Appendix B — The Recipe Cost Card Library

This appendix is the working file the rest of the book keeps pointing at. Eighteen fully costed recipe cost cards for Bellwether, the constructed 68-seat restaurant this book builds — six dinner entrées, four starters, three brunch items, and five bar and dessert items — plus the yield table they depend on, the discipline for keeping them current, and an honest account of everything a cost card cannot tell you.

Copy the format. That is the point of it. A reader who takes nothing else from this appendix but the blank template in B.2 and the habit of making every card foot will be ahead of most operators in their market, because most operators have cost cards for about a third of their menu, costed once, eighteen months ago, by somebody who no longer works there.

Every card here foots. The extended costs sum to the subtotal, the waste allowance computes off that subtotal, the plate cost is the sum of the two, the plate cost divided by the menu price produces the stated food cost percentage, and the menu price minus the plate cost produces the stated contribution margin. If you find a card in this appendix where that is not true, we have failed at the one thing this appendix exists to demonstrate.

B.1 How to read a cost card

A recipe cost card answers exactly one question: what does the food on this plate cost me to put there? It answers that question well. It answers no other question at all, which is the subject of B.9 and the reason this appendix does not end after the eighteenth card.

Here is the anatomy, field by field.

ANATOMY OF A RECIPE COST CARD                            [constructed teaching example]

  ┌─ HEADER ──────────────────────────────────────────────────────────────────┐
  │  Item name · menu section · station · yield · portion size · date costed  │
  └───────────────────────────────────────────────────────────────────────────┘
  ┌─ THE INGREDIENT BLOCK ────────────────────────────────────────────────────┐
  │  ITEM          the ingredient or sub-recipe, named the way you buy it     │
  │  UNIT          the unit the PRICE is expressed in (per lb, per fl oz,     │
  │                per each, per plate) — not the unit the recipe uses        │
  │  AP COST       as-purchased cost: what the invoice says, per that unit    │
  │  QUANTITY      how much this one plate consumes, in that same unit        │
  │  EXTENDED      AP cost x quantity. This column is the card.               │
  └───────────────────────────────────────────────────────────────────────────┘
  ┌─ THE FOOT ────────────────────────────────────────────────────────────────┐
  │  SUBTOTAL             sum of the extended column                          │
  │  + WASTE ALLOWANCE    2% of subtotal (see below)                          │
  │  = PLATE COST         the number you price from                           │
  │  MENU PRICE           what the guest pays, before tax                     │
  │  FOOD COST %          plate cost / menu price                             │
  │  CONTRIBUTION MARGIN  menu price - plate cost                             │
  └───────────────────────────────────────────────────────────────────────────┘

Yield and portion size

Yield is what the recipe produces: two plates from a bird, twelve portions from a cake, one quart of salsa verde. Portion size is what one guest receives. These two fields together are the whole reason a cost card can exist, because they convert a purchase unit into a plate.

The conversion is where almost every costing error lives. You buy chicken by the pound and serve it by the half-bird. You buy olive oil by the three-liter tin and use it by the tablespoon. You buy lemons by the case and use juice by the half-ounce. A card that does not state its portion size in the same unit the price is expressed in is not a cost card; it is a guess with a table around it.

As-purchased versus edible-portion cost

As-purchased (AP) cost is what you paid: the invoice number, per pound, per case, per each. It is a fact. You can look it up.

Edible-portion (EP) cost is what the usable part costs after you have thrown away the part you cannot serve — the silverskin, the carrot peel, the herb stems, the fish frame, the outer leaves. It is a derived number, and it is always higher than AP.

$$\text{EP cost per lb} = \frac{\text{AP cost per lb}}{\text{yield percentage}}$$

Whole beef tenderloin at \$18.48 a pound as purchased, trimmed at a 70% yield, is \$26.40 a pound of usable meat. Not \$18.48. If you cost a seven-ounce steak at the invoice price you will understate that plate by \$3.47, and you will do it on every tenderloin you sell, forever, while congratulating yourself on a 30% food cost you do not have.

Some cards in this library use AP cost directly because the item is used whole or nearly whole — the whole roasted fish, the six oysters, the eggs. Others carry an EP cost with the yield stated. Both are correct as long as the card says which one it is using and why. B.7 holds the yield reference.

The 2% waste allowance — the convention, and its lie

Every card in this appendix adds a flat 2% waste and spillage allowance to the ingredient subtotal. It covers what happens between the walk-in and the pass that is not anybody's fault: the plate dropped on the way out, the sauce that broke, the last four ounces in the container that went past date on a Tuesday, the portion the expo sent back because the garnish wilted under the heat lamp.

Two percent is a convention. It is defensible, it is common, and this book uses it consistently so that eighteen cards can be compared to each other on the same basis.

It is also a simplification that hides something real, and you should know exactly what. A flat percentage assumes waste is proportional to cost. It is not. Waste is proportional to handling, and handling varies enormously by ingredient:

• A case of root vegetables loses real weight to peel, trim, and the two that arrived soft. Your actual loss against the yield you planned might be four or five points on a bad delivery.
• A whole fish is either right or it is a disaster. You do not lose 2% of a branzino; you either serve it or you throw the whole \$9.79 away because someone left it on the hearth for two extra minutes.
• Fresh herbs are the worst offender on any menu. They arrive wet, they are picked by hand, the stems are half the weight, and if you do not use them within four days they are compost.
• Portioned dry goods — flour, sugar, oats, spices — genuinely lose close to nothing.

