Appendix G — Purchasing and Inventory Reference

"The walk-in is where you find out whether the rest of it was true." — constructed; the thing you say to yourself at eleven at night with a clipboard

This is the least glamorous appendix in the book. It is also the one you will open most often — at the receiving door on a Tuesday morning, in the walk-in on the last night of the period, and on the phone with a broadliner who wants to know why you are asking about a fuel surcharge.

Everything here supports Chapter 11 (costing and pricing) and Chapter 13 (purchasing, receiving, storage, and inventory). Nothing here replaces those chapters. What it does is put the tables, the forms, and the arithmetic in one place so you do not have to hunt through the book to find the yield on a whole chicken while a delivery driver is standing in your doorway with a hand truck.

G.1 How to use this appendix

Use it three ways.

At the door. G.4 (receiving) and G.6 (shelf life) are the two sections a manager should find in ten seconds. Print them and put them where the receiving happens, not in the office.

At the desk. G.7 (yield), G.8 (conversions), and G.9 (par levels) are what you use when you are building a cost card, setting an order guide, or trying to understand why your purchases and your usage disagree.

As blank forms. G.14 collects every form here as a copyable table. Put them in a spreadsheet and change them until they fit your restaurant. A form you inherited and never adjusted is a form nobody fills out honestly.

The standing caveat

State this once, then remember it every time you use a number from this appendix:

Every yield, shelf life, and par level in this appendix is a planning figure, not a guarantee.

Yields vary by supplier, by grade, by season, by the size of the individual item, and — enormously — by the person holding the knife. A cook who has broken down two thousand chickens will get five points more meat off a bird than a cook who has broken down twenty, and neither of them is wrong.

Shelf lives vary by how the product was handled before it reached you, by how fast it moved through the distribution chain, by how many times your walk-in door opened during Saturday service, and by whether your walk-in actually holds the temperature its thermostat claims.

Par levels vary by everything.

Verify with your own yield tests, on your own product, cut by your own cooks. The numbers here are for the first draft of a cost card and for sanity-checking a figure that looks wrong. They are not evidence, and you must never put one of them in front of a lender, a partner, or an accountant as though it were measured in your building.

Every dollar figure attached to an ingredient in this appendix is a constructed illustrative figure for a mid-size Midwestern market, consistent with the constructed restaurant this book builds ([the Bellwether plan]). Prices move constantly, by region, by season, and by contract. Yours will be different. Use the arithmetic, not the price.

Food-safety controls — temperature holding, date marking, cooling, cross-contamination, sanitizer concentration, and the inspection checklist — live in Appendix E. This appendix references them where storage and shelf life touch them, and does not re-teach them. When the two disagree, Appendix E wins, and behind Appendix E is your local health code, which wins over everything.

G.2 The purchase specification

What a spec is, and why an unspecified order is a price you cannot compare

A purchase specification — a spec — is a written description of exactly what you want, precise enough that two different suppliers quoting against it are quoting the same thing.

That last clause is the whole point. Most operators think a spec is about quality. It is partly about quality; it is mostly about comparability. Call three vendors and ask "what's your price on chicken" and you will get three numbers that mean nothing, because one is quoting a 2.5-pound water-chilled bird, one a 3.5-pound air-chilled bird, and one boneless thighs. You have not run a bid. You have collected trivia.

A spec turns purchasing from a relationship into a market. It also does four other things:

• It makes receiving possible. You cannot reject a delivery for being wrong if you never said what right was. "This isn't what I wanted" is an argument. "This is not what the spec says, here is the spec" is a credit.
• It protects your cost card. A cost card is built on a yield, and a yield is built on a specific product. Substitute a different size, grade, or pack and the cost card silently becomes fiction.
• It makes your kitchen repeatable. The dish tastes the same in March as it did in October because the product is the same product.
• It survives you. When you are not there, the spec is there.

The anatomy of a good spec

Eight fields. Each one exists because somebody once lost money for the lack of it.

The blank spec template

PURCHASE SPECIFICATION                                       Rev. ____  Date ____

That second-to-last line — the yield and the date of the last yield test — is the one nobody includes and everybody needs. It ties the spec to the cost card. When the spec changes, you know immediately which cost cards became wrong.

