Appendix H — Worked Financial Examples
Twenty problems, carried from a stated situation to a decision, with every arithmetic step shown.
These are not exercises. The exercises are in the chapters, and they are for you to do. These are worked examples — you watch someone competent do the calculation, including the judgment calls the formula does not contain, and including the places where the number turns out to be the beginning of the question rather than the end of it.
Every example follows the same six-part shape.
THE SITUATION — what happened, and what you are being asked. WHAT YOU KNOW — the inputs, and where each came from. THE WORK — every step. No skipped lines. THE ANSWER — the number, plainly. THE DECISION — what the operator does, which is frequently not what the number alone implies. WHAT THE NUMBER DOESN'T TELL YOU — the limits.
That last field is the point of the appendix. A worked example that stops at the answer has taught you arithmetic. The useful thing is knowing what the arithmetic is silent about.
All figures are constructed and illustrative. Where an example uses Bellwether — the book's
running 68-seat restaurant — it is labeled [the Bellwether plan] and its numbers match the plan exactly.
The other restaurants here are one-off constructions, deliberately, so that the technique is tested on
something other than the one business you have been reading about for forty chapters.
One standing rule, because three of these examples turn on it: dollar figures are canonical and percentages are rounded displays. Never reconstruct a line item from a percentage.
H.1 Costing
Example 1 — Costing a dish, with yield and waste
THE SITUATION The Larder is a constructed 74-seat bistro. The chef wants to put a braised lamb shoulder on the autumn menu at \$27 and needs to know whether that price works before it is printed.
WHAT YOU KNOW
Input Value Source Lamb shoulder, bone-in \$7.40 per lb as purchased this week's invoice Butchering yield 62% the kitchen's own yield test Portion size 6 oz edible portion the chef's spec White bean ragout \$0.92 per portion sub-recipe cost card Gremolata \$0.31 per portion sub-recipe cost card Braising liquid and aromatics \$0.44 per portion batch cost ÷ portions Oil, salt, and miscellaneous \$0.16 per plate standing allowance Garnish \$0.19 per plate cost card Waste allowance 2% of subtotal house convention THE WORK
```text STEP 1 — Convert as-purchased cost to edible-portion cost. $7.40 / 0.62 = $11.9355 per usable pound
STEP 2 — Cost the portion. 6 oz = 0.375 lb. 0.375 x $11.9355 = $4.4758 -> $4.48
STEP 3 — Extend every line. Lamb, 6 oz EP .................. $4.48 White bean ragout .............. $0.92 Gremolata ...................... $0.31 Braising liquid and aromatics .. $0.44 Oil, salt, misc ................ $0.16 Garnish ........................ $0.19 ──────────────────────────────────── INGREDIENT SUBTOTAL ............ $6.50
STEP 4 — Waste allowance. 2% x $6.50 = $0.13 PLATE COST = $6.50 + $0.13 = $6.63
STEP 5 — Test the price. Food cost %: $6.63 / $27.00 = 24.556% -> 24.6% Contribution margin: $27.00 - $6.63 = $20.37 ```
THE ANSWER Plate cost \$6.63**. At \$27.00 the dish runs 24.6% food cost and contributes \$20.37**.
THE DECISION Put it on at \$27. And notice why it works: not because lamb is cheap — it is the most expensive line on the card — but because a 62% yield on a bone-in cut still lands a six-ounce portion at \$4.48. Run the yield test before you price anything with a bone in it. Had the yield come in at 54% instead, the protein line would be \$5.14 and the plate \$6.98 — still fine at \$27, which is the other thing worth knowing.
WHAT THE NUMBER DOESN'T TELL YOU The card prices ingredients. It does not price the six hours of braising, the walk-in space the hotel pans occupy overnight, or the fact that this is a dish that punishes a cook who is behind. It also assumes 62% holds — and the yield you get depends on who is breaking down the shoulder, which is a training question dressed as a costing input.
Example 2 — True cost of goods sold, and the adjustment that changes the diagnosis
THE SITUATION Pike & Vine is a constructed 70-seat restaurant. The chef says food cost is running "about 33," which would be three points over target, and wants to change purveyors. The bookkeeper has closed the four-week period. Before anybody calls a purveyor, compute the number properly.
WHAT YOU KNOW
Input Value Beginning food inventory \$18,400 Food purchases \$61,200 Ending food inventory \$17,050 Food sales \$189,000 Staff meals \$2,340 Comps \$1,180 Transfers, kitchen → bar (citrus, cream, herbs) \$640 Transfers, bar → kitchen (wine for cooking) \$310 THE WORK
```text STEP 1 — Usage. This is the formula, and there is no substitute for it. Beginning inventory $18,400 + Purchases +$61,200 - Ending inventory -$17,050 ───────────────────────────────── FOOD USAGE $62,550
STEP 2 — The unadjusted number, which is what most people quote. $62,550 / $189,000 = 33.095% -> 33.1%
STEP 3 — Remove what left the kitchen but was never sold as food. Staff meals -$2,340 Comps -$1,180 Net transfers OUT to the bar ($640 out less $310 back) -$330 ───────────────────────────────────────── Total adjustments -$3,850 ADJUSTED FOOD USAGE $62,550 - $3,850 = $58,700
STEP 4 — The operating number. $58,700 / $189,000 = 31.058% -> 31.1%
STEP 5 — Check it. 33.1% - 31.1% = 2.0 points 2.0% x $189,000 = $3,780, against $3,850 of adjustments. The residual is rounding in the displayed percentages, not a missing item. ```
THE ANSWER Unadjusted 33.1%. Adjusted 31.1%. Two full points of the "problem" are staff meals, comps, and a transfer to the bar.
THE DECISION Do not call the purveyor. The kitchen is running 31.1% against a 30% target — a 1.1-point gap worth about \$2,079 in the period, which is a real conversation about portioning and waste, not a purchasing crisis. Then fix the reporting so this never happens again: the chef gets the adjusted number, because it measures what the kitchen controls. A lender or a landlord gets the adjusted number with the reconciliation attached, because the reconciliation is the evidence that you count.
WHAT THE NUMBER DOESN'T TELL YOU It does not tell you whether \$2,340 of staff meals is the right amount to spend — that is a culture decision with a cost, and it should be a decision rather than a leak. It does not tell you why there were \$1,180 of comps. And it depends entirely on the ending count being right, which is Example 3.