So a 2% flat allowance overstates waste on your dry goods and understates it badly on your proteins and delicate produce. It nets out acceptably across a menu. It nets out very poorly on a single card, and it is one of several reasons why your actual food cost at month-end will not match the sum of your cost cards. Chapter 13 teaches the ideal-versus-actual variance that measures the gap; this appendix simply tells you the gap is there and that the 2% convention is part of it.

The disciplined alternative, for an operator ready to do it: run a real yield test on your five most expensive ingredients, set an item-specific allowance for each, and keep 2% as the default for everything else. That is more work and more accurate. Do it after the menu has settled, not before.

The Q factor

A Q factor is the cost of everything the guest consumes that is not on any cost card: the condiments, the water-glass lemon, the coffee refill, the salt and pepper, the butter the table asks for twice, the sanitizer-clean silverware they use and you rewash. It is real food cost with no menu item attached to it.

Bellwether tracks Q factor per cover, not per plate, and it is not included in any card in this appendix. The planning figure is \$0.35 per cover.

🧮 Run the Numbers

The Q factor is worth more than a point of food cost.

Bellwether's plan runs about 95 dinner covers a night, five nights, plus roughly 110 covers each weekend brunch service:

• Dinner: 95 × 5 = 475 covers a week
• Brunch: 110 × 2 = 220 covers a week
• Total: 695 covers a week

At \$0.35 a cover: 695 × \$0.35 = \$243.25 a week**, or **\$12,649 a year.

Food sales on the plan are 72% of \$1,550,000, which is **\$1,116,000. So the Q factor alone is \$12,649 ÷ \$1,116,000 = 1.13% of food sales** — better than a full point of food cost that appears on no cost card anywhere in this appendix.

What you do with that: either add \$0.35 to every plate cost before you price (crude, but it works), or price from the cards and budget the Q factor as a separate line in your food cost target — which is what Bellwether does. What you must not do is forget it exists and then wonder for eleven months why the walk-in count never agrees with the cards.

Plate cost, menu price, food cost percentage, contribution margin

Plate cost is the subtotal plus the waste allowance. It is the only honest basis for a price.

Menu price comes from the plate cost and a target:

= plate_cost / target_food_cost_pct

The Hearth Chicken: \$8.52 ÷ 0.30 = \$28.40, which Bellwether prices at \$29.00. Note what just happened — the target produced a suggestion, and a human being made a decision. Chapter 11 is about the gap between those two things. The formula is where pricing starts, never where it ends.

Food cost percentage is plate cost ÷ menu price. It is the number everyone quotes.

Contribution margin is menu price − plate cost. It is the number that actually pays your rent.

You bank dollars, not percentages, and the arithmetic is brutal about this. The Hearth-Roasted Cauliflower runs a beautiful 22.0% food cost and contributes \$18.72. The Wood-Grilled Tenderloin runs an alarming 38.0% and contributes \$23.56. If you can only sell one more plate tonight, sell the alarming one. Chapter 12 turns this observation into a system.

B.2 The standard card format

Every card in B.3 through B.6 follows this template exactly. Copy it into a spreadsheet, one tab per item, and never break the column order — you will eventually want to compare fifty of them and the comparison only works if they are identical.

Item: ______ · Menu section: ______ · Station: ______ Yield: ___ portions · Portion size: ___ · Costed: ____ · Costed by: ____

Four conventions this library follows, which you should adopt or consciously reject:

1. The UNIT column describes the price, not the recipe. If you buy butter by the pound, the unit is "per lb" even though the recipe calls for an ounce and a half. The quantity column then reads 0.09375 lb. Ugly, and correct. A card where the unit and the quantity are in different systems is a card that will be extended wrong by whoever updates it next.
2. Compound sub-recipes get their own card and carry to the plate card as a single line — priced per fluid ounce or per portion. Salsa verde, romesco, dukkah, and the anchovy dressing in this library are all costed separately and appear here at their per-ounce cost. That is why they look suspiciously round. They are not round; they are the foot of another card.
3. "Oil, salt, and misc" is a real line, not a fudge. Every plate consumes finishing oil, seasoning, and the small stuff that is genuinely not worth weighing. Give it a number, keep the number small, and revisit it annually. A card with no misc line is understated on every plate.
4. Rounding is to the cent, at every step. Extended costs round to the cent, the subtotal is the sum of the rounded extensions, the allowance rounds to the cent, and the plate cost is the sum of those two. This produces the occasional one-cent artifact and it makes every card auditable by hand, which matters more.

⚠️ Where the Money Leaks

The card that was right in March.

The most common cost card failure is not arithmetic. It is age. A card costed at opening and never touched is a document that describes a restaurant that no longer exists — different supplier, different portion, a garnish somebody added in July, protein up eleven percent.

The specific damage is that a stale card makes your theoretical food cost wrong, and theoretical food cost is the benchmark you compare your actual count against (Chapter 13). If the cards say 29% and the count says 33%, you go hunting for four points of theft and over-portioning that may not exist — because two of those points are simply the cards being fourteen months old.