Six filled examples

All prices are constructed illustrative figures for a mid-size Midwestern market. [the Bellwether plan]

1. Protein — the Hearth Chicken bird

2. Produce — romaine

3. Dairy — butter

4. Dry good — flour

5. Paper good — the to-go container

6. Bar item — the well gin

⚖️ Code and Compliance

Alcohol purchasing is not like food purchasing, anywhere.

Depending on your state, you may be required to buy through a licensed wholesaler, prohibited from buying at retail, prohibited from accepting credit terms from a distributor, required to pay on a statutory schedule, or buying from a state-operated warehouse at a posted price. Some states forbid the transfer of alcohol between two locations you own. Some require you to log every delivery.

None of this is optional and all of it varies. Verify with your state alcohol authority and your attorney before you set up a beverage purchasing program, and re-verify when you open a second location or cross a county line. The penalties for getting this wrong are attached to your liquor license, which is to say to the value of your business.

G.3 The vendor relationship

The four channels, and what each is good at

Most independent full-service restaurants end up with a broadliner as the backbone, one or two specialty houses, and a cash-and-carry account for the Friday afternoon when something is missing. That mix is not a failure of discipline. It is the correct answer.

Bid sheets, and how to run one without becoming a full-time buyer

A bid sheet is a list of specified items sent to two or more vendors, who quote against it for a stated period. It is how you find out what a fair price is.

The trap is that bidding is a job. If you bid everything every week, you will spend eight hours a week buying and still get the same delivery. Five rules keep it proportionate:

1. Bid the top of your spend, not the top of your item count. Sort purchases by annual dollar volume; the top 20 items typically carry most of your food spend. The 4-ounce squeeze bottle of specialty vinegar does not get bid.
2. Bid on a cycle. Quarterly for most categories. Weekly for produce, because produce is genuinely a market and last quarter's price is meaningless.
3. Bid the spec, not the item. Require the vendor to quote against it or state their deviation. A quote that does not reference the spec is not a quote.
4. Bid the delivered cost, not the line cost. Include minimums, drop charges, fuel surcharges, and delivery days — G.14 has a bid comparison form built for exactly this.
5. Tell the incumbent you are bidding. Not a hostile act. It is how the relationship stays honest, and a good vendor expects it.

Price protection, market pricing, and what "market" means on an invoice

Three pricing bases, three very different exposures:

• Fixed / contract price. The vendor holds a price for a stated period. You can cost against it with confidence, and you pay for that certainty: a contract price usually sits a little above the average market price over the period, because the vendor carries the risk. Read the escape clause; most contracts have one for extraordinary market moves, and "extraordinary" is defined by the vendor.
• Price protection. A softer version: the vendor holds your price for a stated number of days after a quote, or caps the increase in a period. Useful for planning a menu print run.
• Market price. The price moves with the vendor's cost. On an invoice, "market" means the price the vendor paid plus their margin, on the day they bought it — not a commodity-exchange price and not anything you can look up. Produce, seafood, and dairy are the usual market categories.

The practical consequence: you cannot cost a market item once and walk away. Either re-cost on a cycle, build a costing allowance into the plate price, or accept a menu item whose margin breathes with the season. The third option is fine as long as you chose it. A market item quietly destroying a cost card is exactly the cost drift Chapter 1 describes.

Minimums, delivery windows, drop charges, and the true cost of a small order

Every one of these is a real cost, and none of them appear on your food cost line where you would notice them.

• Minimum order. The figure below which the vendor will not deliver, or will deliver with a fee.
• Delivery window. The days and the hours — more important than most operators realize, because it sets your par levels (G.9). A vendor delivering Monday, Wednesday, and Friday lets you carry half the inventory of a vendor delivering Tuesday only.
• Drop charge / small-order fee. A flat fee for an order under the minimum.
• Fuel surcharge. A per-delivery or percentage charge that moves with diesel. Negotiable more often than vendors admit, and almost never audited by the restaurant.

🧮 Run the Numbers

What a small order actually costs. (constructed illustrative figures)

You are short on produce Thursday. You place a \$180 order. Your vendor's minimum is \$350, so a \$35 small-order fee applies.

\$35 ÷ \$180 = 19.4%. You just paid a nineteen percent surcharge on that produce.

Now run it forward. Suppose that happens once a week for a year: \$35 × 52 = **\$1,820**.