Example 3 — Chasing a variance, in the right order
THE SITUATION The next period at Pike & Vine. Theoretical food cost — every cost card extended at the actual menu mix — says 29.4%. Actual adjusted food cost says 32.1%. That is 2.7 points on \$196,000 of food sales. The chef's first instinct is that someone is stealing.
WHAT YOU KNOW Theoretical 29.4% · actual adjusted 32.1% · food sales \$196,000 · the diagnostic order this book insists on: arithmetic, then receiving, then portioning, then waste and comps, then theft — which is the order of likelihood.
THE WORK
```text THE GAP 2.7 points x $196,000 = $5,292 for the period x 13 periods = $68,796 a year, if it is structural
STEP 1 — ARITHMETIC. Re-add the count sheet before anything else. Found: 22 cases of chicken counted as 22 POUNDS. Counted: 22 x $3.20 = $70.40 Actual: 22 cases x 40 lb x $3.20 = $2,816.00 Ending inventory understated by $2,745.60 An understated ending inventory OVERSTATES usage by the same amount. $2,745.60 / $196,000 = 1.4 points Corrected actual: 32.1% - 1.4% = 30.7% REMAINING GAP: 30.7% - 29.4% = 1.3 points = $2,548
STEP 2 — RECEIVING. Invoices re-checked against the order guide. Two short deliveries credited, one price increase not caught. Net: clean.
STEP 3 — PORTIONING. Yield check on the lamb. Spec 6 oz; four consecutive plates averaged 7.1 oz. $4.48 x (7.1 / 6) = $5.30 per plate, +$0.82 310 plates in the period x $0.82 = $254
STEP 4 — WASTE AND COMPS. Unlogged spoilage found in the walk-in $980 Comps rung without a reason code $610
STEP 5 — RESIDUAL, after all of the above. $2,548 - $254 - $980 - $610 = $704 ```
THE ANSWER \$2,745.60 of the "variance" was a counting error.** Of the genuine \$2,548 that remained: \$254 portioning, \$980 unlogged waste, \$610 uncoded comps, and \$704 unexplained**.
THE DECISION Fix the count sheet template so units are printed next to every line — that is a fifteen-minute fix that was worth \$2,745 this period. Re-spec and re-train the lamb portion. Put a scale on the line. Require a reason code on every comp. And leave the \$704 alone for one more period. A residual of 0.36 points is inside the noise of any restaurant; chasing it with an accusation costs more than it recovers, and if it is real it will still be there next period and will be bigger.
WHAT THE NUMBER DOESN'T TELL YOU Nothing in this arithmetic distinguishes theft from sloppiness, and it never will. What it does is make theft the last hypothesis instead of the first — which matters, because starting there is usually wrong and always corrosive. Note also that the whole exercise depends on the theoretical number being right, and the theoretical number is only as good as the cost cards, which are only as good as the last time anybody recosted them.
Example 4 — A vendor price increase: absorb it or reprice?
THE SITUATION
[the Bellwether plan]Poultry goes up 12%. The Hearth Chicken is the restaurant's signature dish and its reference cost card. Do you eat the increase or move the price?WHAT YOU KNOW The frozen card: plate cost \$8.35** before waste, **\$8.52 after the 2% allowance, menu price \$29.00**, food cost **29.4%**, contribution margin **\$20.48. The chicken line is \$5.60 — half of a 3.5 lb bird at \$3.20 a pound. The dish sells about 6,400 covers a year.
THE WORK
```text STEP 1 — Reprice the affected line only. $3.20 x 1.12 = $3.584 per lb 1.75 lb x $3.584 = $6.272 -> $6.27 Increase: $6.27 - $5.60 = $0.67
STEP 2 — Rebuild the card. New subtotal: $8.35 + $0.67 = $9.02 Waste at 2%: $0.18 NEW PLATE COST: $9.20 (was $8.52, up $0.68)
STEP 3 — What it does at the current price. Food cost: $9.20 / $29.00 = 31.724% -> 31.7% Contribution margin: $29.00 - $9.20 = $19.80 (was $20.48)
STEP 4 — The annual bill. $0.68 x 6,400 covers = $4,352 a year As a share of revenue: $4,352 / $1,550,000 = 0.28%
STEP 5 — What price restores 29.4%? $9.20 / 0.294 = $31.29 -> $31.00 At $31.00: food cost $9.20 / $31.00 = 29.7% contribution $31.00 - $9.20 = $21.80 ```
THE ANSWER The increase costs \$4,352 a year and moves food cost on the dish from 29.4% to 31.7%. Restoring the margin requires \$31.00, a 6.9% increase on the most visible price on the menu.
THE DECISION Absorb it, for now, and set a trigger. \$4,352 is 1.7% of the plan's operating profit — real money and not an emergency. The signature dish is the one price your regulars know by heart, and moving it 6.9% for a commodity swing that may reverse in a quarter spends trust you cannot rebuy. Write the trigger down: if the poultry contract has not come back under \$3.40 by the next menu change, the price goes to \$31. Then reprice at a menu change, alongside other movement, rather than as a visible single-item increase.
WHAT THE NUMBER DOESN'T TELL YOU It cannot tell you the price elasticity of your own signature dish, and anyone who quotes you one is guessing. It also does not consider that a 12% poultry move rarely arrives alone — check the rest of the protein book before deciding this is a one-line problem.
H.2 Menu
Example 5 — Building the menu-engineering matrix
THE SITUATION Copper & Oak is a constructed neighborhood restaurant with eight entrées. One period's menu mix report is in. Classify every item.
WHAT YOU KNOW 2,000 entrées sold in the period, with prices and plate costs from the cost card file.