You will chase a ghost for a month. Recost quarterly. B.8 is how.

B.3 The cards — hearth and entrées

A necessary statement about these prices. Every ingredient cost in this appendix is a constructed illustrative figure representing plausible delivered pricing for a mid-size Midwestern market. They are not market quotes. They are not drawn from any distributor's price list, and no supplier is named anywhere in this book. They exist to make the arithmetic real and reproducible, and they will not match what you pay in your city this week — nor should you expect them to. Use the format; replace every number with your own invoices.

Dinner runs Tuesday through Saturday. Entrées are priced \$24 to \$38 against an average dinner check of \$46 and a blended food cost target of 30%.

Card 1 — Hearth Chicken ★ THE REFERENCE CARD

Item: Hearth Chicken · Menu section: Entrées · Station: Hearth Yield: 2 portions per bird · Portion size: half a 3.5 lb bird, bone-in Costed: current period · Costed by: the chef-owner

This is the book's costing lab. It appears in Chapters 10, 11, 12, 13, 14, and 24, and its figures are the constant against which every other lesson in this book measures a change. Do not adjust this card; when a cost moves, build a second card and compare it to this one. B.8 shows you how.

Menu-engineering classification: Star — high popularity, high contribution margin. Protect it, never discount it, and do not let anyone "improve" it onto a more expensive garnish.

The chicken is 67% of this card. That single fact governs everything: a ten percent move in poultry costs this plate fifty-six cents, and no amount of discipline on the salsa verde or the garnish will offset it. It is also why the roots and the herb salad are not worth arguing about — combined they are \$1.10, and a cook who spends four extra minutes trimming the parsnips more carefully has just spent labor to chase thirteen cents. Watch the line that is two-thirds of the card and manage the rest by exception.

Card 2 — Wood-Grilled Beef Tenderloin

Item: Wood-Grilled Beef Tenderloin · Menu section: Entrées · Station: Hearth Yield: 1 portion · Portion size: 7 oz, edible portion Costed: current period

Tenderloin is fabricated in house from whole PSMO at \$18.48 per pound as purchased. At a 70% planning yield after silverskin, chain, and side muscle, the edible-portion cost is \$18.48 ÷ 0.70 = **\$26.40 per pound**. That is the number on the card.

Thirty-eight percent food cost, and it is the most profitable plate on the menu in dollars. This card exists to break the reflex that low percentage equals good item. It also demonstrates how completely a yield assumption controls an expensive card: at 70% the protein line is \$11.55, at 65% it is \$12.44, and at 75% it is \$10.78. That is a \$1.66 swing on one assumption nobody in the building has verified. Run the yield test. Then run it again with a different butcher on the bench.

Card 3 — Hearth Pork Chop, apple and fennel

Item: Hearth Pork Chop · Menu section: Entrées · Station: Hearth Yield: 1 portion · Portion size: 12 oz bone-in chop, brined 12 hours

The chop is cut in house from whole bone-in loin at a 90% planning yield, which is why the protein line is affordable enough to support a \$22.32 contribution margin on a \$31 plate. The card also shows the butter line doing more work than usual — \$0.68, eight percent of the subtotal — because the chop is basted. That is a real cost and it is exactly the sort of thing an inherited cost card omits. If a cook does it every time, it belongs on the card.

Card 4 — Whole Hearth-Roasted Branzino

Item: Whole Hearth-Roasted Branzino · Menu section: Entrées · Station: Hearth Yield: 1 portion · Portion size: one whole fish, 1.1 lb, served on the bone

Eighty-four percent of this card is one fish, and the reason the food cost is only 35% is that serving the fish whole means the yield is effectively 100% — you paid for the frame and the head and you sold them. Fillet the same fish at a 47% planning yield and the usable meat costs \$8.90 ÷ 0.47 = **\$18.94 a pound; a 6-ounce fillet would carry \$7.10 of protein cost for less than half the plate presence. The whole-fish preparation is not a style choice on this menu. It is a costing decision that happens to plate beautifully.**

Card 5 — Hearth-Roasted Cauliflower, white bean, romesco

Item: Hearth-Roasted Cauliflower · Menu section: Entrées · Station: Hearth Yield: 1 portion · Portion size: half a large head, roasted whole

The best food cost percentage on the menu and the worst contribution margin on the menu, which is the whole lesson in one line. Note also that the cauliflower is roasted as a half-head, so the card uses AP cost — no trim loss to speak of. Break the same head into florets and you are at a 58% planning yield, an EP cost of \$2.93 a pound, and a materially different card. The vegetable entrée exists because a party of four needs one, and a table that can't seat is worth zero. Do not cut it because the contribution margin is last on the list; price it so it isn't embarrassing.

Card 6 — Bucatini, lamb shoulder ragù

Item: Bucatini with lamb shoulder ragù · Menu section: Entrées · Station: Sauté Yield: 1 portion · Portion size: 5 oz fresh pasta, ~4.5 oz braised lamb

Watch the lamb line: eight ounces raw yields about four and a half ounces braised. The card is costed on the raw weight, which is correct, and every plate therefore carries \$3.24 of a protein the guest experiences as four ounces. The single most common error on a braise card is costing the cooked weight at the raw price, which understates the protein by about 44% — here, by \$1.44 a plate. On 95 plates a week that error is \$7,113 a year of food cost you believe you are not spending.