Put that in the terms this book keeps returning to. On the Hearth Chicken cost card [the Bellwether plan] — plate cost \$8.52, menu price \$29.00, contribution margin \$20.48 — \$1,820 ÷ \$20.48 = 89 additional chickens you must sell just to get back to even. That is close to a full extra Friday night of your best-selling entrée, spent on delivery fees nobody read.

What the disciplined operator does: build the order guide so the Thursday order is on the Wednesday truck. This is a par-level problem (G.9), not a purchasing problem, and it is solved with arithmetic rather than with a phone call.

Rebates, and why the cheapest line item is not always the cheapest program

Broadliners and buying groups offer rebates — a percentage of qualifying purchases returned quarterly or annually, sometimes as cash and sometimes as credit. Buying groups aggregate independent restaurants to negotiate as a larger account.

Rebates make line-item comparison misleading in both directions, and you have to do the arithmetic to know which.

🧮 Run the Numbers

When the rebate wins. (constructed illustrative figures)

Annual food and paper purchases: \$180,000.

You bid 20 high-volume items against a competitor. The broadliner's prices on those 20 items come back 3% higher. Those 20 items are 15% of your spend.

• Extra cost on the bid basket: 15% × \$180,000 = \$27,000 of spend; 3% of \$27,000 = **\$810/year**.
• The broadliner's program returns a 2% rebate on qualifying purchases: 2% × \$180,000 = \$3,600/year.
• Net: \$3,600 − \$810 = \$2,790 in your favor.

When the rebate loses. Change one assumption. Suppose the 20 items you bid are not representative, and the rest of the basket also runs 3% high. Now the extra cost is 3% × \$180,000 = \$5,400**, against a \$3,600 rebate — you are \$1,800 worse off**, and the rebate check arriving in January feels like a win while you lose money every week.

The rule. A rebate is a discount you receive late, on a base somebody else defines. Before you sign, ask two questions in writing: what purchases qualify for the rebate, and what is the comparison price on my full basket, not on twenty items I chose. If a vendor will not answer the second one, that is your answer.

The honest note about relationships

Here is something a purely analytical treatment of this topic will get wrong, and I want to be direct about it.

A good vendor relationship is worth real money on a Friday when you are out of something.

At 3:15 on a Friday afternoon in October, with 142 covers on the books, you discover you are 40 portions short on the protein for the party at 6:30. What happens next depends almost entirely on whether your purveyor's driver will make an unscheduled stop, whether the sales rep will pull product from another route, and whether anyone at that company thinks of you as a customer worth inconveniencing themselves for.

That is not sentimentality. It is a real, quantifiable option that you hold or do not hold, and it is priced in the hundreds of basis points you gave up by not always buying the lowest bid.

So: it is legitimate to not always buy the lowest bid. What is not legitimate is failing to know what the loyalty costs you. Run the bid. Know that you are paying 2.5% more on a category. Decide, deliberately, that the emergency delivery and the honest substitution call and the rep who tells you when a price is about to move are worth 2.5%. Write down the decision.

The operator who never bids is not loyal — they are uninformed, and the vendor knows it. The operator who always takes the lowest bid has traded away every favor they will ever need. The operator who bids every quarter, knows the gap, and chooses to pay it has bought something real at a known price.

G.4 Receiving

Why receiving is the single highest-leverage control in the building

Receiving is where product and money change places. Every other control in this appendix operates on inventory you already own; receiving is the only one that operates on inventory before you have paid for it, which makes it the only control where the cost of being wrong is fully recoverable, on the spot, at no cost to the relationship. It is also the control most often delegated to whoever is nearest the door.

Consider the arithmetic. A restaurant at Bellwether's plan scale buys roughly \$335,000 of food a year [the Bellwether plan]. Suppose the receiving process leaks 1% of that — short weights, uncredited substitutions, price creep against the quoted price, a case of something signed for and never unloaded. That is \$3,350 a year, and it is a straight subtraction from operating profit.

In the terms this book uses: \$3,350 ÷ \$20.48 of contribution margin = 164 Hearth Chickens. You must sell a hundred and sixty-four additional plates of your best item to replace money you gave away at a door, standing up, holding a pen.