THE WORK
Item Units Mix % Price Plate cost CM Total CM A — Roast chicken 470 23.5% \$28.00 | \$8.40 \$19.60 | \$9,212.00 B — Strip steak 250 12.5% \$39.00 | \$14.82 \$24.18 | \$6,045.00 C — Pasta 380 19.0% \$24.00 | \$6.00 \$18.00 | \$6,840.00 D — Fish 210 10.5% \$32.00 | \$11.20 \$20.80 | \$4,368.00 E — Pork 240 12.0% \$29.00 | \$8.12 \$20.88 | \$5,011.20 F — Burger 300 15.0% \$21.00 | \$6.30 \$14.70 | \$4,410.00 G — Vegetarian 90 4.5% \$25.00 | \$7.50 \$17.50 | \$1,575.00 H — Lamb 60 3.0% \$36.00 | \$13.68 \$22.32 | \$1,339.20 Total 2,000 100.0% \$38,800.40 ```text THE TWO THRESHOLDS Popularity threshold = 70% / 8 items = 8.75% of mix Contribution threshold = weighted average CM = $38,800.40 / 2,000 = $19.40
CLASSIFY: high CM = at or above $19.40; high mix = at or above 8.75%
A $19.60 23.5% high / high -> STAR B $24.18 12.5% high / high -> STAR C $18.00 19.0% low / high -> PLOWHORSE D $20.80 10.5% high / high -> STAR E $20.88 12.0% high / high -> STAR F $14.70 15.0% low / high -> PLOWHORSE G $17.50 4.5% low / low -> DOG H $22.32 3.0% high / low -> PUZZLE```
THE ANSWER Four Stars, two Plowhorses, one Puzzle, one Dog. Weighted average contribution \$19.40; popularity threshold 8.75%.
THE DECISION The matrix has told you where to look, not what to do — and the two quadrants that need a decision are the Puzzle and the Dog. H (lamb) contributes \$22.32, the second-best on the menu, and sells sixty. That is a description problem or a position problem before it is a menu problem: move it up the page, have the servers taste it, and re-count in a period. G (vegetarian) contributes \$17.50 and sells ninety, which is the worst of both — but it is the only vegetarian entrée on the menu, and removing it removes a table's whole booking. Reprice it to \$27 (CM \$19.50, above the threshold) and rebuild the dish before you consider cutting it.
WHAT THE NUMBER DOESN'T TELL YOU The matrix cannot see that G is the reason a four-top with one vegetarian chose you. It cannot see that C, a Plowhorse, is what your regulars order on a Tuesday. And it re-sorts itself every time you change a price, so a quadrant is a snapshot of a menu, not a property of a dish.
Example 6 — The trap: food cost improves and the restaurant earns less
THE SITUATION Copper & Oak's manager wants to push the vegetarian entrée, which runs a 30.0% food cost, over the fish, which runs 35.0%. On its face that is exactly what a food-cost target asks for.
WHAT YOU KNOW From Example 5: total revenue \$56,420, total food cost \$17,619.60 (31.23%), total contribution \$38,800.40. The shift moves **100 covers** from **D (fish, \$32.00 / \$11.20 / CM \$20.80) to G (vegetarian, \$25.00 / \$7.50 / CM \$17.50)**.
THE WORK
```text BEFORE Revenue $56,420.00 Food cost $17,619.60 = 31.23% Contribution $38,800.40
AFTER — 100 covers move from D to G Revenue $56,420.00 - (100 x $32.00) + (100 x $25.00) = $56,420.00 - $3,200.00 + $2,500.00 = $55,720.00 Food cost $17,619.60 - (100 x $11.20) + (100 x $7.50) = $17,619.60 - $1,120.00 + $750.00 = $17,249.60 Food cost % $17,249.60 / $55,720.00 = 30.957% -> 31.0% Contribution $38,800.40 - (100 x $20.80) + (100 x $17.50) = $38,800.40 - $2,080.00 + $1,750.00 = $38,470.40
THE TWO RESULTS, SIDE BY SIDE Food cost percentage: 31.23% -> 31.0% IMPROVED by 0.27 points Contribution: $38,800.40 -> $38,470.40 FELL by $330.00 Annualized over 13 periods: -$4,290 ```
THE ANSWER Food cost percentage improves by 0.27 points and the restaurant makes \$330 less per period — \$4,290 a year.
THE DECISION Do not run the promotion. Then do the more useful thing: stop reporting food cost percentage as a performance metric to anyone who can change the menu mix. It is the right tool for costing a dish and the wrong tool for choosing which dish to sell, and a manager compensated on it will reliably make this trade. Report contribution per cover alongside it, every week.
WHAT THE NUMBER DOESN'T TELL YOU It assumes the hundred covers actually transfer, one for one. Some of the guests who would have ordered the fish will order something else entirely, and some would not have come. It also ignores the kitchen: if the fish station is the constraint on a Saturday and the vegetarian is not, moving volume off it may buy throughput worth more than \$330. That is a real argument — but it is a capacity argument, and it should be made on capacity grounds, not smuggled in as a food-cost win.
Example 7 — Pricing a new dish three ways
THE SITUATION Copper & Oak is adding a hearth-roasted half duck. Plate cost \$10.85. What should it cost the guest?
WHAT YOU KNOW Plate cost \$10.85 · house food-cost target 30% · the menu's weighted average contribution \$19.40 (Example 5) · comparable duck dishes in the market run \$33–\$38.
THE WORK
```text METHOD A — Food-cost percentage $10.85 / 0.30 = $36.1667 -> price at $36.00 Result: food cost $10.85 / $36.00 = 30.1% CM = $25.15
METHOD B — Contribution target (beat the menu average) $10.85 + $19.40 = $30.25 -> price at $31.00 Result: food cost $10.85 / $31.00 = 35.0% CM = $20.15
METHOD C — Competitive position Market range $33-$38; place mid-range at $34.00 Result: food cost $10.85 / $34.00 = 31.9% CM = $23.15
NOW THE PART THAT DECIDES IT — volume at each price At $36.00, the kitchen expects ~70 a period: 70 x $25.15 = $1,760.50 At $34.00, the kitchen expects ~120 a period: 120 x $23.15 = $2,778.00 Difference in contribution: +$1,017.50 ```
THE ANSWER \$34.00.** It banks **\$1,017.50 more per period than the price a 30% food-cost target would have produced, despite showing a worse food cost percentage.
THE DECISION Price at \$34 and watch the count for two periods. The three methods are not rivals; they answer different questions. Method A tells you the floor below which the dish cannot carry its share of overhead. Method C tells you the ceiling the market will bear. Method B tells you whether the dish is worth a slot on the menu at all — anything contributing below \$19.40 is taking a line from something better.
WHAT THE NUMBER DOESN'T TELL YOU The volume estimates are the weakest inputs in the calculation and they drive the whole answer. Mark them as low-confidence, and treat the first two periods as the actual experiment. Note also that a duck at \$34 changes what the menu says about the restaurant, which is Chapter 3's subject and does not appear anywhere in this arithmetic.