B.4 The cards — starters and shared plates

Starters run \$11 to \$17. They matter more than their price suggests: they are the check-average lever, and a table that orders two starters instead of one moves per-person average by roughly seven dollars on a four-top.

Card 7 — Hearth Bread, cultured butter, honey

Item: Hearth Bread · Menu section: Starters · Station: Hearth / pastry Yield: 1 shared portion · Portion size: half loaf, 3 oz butter

The butter costs almost twice what the bread costs. That is true of essentially every bread service in America and it is why operators who give bread away and think they are giving away flour are wrong by a factor of two. Bellwether sells it, which converts a Q-factor cost into an \$8.36 contribution margin on roughly half the tables in the room. Selling bread is not stingy; it is the difference between a garnish line and a menu item. Chapter 23 argues the hospitality side of that decision, which is not a settled question.

Card 8 — Chicories Caesar, cured egg yolk

Item: Chicories Caesar · Menu section: Starters · Station: Garde manger Yield: 1 portion · Portion size: 6 oz trimmed leaves

Seven and a half ounces purchased to plate six. That 80% yield is a planning figure and it is the most fragile number on this card, because leafy-green yield is a function of the delivery, the season, and how patient the prep cook is feeling. A bad case at 65% pushes the lettuce line to \$2.08 and the plate cost to \$4.20 — 30.0% food cost instead of 27.0%, from nothing you did. Note the croutons: \$0.14, made from bread trim that would otherwise be waste. That is what cross-utilization looks like on a card, and it is worth more than it appears in B.9.

Card 9 — Half Dozen Oysters, mignonette

Item: Half Dozen Oysters · Menu section: Starters · Station: Raw / bar Yield: 1 portion · Portion size: 6 oysters, shucked to order

Ninety-three percent of this card is the oysters, and 38% is the highest food cost on the food menu — tied with the tenderloin. But look at the contribution margin: \$10.54, higher than the \$14 Caesar, higher than the \$13 carrots, higher than the bread. A 38% starter that a third of your tables order is a better business than a 24% starter nobody wants. The 2% waste allowance also lies harder here than anywhere else in this appendix: oyster loss is a count problem, not a percentage one. Two dead in a case of a hundred is real money, and the way you manage it is by counting the case on receipt (Chapter 13), not by adjusting a percentage on a card.

Card 10 — Hearth-Charred Carrots, whipped feta, dukkah

Item: Hearth-Charred Carrots · Menu section: Starters · Station: Hearth Yield: 1 portion · Portion size: 10 oz carrots AP, roasted

This is the card that proves cost cards do not price labor. Sixty-nine cents of carrots and a 23% food cost, and the dish is the most labor-intensive thing on the starter list: the carrots are scrubbed and blanched and charred in batches, the feta is whipped, the dukkah is toasted and ground in house, and the herb salad is picked leaf by leaf — the \$0.42 herb line represents maybe eight minutes of a prep cook's day per batch, which at a fully loaded \$20 an hour is \$2.67 of labor attached to forty-two cents of food. The card says this is your third-best percentage. The kitchen knows it is the reason prep runs long on Thursdays. Both are true, and only one of them is on the card.

B.5 The cards — brunch

Brunch is Saturday and Sunday against an average check of \$24 — roughly half the dinner check on a similar cover count. That arithmetic is unforgiving, and it means brunch lives or dies on plate cost and on beverage attachment, not on price. Items run \$13 to \$19.

Card 11 — Hearth Breakfast Sandwich

Item: Hearth Breakfast Sandwich · Menu section: Brunch · Station: Flat top / hearth Yield: 1 portion · Portion size: house muffin, 4 oz sausage, 2 eggs, potatoes

Note the two lines that round up — sausage at \$0.975 and cheddar at \$0.675 both become a penny more. That is the rounding convention from B.2 doing its job, and across a menu it costs you almost nothing while making every card checkable by hand. The card's real exposure is eggs: at \$0.31 each they are 16% of the subtotal, and egg pricing is the single most volatile line in this entire appendix. A doubling — which the market has delivered more than once in recent years — puts \$0.62 on this plate and takes it straight out of the contribution margin.

Card 12 — Sourdough Buttermilk Pancakes

Item: Sourdough Buttermilk Pancakes · Menu section: Brunch · Station: Flat top Yield: 1 portion · Portion size: 3 cakes, 2 fl oz maple

A brunch item that looks cheap and is. Twenty-two percent food cost, \$11.70 of contribution, and it cooks in six minutes on a flat top by anyone you can train in a shift. But look at what dominates: the maple syrup is \$1.34 — 41% of the whole card, more than four times the flour and sugar. Put a syrup pitcher on the table and let guests pour freely and you will run three or four ounces, not two, which is another \$0.67 to \$1.34 straight off the margin. Portion it in a two-ounce vessel, refill on request, and say yes cheerfully when they ask. That is a costing decision and a hospitality decision at the same time, and the card only sees half of it.