The procedure, in order

1. Have the order in front of you. The purchase order, the order guide, or the confirmed order email. If you do not know what you ordered, you cannot receive it — you can only accept it.
2. Do not sign first. The driver is in a hurry. The driver is always in a hurry. That is a scheduling problem between the driver and their employer, and it is not a reason to sign an invoice you have not checked.
3. Temperature first, on the refrigerated and frozen items. Take a product temperature — not an air temperature — with a calibrated thermometer, on the item most likely to be out of range. Refrigerated TCS product at 41°F or below; frozen product frozen solid with no evidence of thaw and refreeze. Log it. Reject at the door, not after the truck leaves. (Appendix E covers the food-safety detail and the local variation.)
4. Count everything. Cases, units within a case, bottles within a case. Open the case if the count is inside it. Yes, this takes longer.
5. Weigh anything billed by weight. Catch-weight proteins, cheese, produce billed by the pound. Use a calibrated scale. A scale that has not been calibrated is a decorative object.
6. Inspect condition and dates. Damage, temperature abuse, crushed cases, torn bags, code dates with insufficient life remaining, evidence of pests. Your spec (G.2) says what "acceptable" is; this is where that sentence earns its keep.
7. Check price against the order. Line by line, against the quoted price. Market items will move — that is what market means — but a market move should be a market move, not a quiet 6% on an item you have not looked at in a year.
8. Note every discrepancy on the invoice before you sign, and get the driver's initials on it. A note the driver did not see is a note the vendor will not honor.
9. Sign, take the copy, and put it away immediately.
10. Put it away in the right place, immediately. Produce and protein go in first. Every minute a case of chicken spends on the loading dock is shelf life you paid for and threw away.
11. Reconcile the invoice against the order in the office the same day, and file the credit request before you forget what happened.

Short shipments, substitutions, and credits

• Short shipment. You ordered four, three arrived. Note it on the invoice, get the driver's initials, and request a credit. Do not assume the driver's word that "they'll fix it on the next one." Sometimes they do. Sometimes there is no record.
• Substitution. The vendor was out and sent something else. Your spec (G.2) says whether you accept it and at what price. A substitution accepted without a price conversation is a price increase you agreed to. It is also, frequently, a yield change — whole romaine heads in place of hearts is not a like-for-like swap, and your salad cost card just moved.
• Credit. A credit memo is a document. It has a number. It shows up on a statement. Track credits requested against credits received, in writing, on a simple log. Unclaimed credits are one of the purest forms of leaked money in a restaurant, because there is no operational trade-off at all — you are simply not collecting money the vendor already agrees they owe you.

The rule

The person who orders should not be the only person who receives.

This is a separation-of-duties principle, and it is not primarily about theft, although it does make theft substantially harder. It is about error and self-verification. A person checking their own order is checking against their own memory of what they meant to do, which is the least reliable document in the building. A second person checks against the written order, which is what you wanted.

In a small restaurant with four managers, this is entirely achievable: the chef writes the order, the sous or the AM manager receives it. In a restaurant with two people, it is harder — and if it genuinely cannot be separated, then the compensating control is that the owner reconciles invoices to orders weekly, personally, at least for the first year.

⚠️ Where the Money Leaks

The invoice signed without a count.

This is the most expensive habit in the American restaurant industry and it costs nothing to fix.

Here is how it happens. The delivery arrives at 7:40 a.m., during prep. The prep cook is the only person at the back door. The driver hands over a clipboard and says the total. The prep cook signs, because the prep cook is not the person who is going to argue with a driver, and has never been told that arguing with the driver is part of the job.

Nobody is stealing. The driver is not a criminal. What actually happens is ordinary: a case gets left on the truck, a catch weight gets keyed from the wrong scale ticket, a substitution gets billed at the original item's price, a promotional price expires and nobody notices.

And then it happens again next week. And the week after. That is the part that turns an annoyance into a number: an unchecked receiving process does not produce one error, it produces a rate. A restaurant taking three deliveries a week takes 156 deliveries a year, and every one of them is an opportunity that a signed clipboard closes.

What the disciplined operator does: a receiving log (G.14), a calibrated scale, a calibrated thermometer, and a written rule that the invoice is signed by a manager, after the count, or the driver waits. Post the rule at the door. It survives about two uncomfortable conversations and then it is simply how your building works, and every driver on the route knows it.