H.3 Labor
Example 8 — Building the labor line from a schedule
THE SITUATION The Rookery is a constructed 60-seat restaurant doing \$1,380,000. Its plan says "labor 34%." Nobody has checked that against a schedule. Build it from the bottom.
WHAT YOU KNOW The posted schedule, the pay rates, and a 20.5% burden (payroll taxes, workers' compensation, benefits, meals, uniforms, training).
THE WORK
Position Shifts/wk Hrs/shift Hrs/wk Rate Weekly wages Line cook A 5 9 45 \$20.00 | \$900.00 Line cook B 5 9 45 \$18.50 | \$832.50 Prep cook 5 8 40 \$17.00 | \$680.00 Dishwasher 6 7 42 \$15.00 | \$630.00 Servers (3) 15 6 90 \$9.00 | \$810.00 Bartender 5 8 40 \$16.00 | \$640.00 Host 5 5 25 \$16.00 | \$400.00 Busser 5 5 25 \$15.50 | \$387.50 Hourly total 352 \$5,280.00 ```text STEP 1 — Annualize the hourly wages. $5,280.00 x 52 = $274,560
STEP 2 — Add the salaried positions. Chef $62,000 General manager $58,000 Salaried subtotal $120,000
STEP 3 — Total wages. $274,560 + $120,000 = $394,560
STEP 4 — Burden. Wages are not labor cost. $394,560 x 0.205 = $80,884.80 -> $80,885
STEP 5 — Total labor, all in. $394,560 + $80,885 = $475,445
STEP 6 — NOW compute the percentage. It is an output. $475,445 / $1,380,000 = 34.452% -> 34.5% ```
THE ANSWER \$475,445 — 34.5% of sales. The plan said 34%. The schedule says 34.5%, and the difference is \$6,955 a year.
THE DECISION Publish the \$475,445 and the schedule that produced it, and change the plan's percentage to match — not the other way round. Then look at the two questions the schedule raises: the 352 hourly hours include no coverage for anybody's day off, and the servers' \$9.00 rate is a tipped rate that assumes a tip credit is lawfully available in this jurisdiction, which is Chapter 20's problem and must be confirmed before this schedule is a plan.
WHAT THE NUMBER DOESN'T TELL YOU A schedule is a plan for a normal week, and there are no normal weeks. It contains no overtime, no call-ins, no double when someone does not show, and no training overlap when someone new starts. Every one of those is real, and every one of them is additive. The honest way to use this figure is as a floor, not a forecast.
Example 9 — A fixed/variable split, tested where it matters
THE SITUATION The Rookery wants to know what labor costs if sales come in at \$1,100,000 instead of \$1,380,000. Two splits are on the table.
WHAT YOU KNOW Total labor \$475,445 at \$1,380,000 (Example 8). The honest split, built from the schedule: salaried \$120,000 burdened at 20.5% = \$144,600, plus an open-and-close hourly floor of 20 hours a week at \$16.50 = \$17,160, burdened = \$20,678. The casual split: "labor is 34.5%, of which about 30% is fixed."
THE WORK
```text THE HONEST SPLIT Fixed labor: $144,600 + $20,678 = $165,278 Variable: $475,445 - $165,278 = $310,167 Variable rate: $310,167 / $1,380,000 = 22.48%
THE CASUAL SPLIT Fixed: 0.30 x $475,445 = $142,634 Variable: $475,445 - $142,634 = $332,811 Variable rate: $332,811 / $1,380,000 = 24.12%
TEST 1 — at the volume they were built on. $1,380,000. Honest: $165,278 + 0.2248 x $1,380,000 = $165,278 + $310,224 = $475,502 Casual: $142,634 + 0.2412 x $1,380,000 = $142,634 + $332,856 = $475,490 Both reproduce $475,445 to within rounding. BOTH LOOK CORRECT.
TEST 2 — at the volume the question was actually about. $1,100,000. Honest: $165,278 + 0.2248 x $1,100,000 = $165,278 + $247,280 = $412,558 (37.5%) Casual: $142,634 + 0.2412 x $1,100,000 = $142,634 + $265,320 = $407,954 (37.1%) DIFFERENCE: $4,604 - and the casual split is the optimistic one. ```
THE ANSWER The two splits agree at \$1,380,000 and disagree by **\$4,604** at \$1,100,000 — with the casual split understating the cost.
THE DECISION Use the schedule-built split. And take the general lesson, which is worth more than this restaurant: a model validated only at its calibration point is untested. Both splits reproduce the known answer; only one of them survives the question you built it to answer. If you never test a split away from the volume you already know, you have not checked it — you have only confirmed that you can do arithmetic backwards.
WHAT THE NUMBER DOESN'T TELL YOU Neither split is linear in reality. At \$1,100,000 you would not simply run fewer hours at the same rates — you would cut a position, which moves a chunk of "variable" cost in one step and changes what the room can execute. A fixed/variable split is a planning approximation, and it degrades the further you get from the schedule that produced it.
Example 10 — What one turnover event costs, and what a year of them costs
THE SITUATION
[the Bellwether plan]A line cook gives notice. The chef says replacing a line cook "costs nothing, we always find someone." Price it.WHAT YOU KNOW Bellwether's frozen figures: \$2,180** to replace one line cook, and **\$38,070 of turnover cost across the restaurant in a year. Manager time is loaded at \$34.62 an hour; the burden is 20.5%.
THE WORK
```text ONE LINE COOK Recruiting, posting, screening ............................... $185 Manager and chef interview time (5 hrs combined) ............. $173 Paid stage, 5 hrs at $20.00 + 20.5% burden ................... $121 Training: 30 hrs of trainee wage at $19.00 x 1.205 ........... $687 Trainer overlap: 20 hrs at $22.00 x 1.205 .................... $530 Overtime and reduced productivity during the gap ............. $404 Uniform, food-handler card, onboarding ........................ $80 ────────────────────────────────────────────────────────────────── TOTAL PER LINE COOK ........................................ $2,180
THE RESTAURANT, FOR A YEAR Total turnover cost $38,070 As a share of revenue: $38,070 / $1,550,000 = 2.5% As a share of operating profit: $38,070 / $261,020 = 14.6% In dinner covers at $19.37 of contribution each: $38,070 / $19.37 = 1,965 covers a year 1,965 / 364 services = about 5.4 covers EVERY NIGHT ```
THE ANSWER \$2,180** for one line cook. **\$38,070 a year for the restaurant — 14.6% of planned operating profit, or 5.4 covers a night that exist only to pay for churn.