Card 13 — Hearth Chicken Hash

Item: Hearth Chicken Hash · Menu section: Brunch · Station: Flat top Yield: 1 portion · Portion size: 4 oz chicken, 2 eggs

The dark meat on this plate comes from the same birds bought for the Hearth Chicken. The dinner card uses the half-bird breast-and-wing side; brunch uses the legs and thighs. The card values that meat at the full AP price of \$3.20 per pound — it is not free just because you already own it — but the bird is now sold twice, which is what makes Bellwether's poultry order work.

The highest contribution margin at brunch, and the reason is cross-utilization: two menu items, two dayparts, one purchase order, one delivery, one place in the walk-in. The salsa verde is shared with the dinner card as well, which means one sub-recipe supports two plates and the batch size gets big enough to be efficient. Cross-utilization does not show up as a line on any cost card. It shows up as a smaller walk-in, fewer spoilage write-offs, and a produce order you can actually predict. Chapter 13 makes the case in inventory terms; the card just quietly benefits.

B.6 The cards — bar, cocktails, and dessert

Bellwether's beverage program runs a 22% blended pour cost against a sales mix of 72% food and 28% beverage. Individual drinks vary widely, and they are supposed to.

Note on tax: none of these prices includes the 7% sales tax. Sales tax is never in a menu price and never on a cost card. It is collected on top and remitted; it was never your money.

Card 14 — Rivermill Old Fashioned

Item: Rivermill Old Fashioned · Menu section: Cocktails · Station: Bar Yield: 1 drink · Portion size: 2 fl oz spirit, one large cube

The pour cost is exactly on target and the ice program is not. That \$0.26 cube is an allocation of the block ice you buy or cut, and it is the most understated line in this appendix. It does not include the freezer space the block occupies, the electricity to hold it, the twelve to fifteen minutes a day someone spends cutting and storing cubes, the cubes that shatter, or the fact that a cube costs the same whether the drink is \$14 or \$9. If you are cutting your own ice, cost the labor before you decide the program is worth it. Then decide it is worth it anyway, if the drink is your signature — but decide it, rather than discovering it.

Card 15 — Rivermill Spritz

Item: Rivermill Spritz · Menu section: Cocktails · Station: Bar Yield: 1 drink · Portion size: 5.25 fl oz built, over ice

Four points above the pour-cost target and worth every one of them, because the contribution margin is eighteen cents higher than the on-target Old Fashioned. What pushes the percentage is fresh juice: citrus yields roughly 35–45% of its purchase weight as juice, so a dollar of grapefruit buys you about forty cents of usable liquid, and any juice you squeeze on Thursday for a Saturday you have already thrown away. Squeeze to par, daily, and cost it as if you will lose some — because you will. Chapter 15's par-level work applies here more than anywhere else in the building.

Card 16 — House Red by the Glass

Item: House Red by the Glass · Menu section: Wine · Station: Bar Yield: 5 pours per 750 ml bottle · Portion size: 5 fl oz

This card's yield assumption is the one most likely to be false in your restaurant. Five five-ounce pours from a 25.4-ounce bottle leaves four-tenths of an ounce of slack — which is to say none. A bartender pouring by eye at 9:30 on a Saturday gets five and a half ounces, and five and a half ounces gives you four pours and a splash. Four pours means \$3.70 a glass, a 28.5% pour cost after allowance, and \$1.10 a glass of margin gone. The fix is not a lecture; it is a lined glass or a measured pour. Note too that the 2% waste allowance is genuinely too small here — bottles-by-the-glass lose meaningfully more to oxidation and the last-glass problem, and an operator with real data should carry 8–10% on this card specifically.

Card 17 — Hearth Apple Crisp, brown butter ice cream

Item: Hearth Apple Crisp · Menu section: Dessert · Station: Hearth / pastry Yield: 1 portion · Portion size: individual skillet, 2.5 fl oz ice cream

Apples are costed AP because they are peeled and cored to a 78% planning yield already reflected in the eight ounces this card buys to plate about six. The house-made ice cream is the interesting line: at \$0.38 a fluid ounce it beats a decent purchased pint on cost, but only if you actually churn it — and a pastry program that churns ice cream is a pastry program that needs someone in the building at seven in the morning. Every "we should make this in house" decision moves cost from the food half of prime cost to the labor half. The card sees one half.

Card 18 — Olive Oil Cake, macerated citrus

Item: Olive Oil Cake · Menu section: Dessert · Station: Pastry Yield: 12 portions per cake · Portion size: 1 slice Costing basis: batch of 12, divided to the portion

A batch card, and the batch is the risk. The arithmetic above assumes you sell all twelve slices. Sell nine and throw away three, and the cost per slice sold is not \$2.74 but \$3.65 — a 41.4% food cost after allowance instead of 31.0%. This is the single most common invisible loss in a dessert program, and no cost card in the world will surface it. What surfaces it is a waste log and a production sheet built off yesterday's covers (Chapter 13). Bake to the forecast, not to the recipe.

