide, the SPLH and CPLH targets, and the labor line built bottom-up from actual positions instead of assumed from a percentage.

What goes in the plan

1. The staffing guide (Figures 19.3 and 19.4), stated as a labor matrix in cover bands:

2. The roster: 24 positions — 3 salaried, 21 hourly. Eight back-of-house hourly, thirteen front-of-house hourly.

3. The targets:

4. The labor line, built bottom-up:

What this section does not settle, stated plainly

The plan's labor line is \$500,000 and the schedule costs \$570,461. The plan's number is not changed here — every projection in the document depends on it, and changing it to match the schedule would simply move the argument somewhere less visible. Instead the plan carries the gap explicitly:

• The gap is \$70,461 — 4.5 points of sales — and it is 84 hourly hours a week.
• In weeks 14–52 the gap is worse, because the ramp already spent the cushion: the plan allows 30.3% where the schedule costs 36.0%. That is 110 hours a week.
• Disciplined scheduling closes about 1.2 points of it. Revenue above plan closes about 2.3 more. The remaining 2.3 points must come from the kitchen's production model, from the owners' compensation, or from sales roughly 17% above plan — and the plan should say which, in writing, before anyone is hired.

That is not a comfortable page to put in a business plan. It is a far better page than a labor line that foots because someone assumed a percentage. A reader who can trace 453.5 hours to \$570,461 will believe your other numbers. A reader who finds "labor: 32.3%" with no schedule behind it has no reason to believe anything in the document.

Open questions carried forward

1. Which of the three structural levers does Bellwether choose — production model, owner compensation, or volume? (Chapters 24 and 40)
2. What does the wage model look like once the tip credit, tip pooling, and the exempt/non-exempt question are settled — and does the sous chef's overtime in Figure 19.7 tell us something about how that position is classified? (Chapter 20)
3. At 75% industry turnover, how many of the 24 positions will be filled by someone new in year one, and what does the resulting training inefficiency do to SPLH? (Chapters 17, 18, 21)
4. Does a schedule that runs a sous chef fifteen hours on a Friday survive its own staff? (Chapter 21)
5. Can the room and the menu produce the 17% of additional revenue that would make the labor line work without cutting anything? (Chapters 22, 23, 24)

Conclusion

The schedule is a financial document, and it is the only one in the building you write before the money moves rather than after.

We built Bellwether's labor line from the bottom: 24 positions, 453.5 hourly hours a week, \$496,541 of wages, \$45,885 of payroll taxes, \$12,035 of workers' compensation, \$16,000 of benefits — **\$570,461**, or 36.8% of sales. The plan says \$500,000. The gap is real, it is 4.5 points, and it is worth more than the restaurant's entire projected operating profit.

Three things made that gap, and all three are ordinary.

Labor was modeled as a percentage of sales instead of built from a schedule, which produced a model that funds the most hours in the quarter with the fewest guests and the fewest hours in the quarter with the most. The fixed floor is larger than it looks — \$191,895 of salaries, taxes, insurance, and benefits that do not move, 12.4% of sales, spent before a single hourly hour is scheduled. And staffing is a staircase, so the cost of the covers between 110 and 130 bears no resemblance to the cost of the covers between 85 and 110, and a schedule built on an average is wrong in both directions at once.

The chapter's harder lesson is the one from the second Friday in October. That night cost \$116.70 in hard dollars — 1.7% of its own sales — and the restaurant came out twelve dollars ahead on wages, because a cook who did not show up costs less than a cook who does. Read only the labor report and it was a good night. Every single thing that actually happened — the fifteen-hour day, the eleven tables past standard, one cook alone on two stations for four hours — appears on no report Bellwether produces. A labor report measures dollars paid. It does not measure work done, and it does not measure damage. If it is your only instrument, it will reward your worst services.

So: can a staffing guide deliver 30.2% in weeks 14–52? Honestly, no — not by itself. Scheduling discipline is worth about 1.2 points. Revenue above plan is worth about 2.3 more. The last 2.3 points comes out of the kitchen's production model, the owners' paychecks, or a dining room selling 17% above plan. Those are real options and an operator can choose among them. What an operator cannot do is write a tighter schedule and expect the arithmetic to work, because the hours that would have to disappear are the garde manger, the bussers, the barback, the second dishwasher, and a server on each of the two biggest nights — and that restaurant has a different product.

Chapter 20 takes up the law that sits underneath everything in this chapter: what the Fair Labor Standards Act requires, how the tip credit works and where it does not exist, who may participate in a tip pool, which salaried manager is actually owed overtime, and what predictive-scheduling ordinances demand of the document you just learned to write. Every technique in this chapter has a compliance boundary, and the boundary is not optional.

Key Terms

Labor cost percentage — total labor cost — wages, employer payroll taxes, workers' compensation, and benefits — divided by total sales. The half of prime cost an operator can move this week. State the denominator you used, every time. (Ch. 19)

Fixed labor — labor that does not vary with volume: salaried positions and the hourly coverage a restaurant must have to open its doors at all. Bellwether's fixed floor is \$191,895, or 12.4% of sales. (Ch. 19)

Variable labor — labor that moves with volume, in steps rather than smoothly: the fourth server, the second dishwasher, the barback. (Ch. 19)

Sales per labor hour (SPLH) — net sales divided by labor hours. The productivity number a schedule is managed to; comparable across your own history but sensitive to check average, so it is not comparable across restaurants at different price points. (Ch. 19)

Covers per labor hour (CPLH) — covers divided by labor hours. The production number; comparable across similar concepts and the better measure of whether a kitchen can physically produce the night you booked. (Ch. 19)

Labor forecasting — projecting covers and sales for each future service period in enough detail to staff it, built from day-of-week history, the reservation book adjusted by conversion, the calendar, the weather, and judgment applied last. (Ch. 19)

Staffing guide (also labor matrix) — a document stating, for each service and each band of forecast volume, exactly which positions are scheduled and for how many hours. It converts a forecast into a schedule mechanically, so the same volume produces the same staffing every time. (Ch. 19)

Overtime — hours beyond a defined threshold, generally more than 40 in a workweek under federal law, paid at a premium rate. All-in, an overtime hour at Bellwether costs about 75% more than the scheduled hour that would have prevented it. (Ch. 19)

Split shift — one employee working two separated blocks in a single day with unpaid time between. A real tool for matching prep and service demand, and a real cost in retention. (Ch. 19)

Cutting the floor — releasing staff early as volume declines. The primary real-time labor lever; it should be executed against a written cut order with objective conditions, set before service rather than during it. (Ch. 19)

Spaced Review

1. Without looking back: name the four components of labor cost, and state which two most operators leave out of the number they quote at a party.
2. Bellwether's Sunday brunch runs 48.5% labor — the worst service on the schedule. Explain, using contribution rather than percentage, why closing it would make the restaurant's labor percentage worse.
3. Prime cost, from Chapter 1. An operator moves from butchering whole chickens in house to buying pre-portioned chicken. Their labor percentage falls 2.4 points and their food cost rises 2.4 points. Has anything improved? What would you need to measure to find out?
4. A Friday forecast at 112 covers sits just above the 111-cover riser in Figure 19.5. Explain why 112 is the worst forecast number on Bellwether's board, and name two things a manager can do about it that are not "cut hours."
5. The recurring question. Figure 19.7 shows a night on which the restaurant came out \$12 ahead on wages and the labor report improved. Name three costs that night generated that appear on no report in the building, and say for each one which later chapter has to price it.