THE DECISION Take the number to the schedule, because that is where turnover is actually made. If \$38,070 buys 5.4 covers a night of nothing, then a retention measure costing \$15,000 — better scheduling notice, a shift-meal upgrade, a real training path — only has to cut turnover by 40% to pay for itself. That is not a soft argument. It is the same argument you would make about a piece of equipment.
WHAT THE NUMBER DOESN'T TELL YOU It does not price the guest experience during the six weeks a new cook is learning the station, which is real and is not in any of these lines. It does not price what happens to the rest of the crew when a good cook leaves. And it says nothing about why the cook left — which is the only input that would let you do something about it, and which is not a number.
Example 11 — A reclassification, traced all the way through
THE SITUATION
[the Bellwether plan]Bellwether's sous chef is salaried at \$48,000 and works about 55 hours a week. Review concludes the position does not satisfy the duties test and must be treated as non-exempt. Trace the effect all the way to covers per night.WHAT YOU KNOW Sous chef salary \$48,000 · about 15 overtime hours a week · Chapter 19's labor line **\$570,461 · total fixed cost \$437,635 · contribution margin ratio 40.53% · accounting break-even \$1,079,815 = 66 dinner covers a night**.
THE WORK
```text STEP 1 — The regular rate, and the overtime premium. $48,000 / 52 = $923.08 a week Regular rate: $923.08 / 40 = $23.077 an hour Overtime rate: $23.077 x 1.5 = $34.615
STEP 2 — The annual overtime bill. 15 hrs x $34.615 x 52 weeks = $27,000
STEP 3 — The labor line. $570,461 + $27,000 = $597,461 36.8% -> 38.5% of sales
STEP 4 — Prime cost. $430,280 + $597,461 = $1,027,741 = 66.3% of sales
STEP 5 — Fixed cost. The sous chef's hours are coverage, not volume, so the $27,000 lands in the FIXED bucket. $437,635 + $27,000 = $464,635
STEP 6 — Break-even, and this is the number that matters. $464,635 / 0.4053 = $1,146,435 Increase: $1,146,435 - $1,079,815 = $66,620
STEP 7 — Check it against the multiplier. Every $1 of new fixed cost raises break-even by 1 / CM ratio = $2.47 $27,000 x $2.47 = $66,690, against the $66,620 computed above. Agreement to within rounding.
STEP 8 — In covers. $1,146,435 / $1,079,815 = 1.0617 66 covers x 1.0617 = 70.1 -> 70 dinner covers a night ```
THE ANSWER \$27,000** of overtime. Labor **36.8% → 38.5%**, prime **66.3%**, break-even **\$1,079,815 → \$1,146,435, and the floor rises from 66 to 70 dinner covers a night.
THE DECISION Reclassify, and then decide what to do about the 55 hours — because reclassifying does not reduce them, it prices them. The choices are to hire coverage, to cut the hours, or to pay the \$27,000; and the third is a decision, not a default. What you may not do is leave the classification wrong, and it is worth being clear-eyed about why: the exposure is back wages, liquidated damages equal to the back wages, taxes, and penalties, which is a multiple of \$27,000.
WHAT THE NUMBER DOESN'T TELL YOU Compliance is not a constraint on the model. It is an input to the model. A plan built on a misclassification is not a plan with a legal footnote; it is a plan whose break-even is wrong by four covers a night. Note also that exemption thresholds and duties tests vary by state, several states are stricter than federal law, and the thresholds change — verify the specific position with an employment attorney in your jurisdiction.
H.4 The statement
Example 12 — Reading a full-year P&L
THE SITUATION Marrow is a constructed 64-seat restaurant, three years open. The year has closed. The owner says it "felt like a good year." Read the statement.
WHAT YOU KNOW Revenue \$1,180,000, split 73% food / 27% beverage. Food cost 31.0% of food sales, pour cost 23.0% of beverage sales.
THE WORK
```text BUILD IT Food sales 0.73 x $1,180,000 = $861,400 Beverage sales 0.27 x $1,180,000 = $318,600
Food COGS 0.310 x $861,400 = $267,034 Beverage COGS 0.230 x $318,600 = $73,278 TOTAL COGS $340,312 28.84%THE STATEMENT Revenue $1,180,000 100.00% Cost of goods sold $340,312 28.84% Labor, all-in $424,800 36.00% ────────────────────────────────────────────────────── PRIME COST $765,112 64.84% Occupancy $106,200 9.00% Other operating $165,200 14.00% General & administrative $35,400 3.00% ────────────────────────────────────────────────────── Total costs $1,071,912 90.84% OPERATING PROFIT $108,088 9.16%
FOOT IT. Always. $765,112 + $106,200 + $165,200 + $35,400 = $1,071,912 $1,180,000 - $1,071,912 = $108,088 ✓
CONTROLLABLE INCOME Revenue - COGS - labor - controllable other operating ($98,000) $1,180,000 - $340,312 - $424,800 - $98,000 = $316,888 26.86% ```
THE ANSWER Prime cost 64.84%. Operating profit \$108,088, or 9.16%.** Controllable income **\$316,888.
THE DECISION Two lines are out of band and they are not equally fixable. Prime at 64.84% is roughly five points above where a full-service casual restaurant should run — about \$59,000 a year — and it is controllable this month, in the schedule and on the cost cards. Occupancy at 9.0% is high, and it is not controllable at all until the lease renews; the only lever on it is the sales line underneath. So the sequence is: work prime now, and understand that the occupancy ratio improves by growing revenue, not by negotiating.
WHAT THE NUMBER DOESN'T TELL YOU Whether 9.16% was a good year depends on what was invested and what the owner was paid, neither of which appears here. It does not show the seasonality inside the year — twelve months at 9.16% and a year that lost money for four months and made it back are the same statement. And it says nothing about cash, which is Examples 13 and 20.
Example 13 — EBITDA is not cash, in both directions
THE SITUATION Marrow's owner reads \$108,088 of operating profit and proposes a \$90,000 distribution.
WHAT YOU KNOW Operating profit (EBITDA) \$108,088 · depreciation \$52,000 · interest \$18,900 · debt principal \$61,400.