🧾 Read the Numbers

```text FIGURE B.1 — "Does the entrée menu actually blend to 30%?" [constructed teaching example] THE ARTIFACT Weekly menu-mix and cost extension across the six dinner entrée cards in B.3, at planned unit volumes for a five-night dinner week. THE CONTEXT Bellwether, 475 dinner covers a week, roughly one entrée per cover.

ITEM               PRICE    PLATE    UNITS    REVENUE     COST      CM/UNIT
Hearth Chicken     $29.00   $ 8.52     150   $ 4,350.00  $1,278.00   $20.48
Bucatini           $26.00   $ 6.76      95   $ 2,470.00  $  642.20   $19.24
Pork Chop          $31.00   $ 8.68      80   $ 2,480.00  $  694.40   $22.32
Branzino           $34.00   $11.90      55   $ 1,870.00  $  654.50   $22.10
Cauliflower        $24.00   $ 5.28      50   $ 1,200.00  $  264.00   $18.72
Tenderloin         $38.00   $14.44      45   $ 1,710.00  $  649.80   $23.56
───────────────────────────────────────────────────────────────────────────
TOTALS                                 475   $14,080.00  $4,182.90

Blended food cost   $4,182.90 / $14,080.00        = 29.7%
Total contribution  $14,080.00 - $4,182.90        = $9,897.10
CM per entrée sold  $9,897.10 / 475              = $20.84

WHAT IT SHOWS Six cards ranging from 22.0% to 38.0% blend to 29.7% — three-tenths of a point under the 30% target — because the mix does the work, not any one card. The Hearth Chicken is 31.6% of units and 31.0% of total contribution margin: it is carrying the menu by volume, not by percentage. WHAT IT DOESN'T It excludes starters, brunch, dessert, the beverage program, and the Q factor. It assumes exactly one entrée per cover and assumes the planned mix holds — which it will not in January, and will not the week the tenderloin runs out at 8:15 on a Saturday. THE DECISION Do not "fix" the tenderloin's 38%. Fix the mix: if tenderloin units rise to 90 and cauliflower falls to 5, blended food cost climbs and you re-cost or reprice that one item. Watch the mix weekly. THE LESSON A menu where every item hits 30% is a menu nobody costed. You manage the blend; you do not manage each card to a target. ```

CONTRIBUTION MARGIN, DINNER ENTREES — dollars, not percentages   [constructed teaching example]

  Tenderloin      $23.56  ███████████████████████    45/wk  =  $1,060
  Pork Chop       $22.32  ██████████████████████     80/wk  =  $1,786
  Branzino        $22.10  ██████████████████████     55/wk  =  $1,216
  Hearth Chicken  $20.48  ████████████████████      150/wk  =  $3,072
  Bucatini        $19.24  ███████████████████        95/wk  =  $1,828
  Cauliflower     $18.72  ██████████████████         50/wk  =    $936
  ──────────────────────────────────────────────────────────────────
  Weekly entree contribution margin, all six                =  $9,897
  (weekly dollar figures rounded to the nearest dollar)

  Read the two columns against each other. The item with the HIGHEST margin per
  plate contributes the LEAST in dollars, because it sells 45 times. The item in
  fourth place on margin contributes nearly a third of the total. That gap is
  what menu engineering is for.

B.7 Yield and conversion notes

These are the as-purchased-to-edible-portion yields the eighteen cards depend on. Every one is a planning yield to be verified by your own yield test — a starting assumption for building a card before you have data, not a guarantee, not a standard, and emphatically not a number to defend to your chef when the walk-in says otherwise.

Yield varies by season, by grade, by supplier, by how the case traveled, by who is holding the knife, and by how fast that person is being asked to work. Two prep cooks butchering the same case of tenderloin will return yields four or five points apart, and neither of them is doing anything wrong.

👨‍🍳 On the Line

How to run a yield test, in about twenty minutes.

You need a scale, a piece of paper, and one honest hour on a slow Tuesday.

1. Weigh the case as it arrives, before anyone touches it. Write it down. Write the invoice price down next to it. This is your AP weight and AP cost.
2. Have the person who normally does this job do it, at normal speed. Not the chef, not you. The yield you want is the yield you actually get on a Thursday, not the yield a careful person gets when they know they are being measured. Tell them what you are doing and why.
3. Weigh the usable product. That is EP weight.
4. Weigh the trim separately and decide honestly whether it is used. Chicken frames that become stock are recovery. Chicken frames that go in the bin are loss. Do not credit yourself for a stock you do not make.
5. Yield % = EP weight ÷ AP weight. EP cost per pound = AP cost per pound ÷ yield %.
6. Do it three times, on three deliveries, and average them. One test is an anecdote.

Do this for your five most expensive ingredients before you do it for anything else. On a menu like Bellwether's that is tenderloin, branzino, oysters, lamb shoulder, and chicken — which together are most of the food cost in the building.

B.8 Recosting: how and when

The cards in this appendix were costed once. Yours have to be costed forever, and the discipline is simple enough to state in a sentence: recost the whole menu quarterly, recost individual cards whenever a price moves more than 10%, and recost every card that touches an item whenever you change a supplier or a spec.

A reminder that belongs in this section as much as in B.3: the ingredient prices throughout this appendix are constructed illustrative figures for a mid-size Midwestern market. They are not quotes, they are not sourced from any distributor, and they will not be right for you. The method in this section is the transferable part.

The quarterly cycle