THE WORK
```text THE TWO ADJUSTMENTS THAT MOVE IN OPPOSITE DIRECTIONS
Depreciation is an EXPENSE THAT CONSUMES NO CASH. Debt principal is CASH THAT IS NOT AN EXPENSE.NET INCOME (what the tax return will roughly show) EBITDA $108,088 Less interest -$18,900 Less depreciation -$52,000 ───────────────────────────────────── NET INCOME $37,188
CASH GENERATED (what the bank account will show) EBITDA $108,088 Less interest -$18,900 Less principal -$61,400 ───────────────────────────────────── CASH BEFORE DRAWS AND CAPEX $27,788
THE GAP, AND WHERE IT COMES FROM $37,188 - $27,788 = $9,400 Principal $61,400 - depreciation $52,000 = $9,400 ✓ ```
THE ANSWER EBITDA \$108,088**. Net income **\$37,188. Cash actually generated \$27,788**. A \$90,000 distribution is not merely aggressive — it is more than three times the cash the business produced.**
THE DECISION Distribute nothing this year, or distribute against the \$27,788 with a reserve held back. Then fix the reporting: put a cash line on the monthly statement, below operating profit, showing principal and capital spending. An owner who only ever sees EBITDA will keep proposing distributions the business cannot fund, and will be genuinely surprised each time.
WHAT THE NUMBER DOESN'T TELL YOU \$27,788 is cash before capital spending, and a three-year-old restaurant has a hood, a walk-in compressor, and a dish machine all moving toward replacement. It is also an annual figure, and the year does not arrive evenly — which is exactly how a business with positive annual cash runs out of money in February.
Example 14 — Five Fridays: the increase that is a calendar
THE SITUATION Marrow's March sales are \$104,900 against February's \$88,400. The manager reports "sales up 18.7%" and asks for a bonus. Check it.
WHAT YOU KNOW February: 4 Fridays, 4 Saturdays, 24 services. March: 5 Fridays, 5 Saturdays, 26 services. The restaurant averages \$6,100** per weekend service and **\$2,450 per weekday service.
THE WORK
```text WHAT THE CALENDAR ALONE PREDICTS February: 8 weekend x $6,100 = $48,800 16 weekday x $2,450 = $39,200 ─────────────────────────────── Calendar-explained $88,000 Reported $88,400 (+$400)
March: 10 weekend x $6,100 = $61,000 16 weekday x $2,450 = $39,200 ─────────────────────────────── Calendar-explained $100,200 Reported $104,900 (+$4,700)DECOMPOSE THE INCREASE Total increase: $104,900 - $88,400 = $16,500 Explained by two extra weekend services: 2 x $6,100 = $12,200 Genuine improvement: $4,700 - $400 = $4,300
THE HONEST COMPARISON — per service February: $88,400 / 24 = $3,683 March: $104,900 / 26 = $4,035 Change: +9.5% ```
THE ANSWER Not 18.7%. \$12,200 of the \$16,500 increase is the calendar. On a per-service basis March was up 9.5%, and the genuine improvement is \$4,300.
THE DECISION Pay the bonus — 9.5% per service is a real result and worth recognizing. But change the report: month-over-month revenue comparison is close to meaningless in a restaurant, because months contain different numbers of the services that carry the revenue. Report per-service and per-cover figures, and compare like periods. The same arithmetic runs the other way in a month with four weekends, and an operator who accepted the 18.7% will panic at a decline that is also a calendar.
WHAT THE NUMBER DOESN'T TELL YOU The \$6,100 and \$2,450 averages are themselves period averages, and a single 40-top or one closed Saturday moves them. This decomposition is a sanity check, not an attribution — it tells you how much of the change cannot be performance, which is a narrower and more defensible claim than telling you how much was.
H.5 Break-even and sensitivity
Example 15 — Break-even, in dollars and in covers
THE SITUATION
[the Bellwether plan]How much revenue does Bellwether have to do before it earns anything, and what does that look like on a Tuesday night?WHAT YOU KNOW ⚠️ This restaurant publishes two internally consistent cost structures, and they must not be mixed. Both carry the same \$437,635** of fixed cost. At the plan's \$500,000 labor line the contribution margin ratio is 45.07%; at Chapter 19's bottom-up \$570,461 line it is 40.53%. Chapter 32 does its break-even work on the second, because Chapter 31 settled that the bottom-up schedule is the honest one. Everything below uses 40.53%**.
THE WORK
```text FIXED COST — five lines, and they must foot Fixed labor (3 salaried $168,935 + open/close floor $22,960) .. $191,895 Occupancy ($28/sq ft base + $6/sq ft NNN on 2,800 sq ft) ....... $95,200 Other operating, genuinely fixed .............................. $60,950 Other operating, fixed base of the semi-variable lines ........ $43,090 General & administrative ...................................... $46,500 ────────────────────────────────────────────────────────────────────── TOTAL FIXED COST $437,635
ACCOUNTING BREAK-EVEN $437,635 / 0.40528645 = $1,079,815
Round-last note: dividing by the DISPLAYED 0.4053 gives $1,079,780, about $35 low. Carry full precision and round only what you print.IN COVERS $1,079,815 is 69.7% of the plan's $1,550,000 0.697 x 95 covers a night = 66 dinner covers a night
MARGIN OF SAFETY $1,550,000 - $1,079,815 = $470,185 $470,185 / $1,550,000 = 30.3%
CASH BREAK-EVEN — because break-even does not know debt exists ($437,635 + $69,500) / 0.40528645 = $1,251,298 80.7% of plan -> 77 dinner covers a night ```
THE ANSWER Accounting break-even \$1,079,815 — 66 dinner covers a night.** Cash break-even **\$1,251,298 — 77 covers. Margin of safety \$470,185, or 30.3%.
THE DECISION Put 77 on the wall, not 66. The accounting break-even is the point at which the statement stops showing a loss; the cash break-even is the point at which the bank account stops shrinking, and the second is the one that closes restaurants. Then note what 77 means operationally: the plan is 95 covers a night, so the entire cushion is 18 covers — about four tables.
WHAT THE NUMBER DOESN'T TELL YOU Break-even is an annual average, and no week is average. Chapter 9's first quarter runs at 66.6% prime cost, which puts Q1's break-even far above this figure — so a restaurant that is "above break-even for the year" can be below it for three months, which is precisely when it has the least cash. Neither figure funds capital replacement, an owner's draw, or a tax bill.