Put four dates in the calendar — one in each quarter, on a Monday, in the morning, before service thinking starts. Block ninety minutes. Pull the last four weeks of invoices and update the AP cost on every line in the library. Do not re-derive the recipes; only the prices move.

Ninety minutes for a fifty-item menu is realistic once the spreadsheet exists, because you are updating an ingredient price list, not fifty cards. Build the library so that every card pulls its prices from one master ingredient sheet. Change chicken once and eleven cards update. If your cards each carry their own hard-typed prices, quarterly recosting takes a day, which means it will not happen, which is how cards go stale.

Which lines move and which do not

The practical consequence: most of your quarterly recosting effort belongs on ten or fifteen ingredients. The chicken, the tenderloin, the branzino, the oysters, the lamb, the eggs, the butter, the cream, the oil, and whatever produce is in the middle of your menu right now. The rest of the sheet you update because you are already there.

Absorb or reprice

Not every increase is a price change. The question is not "did the cost go up?" — the question is "is this move structural or seasonal, and how much contribution margin does it actually take?"

THE RECOSTING DECISION                                  [constructed teaching example]

   A cost line moves.
        |
        +--> Is the move under 5% of the PLATE cost?
        |         YES --> ABSORB. Note it. Watch it next quarter.
        |
        +--> Is it seasonal or weather-driven, and will it revert?
        |         YES --> ABSORB, or swap the ingredient. Do not reprint a menu
        |                 for a spring produce spike.
        |
        +--> Is it structural: a contract renewal, a permanent supplier change,
        |    a commodity shift that has held for two quarters?
        |         YES --> Consider REPRICE, RE-PORTION, or RE-SPEC.
        |
        +--> Does the item still clear its contribution margin target
             even at the higher cost?
                  YES --> You may still hold price. CM is what you bank.
                  NO  --> Reprice, or take the item off the menu.

🧮 Run the Numbers

Two increases on the Hearth Chicken, and two different answers.

The reference card: subtotal \$8.35, allowance \$0.17, plate cost \$8.52, price \$29.00, food cost 29.4%, contribution margin \$20.48.

Case one — the salsa verde gets expensive. A wet spring pushes herb cost up and the salsa verde sub-recipe goes from \$0.42 to \$0.50 a fluid ounce, a 19% jump.

• Salsa verde line: 2.5 fl oz × \$0.50 = **\$1.25** (was \$1.05)
• New subtotal: \$8.35 + \$0.20 = \$8.55
• Allowance at 2%: \$8.55 × 0.02 = **\$0.17**
• New plate cost: \$8.72
• At \$29.00: \$8.72 ÷ \$29.00 = **30.1%** food cost; contribution margin **\$20.28**

You lost twenty cents a plate. Absorb it. Herb prices revert, the item is still a Star, and reprinting a menu to recover \$0.20 on 150 plates a week is not worth the printing.

Case two — the poultry contract renews 10% up. Chicken goes from \$3.20 to \$3.52 a pound.

• Chicken line: 1.75 lb × \$3.52 = **\$6.16** (was \$5.60)
• New subtotal: \$8.35 + \$0.56 = \$8.91
• Allowance at 2%: \$8.91 × 0.02 = **\$0.18**
• New plate cost: \$9.09
• At \$29.00: \$9.09 ÷ \$29.00 = **31.3%** food cost; contribution margin **\$19.91**

That is \$0.57 a plate. At 150 plates a week: \$85.50 a week, \$4,446 a year of contribution margin gone from one item on one contract renewal.

This one is structural, so you decide deliberately. To restore a 29.4% food cost: \$9.09 ÷ 0.294 = \$30.92, which prices at \$31.00 — food cost 29.3%, contribution margin \$21.91.

But note the honest alternative: at \$29.00 the item still contributes \$19.91, which is more than the bucatini and the cauliflower contribute at their original costs. Holding the price is a defensible answer. What is not defensible is holding the price without knowing you did.

And note what you must never do: change the reference card. Build the new card, keep the old one, and let the pair of them tell you what a contract renewal costs.

The two things quarterly recosting is not

It is not a menu price review. Price is a market decision, a positioning decision, and a guest decision; cost is only one of its inputs. Chapter 11 separates the two properly. You can recost four times a year and change prices once.

It is not a substitute for counting. Cards give you theoretical cost. Only inventory gives you actual cost. If your cards say 29.7% and your count says 33.5%, the cards are not wrong — the kitchen is telling you something, and the gap is the lesson. Chapter 13 owns that variance and it is the most valuable number in this entire discipline.

B.9 What a cost card cannot tell you

A cost card prices an ingredient list. That is all it does, and it does it precisely enough that operators forget how narrow it is. Here is what these eighteen cards do not know, stated operationally.

They do not price the labor to execute the dish. Card 10 — the charred carrots — costs \$2.99 and runs a 23.0% food cost, which is the third-best number in this appendix. It is also the most labor-expensive plate on the starter list. Blanching in batches, whipping the feta, toasting and grinding the dukkah, and picking the herb salad leaf by leaf is roughly eight minutes of prep per batch on the herbs alone; at a fully loaded \$20 an hour that is \$2.67 of labor attached to \$0.42 of herbs. Prime cost is food plus labor, and every "in house" decision on these cards is a transfer from one to the other. House pasta, house ice cream, house English muffins, house butter, house dukkah, hand-cut ice — Bellwether's menu is a series of decisions to buy a low food cost with labor, and only the P&L sees the whole trade.