Example 16 — Sensitivity on three variables
THE SITUATION
[the Bellwether plan]What happens to Bellwether if revenue disappoints, food cost slips, and labor runs over — separately, and together?WHAT YOU KNOW Fixed cost \$437,635 · CM ratio 40.53% (variable 59.47%) · plan revenue \$1,550,000 · food sales are 72% of revenue. Baseline operating profit on this cost base is \$190,559, not the plan's \$261,020 — that figure belongs to the plan's \$500,000 labor line.
THE WORK
```text BASELINE $1,550,000 x 0.40528645 = $628,194 of contribution $628,194 - $437,635 = $190,559
(a) REVENUE DOWN 8% -> $1,426,000 $1,426,000 x 0.40528645 = $577,939 $577,939 - $437,635 = $140,304 change: -$50,255
(b) FOOD COST UP 2 POINTS (30% -> 32% on food sales) Food sales: 0.72 x $1,550,000 = $1,116,000 2% x $1,116,000 = $22,320 of extra COGS $190,559 - $22,320 = $168,239 change: -$22,320
(c) LABOR UP 2 POINTS OF SALES 2% x $1,550,000 = $31,000 $190,559 - $31,000 = $159,559 change: -$31,000
(d) ALL THREE AT ONCE — and this is the one that matters New variable rate: 59.47% + 1.44 (food, 2 pts on a 72% share) + 2.00 (labor) = 62.91% New CM ratio: 37.09% $1,426,000 x 0.3709 = $528,903 $528,903 - $437,635 = $91,268 change: -$99,291 ```
Scenario Operating profit Change Baseline \$190,559 — Revenue −8% \$140,304 | −\$50,255 Food cost +2 points \$168,239 | −\$22,320 Labor +2 points \$159,559 | −\$31,000 All three together \$91,268** | **−\$99,291 THE ANSWER Any one of the three is survivable. All three together cost \$99,291 — 52% of the operating profit — and the business is still profitable.
THE DECISION Two conclusions, and the second is the useful one. First, revenue is the most dangerous single variable: an 8% miss costs more than two full points of food cost. Second, the combined case is not the sum of the individual cases in any intuitive way — it is worse, because the revenue decline shrinks the base that the cost increases apply to and the cost increases shrink the margin on the revenue that remains. Model the combination. Downside cases that move one variable at a time are systematically optimistic.
WHAT THE NUMBER DOESN'T TELL YOU Nothing here says how likely any of it is, and a sensitivity table is often mistaken for a probability. It also holds the fixed costs fixed, which is the assumption most likely to fail: an 8% revenue miss that persists usually produces a schedule change, which moves \$437,635.
Example 17 — Cash break-even versus accounting break-even
THE SITUATION
[the Bellwether plan]Why is there more than one break-even, and which one should be on the office wall?WHAT YOU KNOW Fixed cost \$437,635 · CM ratio 40.53% · annual debt service \$69,500.
THE WORK
```text ACCOUNTING BREAK-EVEN — where the statement stops showing a loss $437,635 / 0.40528645 = $1,079,815 69.7% of plan 66 covers/night
CASH BREAK-EVEN — where the bank account stops shrinking ($437,635 + $69,500) / 0.40528645 = $1,251,298 80.7% of plan 77 covers/night
THE GAP $1,251,298 - $1,079,815 = $171,483
CHECK IT — the gap should be the debt service divided by the CM ratio. $69,500 / 0.40528645 = $171,483 ✓
OR, USING THE MULTIPLIER Every $1 of fixed obligation needs 1 / 0.4053 = $2.47 of sales $69,500 x $2.47 = $171,665, against $171,483 computed above. Agreement to within the rounding of the multiplier.
IN COVERS 77 - 66 = 11 more dinner covers EVERY NIGHT, all year, purely to service the debt that built the room. ```
THE ANSWER \$171,483 of additional annual revenue — 11 dinner covers a night — separates "not losing money" from "not running out of money."
THE DECISION The cash figure goes on the wall. The accounting figure goes in the plan, clearly labeled, because a reader will want both and will trust you more for showing the difference. And a third figure belongs beside them: at Chapter 20's lawfully reclassified labor line, cash break-even rises to \$1,317,918 — 81 covers a night. Carrying all three side by side is not indecision; it is the honest range.
WHAT THE NUMBER DOESN'T TELL YOU Even the cash figure omits capital replacement, owner draws, and the sales tax you are holding for the state. A restaurant sitting exactly at cash break-even is not stable — it is a restaurant with no capacity to absorb a compressor failure.
H.6 Cash
Example 18 — Building a short forecast and finding the trough
THE SITUATION Saltgrass is a constructed restaurant that opened eight weeks ago with \$46,000 in the operating account. Sales are ramping on plan. The owner wants to know whether the account is safe.
WHAT YOU KNOW Opening balance \$46,000 · food and beverage paid weekly at 28% of the prior week's sales, with \$16,000 of opening inventory in week 1 · payroll biweekly, disbursed in weeks 2, 4, 6, and 8 · rent \$7,933 in weeks 1 and 5 · other operating \$3,200 a week.
THE WORK
Wk Open Receipts Food/bev Payroll Rent Other Net Close 1 \$46,000 | \$14,200 \$16,000 | — | \$7,933 \$3,200 | −\$12,933 \$33,067 2 \$33,067 | \$21,600 \$3,976 | \$24,200 — \$3,200 | −\$9,776 \$23,291 3 \$23,291 | \$27,400 \$6,048 | — | — | \$3,200 +\$18,152 | **\$41,443** 4 \$41,443 | \$29,800 \$7,672 | \$22,800 — \$3,200 | −\$3,872 \$37,571 5 \$37,571 | \$31,200 \$8,344 | — | \$7,933 \$3,200 | +\$11,723 \$49,294 6 \$49,294 | \$32,600 \$8,736 | \$23,600 — \$3,200 | −\$2,936 \$46,358 7 \$46,358 | \$33,400 \$9,128 | — | — | \$3,200 +\$21,072 | **\$67,430** 8 \$67,430 | \$34,000 \$9,352 | \$24,200 — \$3,200 | −\$2,752 \$64,678 ```text READ THE SIGN OF THE "NET" COLUMN, NOT THE SALES COLUMN.
Weeks that FELL: 1, 2, 4, 6, 8 Weeks that ROSE: 3, 5, 7
Week 1 fell on opening inventory plus rent. Weeks 2, 4, 6 and 8 are EVERY payroll week, without exception. Sales rose every single week of the eight.