They do not price the equipment time the dish occupies. The hearth has a fixed number of usable positions and a fixed recovery rate. A whole branzino holds a position for about fourteen minutes; a half chicken holds one considerably longer. On a Saturday at 7:40, the constraint is not cost, it is hearth minutes — and a menu whose two most popular items both live on the hearth will produce ticket times that no cost card predicted and a dining room that turns slower than the plan assumed. A seat you cannot turn is revenue you cannot store.

They do not price the fuel. Bellwether burns hardwood. At roughly \$210 a week that is \$10,920 a year, about 0.7% of planned sales — and it lands in other operating expense, not in COGS, so it appears on no cost card in this appendix. An operator who prices exclusively from the cards under-recovers it on every hearth item. It is not an error to keep it out of COGS; it is an error to forget it exists.

They do not price the skill the dish demands, or what happens when that skill is off. Every card here assumes the dish is executed correctly the first time by someone who has made it two hundred times. Three specific costs arrive when that is not true:

• Refires. Assume a 2% refire rate on the Hearth Chicken — three plates a week remade. Three × \$8.52 = **\$25.56 a week, \$1,329 a year** in food alone, plus the ticket that now runs twenty-five minutes and the table that now needs recovering.
• Over-portioning. A cook who runs half a fluid ounce heavy on the salsa verde adds 0.5 × \$0.42 = \$0.21 a plate**. On 150 plates that is \$31.50 a week, \$1,638 a year**, from a hand that is being generous.
• Yield drift. A new prep cook takes the root-vegetable yield from 80% to 70%. To plate the same 8 ounces you now buy 11.43 ounces AP, and the root line moves from \$0.95 to **\$1.09 — fourteen cents a plate, \$21 a week, **\$1,092 a year, from nothing anybody did wrong on purpose.

Those three lines together are just under \$4,100 a year on one menu item, none of which is visible on the card, all of which is visible in your ideal-versus-actual variance. That is what variance is.

They do not price the cook who makes it well being off. This is the operational version of the point above and it deserves its own line. Every restaurant has a dish that one person makes correctly and everyone else makes approximately. When that person is on vacation, the cost of the dish rises — in trim, in refires, in comps, in the table that does not come back — and it rises silently, because the cost card did not change. The countermeasure is not a better card. It is a standardized recipe with a scaled photograph of the plate and a written spec (Chapter 10), executed by a station that has been cross-trained (Chapter 21). A cost card is only accurate to the extent that the recipe is actually followed.

They do not know your comp and void rate. Every comped plate is a plate you bought and did not sell. Card 1 says \$8.52; a comped Hearth Chicken costs \$8.52 and returns \$0.00, and the card has no field for that.

They do not know spoilage beyond the 2% convention. Card 18 — the olive oil cake — shows this plainly: bake twelve, sell nine, and the cost per slice sold goes from \$2.74 to \$3.65, a jump from 31.0% to 41.4% food cost after allowance. Nothing on the card moved.

They do not know what your guest will pay. The card produces a suggested price from a target percentage. It knows nothing about your neighborhood, your competitive set, what a \$38 entrée signals on a menu whose other five items top out at \$34, or the fact that guests read prices in relation to each other rather than absolutely. Chapter 11 handles pricing; Chapter 12 handles what happens after the guests vote.

They do not know what the dish does to the other dishes. The Hearth Chicken Hash on Card 13 exists because the Hearth Chicken on Card 1 leaves legs and thighs behind. The croutons on Card 8 exist because the bread on Card 7 leaves ends. Neither of those relationships appears in a single extended cost. Cross-utilization shows up as a smaller walk-in, fewer write-offs, a produce order you can predict, and a delivery you can actually receive properly — real money, none of it on a card.

And they do not tell you whether the dish should exist. That question is answered by units sold and contribution margin, and it is a menu-engineering question rather than a costing one. A card can tell you a dish is expensive. Only the mix can tell you it is a Dog.

⚠️ Where the Money Leaks

The confidence a cost card creates is the most expensive thing about it.

An operator with a full library of costed cards, all footing, all current, will tell you their food cost is 29.7% — because the cards say so, and the cards are right about what they measure.

Then the count comes in at 33.5%.

Nothing in this appendix is wrong. The gap is the Q factor, the refires, the over-portioning, the yield drift, the comps, the batch you baked and threw away, the case that came in short and nobody checked, and the two percent flat allowance that was never going to be right on a whole fish.

Cost cards do not measure your food cost. They give you the number to compare it against. That is a smaller claim than most operators make for them, and it is worth far more, because a theoretical you trust is the only way a variance means anything at all.

Where to go next. Chapter 10 builds the standardized recipes these cards depend on. Chapter 11 turns a plate cost into a price. Chapter 12 puts the results on the menu-engineering matrix. Chapter 13 counts the walk-in and tells you whether any of this was true. And Appendix C is the blank Business Plan Workbook, where your own costed menu belongs.

Start with the five most expensive ingredients on your menu. Weigh a case. Write down the number. Everything in this appendix follows from that one act of counting.