THE TROUGH Week 2, at $23,291 Drawdown from opening: $46,000 - $23,291 = $22,709 ```
THE ANSWER The trough is week 2, at \$23,291 — a \$22,709 drawdown — and it happens while sales are rising every week.
THE DECISION Nothing here needs fixing this month; the account recovers and ends at \$64,678. What needs fixing is the owner's mental model. The trough was not caused by a bad week. It was caused by the payroll calendar — a biweekly disbursement covering two weeks of a fully staffed floor, landing against a ramp that had only produced one week of revenue. Every falling week in the eight is a payroll week. Once you have seen that, you schedule the rent, the debt service, and the tax remittance so they do not stack on a payroll week, which costs nothing and moves the trough by thousands.
WHAT THE NUMBER DOESN'T TELL YOU Eight weeks is not long enough. This forecast never meets a slow February, a quarterly insurance premium, or the first sales-tax remittance. Thirteen weeks is the shortest forecast worth building, and the reason is that a quarter is the shortest period that contains a full cycle of the obligations that arrive on someone else's schedule.
Example 19 — Sizing working capital against the trough
THE SITUATION Saltgrass budgeted \$46,000 of working capital because it was what was left after the build. Was it enough, and how would you have known?
WHAT YOU KNOW Modeled trough drawdown **\$22,709** (Example 18) · rent \$7,933 a month · other operating \$3,200 a week · fixed labor about \$9,600 a month.
THE WORK
```text STEP 1 — The 30-day operating floor: what you owe if sales stop. Rent .................................... $7,933 Other operating ($3,200 x 52 / 12) ..... $13,867 Fixed labor ............................. $9,600 ───────────────────────────────────────────────── MONTHLY FLOOR $31,400
METHOD 1 — Trough plus one month of floor $22,709 + $31,400 = $54,109
METHOD 2 — The 60-day rule of thumb 2 x $31,400 = $62,800
CROSS-CHECK — do two independent methods agree? Difference: $62,800 - $54,109 = $8,691 $8,691 / $54,109 = 16.1% Two methods within 16% of each other. That is agreement.
TAKE THE HIGHER, AND ROUND UP. Working capital required: $63,000 Budgeted: $46,000 SHORTFALL: $17,000
IN DAYS OF RUNWAY $46,000 / $31,400 = 1.5 months $63,000 / $31,400 = 2.0 months ```
THE ANSWER \$63,000 was required against \$46,000 budgeted — a \$17,000 shortfall, and the difference between 1.5 and 2.0 months of runway.
THE DECISION Raise or reserve the \$17,000 now, while the business is healthy and a lender will still take the call. And note the general form of the argument: "\$46,000 of working capital" is an amount, not a plan. "\$46,000 against a modeled trough of \$22,709 plus a \$31,400 monthly floor" is a plan, because it can be checked and it can be wrong in a way you would notice.
WHAT THE NUMBER DOESN'T TELL YOU Both methods assume the trough you modeled is the worst one, and it is only the worst one in the eight weeks you looked at. Neither method contemplates an equipment failure, a health-department closure, or a landlord dispute. Working capital sized to the modeled trough is the minimum defensible answer, not a comfortable one.
Example 20 — Profitable, and out of cash
THE SITUATION Marrow's owner opens the monthly statement: operating profit +\$9,140. Then opens the bank statement: cash down \$12,144. The two numbers describe the same month. Explain the gap.
WHAT YOU KNOW Operating profit (EBITDA) \$9,140 · monthly debt principal \$5,117 · monthly interest \$1,575 · inventory built up \$6,200 ahead of a holiday · owner draw \$8,000 · sales tax collected \$9,440 during the month against \$9,832 remitted for the prior month.
THE WORK
```text THE BRIDGE — every line names one thing
Operating profit (EBITDA) +$9,140 Less interest — a real expense, below EBITDA -$1,575 Less debt PRINCIPAL — cash, but not an expense -$5,117 Less inventory build — cash converted to product sitting in the walk-in -$6,200 Less owner draw — cash, never an expense -$8,000 Less sales tax timing — remitted more than collected ($9,832 out, $9,440 in) -$392 ───────────────────────────────────────────────────────── NET CHANGE IN CASH -$12,144CHECK IT. The divergence must equal the sum of the named items. $9,140 - (-$12,144) = $21,284 $1,575 + $5,117 + $6,200 + $8,000 + $392 = $21,284 ✓ ```
THE ANSWER **A \$21,284 divergence, and not one dollar of it is mysterious.** Principal \$5,117, interest \$1,575, inventory \$6,200, owner draw \$8,000, tax timing \$392.
THE DECISION Nothing here is a crisis and nothing here is an error — the month was genuinely profitable and the account genuinely fell. The action is to put this bridge on the monthly statement permanently, as five standing lines below operating profit. An owner who sees it every month stops being surprised, and — more usefully — starts noticing when a line changes: an inventory build that does not unwind next month is a purchasing problem, and a draw that exceeds the cash generated two months running is a decision the business has not agreed to.
WHAT THE NUMBER DOESN'T TELL YOU The bridge explains the past month. It does not forecast the next one, and the items on it behave very differently: the principal is contractual and known, the tax timing is mechanical and self-correcting, the inventory build is discretionary, and the draw is a choice. Treating them as one lump — "cash was down twelve grand" — is how an owner concludes the business is failing when what actually happened is that they bought turkeys and paid themselves.
What all twenty have in common
Read back through them and one pattern recurs: in fourteen of the twenty, the arithmetic was not the hard part. The hard part was choosing which number to compute, and then knowing what it was silent about.
The costing examples turn on adjusting for what left the kitchen without being sold. The menu examples turn on measuring contribution rather than percentage. The labor examples turn on building from a schedule instead of asserting a rate. The statement examples turn on separating what a period earned from what it collected. The break-even examples turn on naming which cost base you are standing on. The cash examples turn on the calendar rather than the volume.
None of those is a formula. All of them are a decision about what question to ask — and that decision is made before the calculator comes out, which is why a book about restaurant management spends forty chapters on the operation and one appendix on the arithmetic.
A last practical note. Every calculation in this appendix foots, and you should check them. If one of them does not reconcile for you, the likeliest explanation is that you rounded somewhere in the middle. Carry the precision; round what you print.