Reading one

🧾 Read the Numbers

```text FIGURE G.1 — "Tuesday's broadliner delivery" [constructed teaching example] THE ARTIFACT One delivery invoice, Tuesday 7:40 a.m., with the receiving manager's notes. Deliberately dense with problems — a real invoice usually has one of these, not four. THE CONTEXT 68-seat full-service restaurant, mid-size Midwestern market. Three deliveries a week from this vendor. Illustrative prices throughout.

                 LINE  QTY    ITEM                              UNIT      EXTENDED
                 1     2 cs   Chicken, whole air-chilled 6/cs   $3.20/lb   $136.32
                                (billed catch weight 42.6 lb)
                 2     1 cs   Romaine hearts 12/3 ct            $34.50      $34.50
                 3     6 gal  Milk, whole                       $4.95/gal   $29.70
                 4     1 cs   Butter, unsalted 36/1 lb          $113.40    $113.40
                 5     2 cs   Onions, yellow jumbo 50 lb        $34.00      $68.00
                 6     1 cs   Napkins, dinner 2-ply 3,000/cs    $46.75      $46.75
                       Fuel surcharge                                       $12.00
                       ─────────────────────────────────────────────────────────────
                       INVOICE TOTAL AS PRESENTED                          $440.67

                 AT THE DOOR
                 Line 1  Scale reads 40.9 lb, not 42.6 lb. 1.7 lb x $3.20     -$5.44
                 Line 2  WHOLE romaine 24 ct substituted, not ordered,
                         not priced as a substitute. Rejected.               -$34.50
                 Line 3  Product temperature 47°F at the door. Rejected.     -$29.70
                 Line 4  Order guide quoted $106.20. Billed $113.40.          -$7.20
                         Held pending vendor confirmation.
                       ─────────────────────────────────────────────────────────────
                       CREDITS NOTED AND INITIALED BY DRIVER                 $76.84
                       CORRECTED INVOICE                                     $363.83

WHAT IT SHOWS $76.84 on a $440.67 invoice — 17.4% — recoverable in about six minutes with a scale, a thermometer, and the order in hand. Three of the four problems are ordinary logistics, not misconduct. The fourth (the butter price) may be a legitimate market move; it becomes a conversation, not an accusation. WHAT IT DOESN'T It does not tell you whether the butter price is a market move or a quiet increase — only the vendor's quote sheet settles that. It does not show what the rejected milk costs you in service today, which is a real cost of rejecting. And it does not show the substitution's second-order effect: whole romaine yields less usable leaf than hearts, so accepting it would have moved a cost card as well as a price. THE DECISION Reject lines 2 and 3, note the short weight on line 1, flag line 4 for the rep before noon. Call the produce house for romaine hearts on today's afternoon run. Log all four in the credit tracker and check the statement. THE LESSON Receiving is the only control that recovers money before you have spent it. Six minutes at the door beats six hours of variance analysis at the end of the period, because at the end of the period the product is gone and all you have is a number you cannot explain. ```

G.5 Storage

The four zones and what each should hold

Temperature and time controls, cooling procedures, date marking, and thermometer calibration are in Appendix E and, above it, in your local adoption of the FDA Food Code. Verify locally.

The storage hierarchy, and a correctly loaded rack

Raw animal products are stacked in order of their minimum required cooking temperature — lowest on top, highest on the bottom — so that a drip from anything above cannot contaminate something that will be cooked to a lower temperature or not cooked at all. Ready-to-eat food goes above everything.

FIGURE G.2 — A correctly loaded walk-in rack, top to bottom
                                                        [see Appendix E for the safety detail]

   ┌──────────────────────────────────────────────────────────────────────┐
   │  SHELF 1  (top)   READY-TO-EAT / COOKED / WASHED PRODUCE             │
   │                   cooked proteins, prepped sauces, dressed greens,   │
   │                   dairy, desserts, anything served without cooking   │
   ├──────────────────────────────────────────────────────────────────────┤
   │  SHELF 2          SEAFOOD                       (cook to 145°F)      │
   ├──────────────────────────────────────────────────────────────────────┤
   │  SHELF 3          WHOLE CUTS — beef, pork       (cook to 145°F)      │
   ├──────────────────────────────────────────────────────────────────────┤
   │  SHELF 4          GROUND MEAT AND GROUND FISH   (cook to 155°F)      │
   ├──────────────────────────────────────────────────────────────────────┤
   │  SHELF 5 (bottom) POULTRY, whole and ground     (cook to 165°F)      │
   └──────────────────────────────────────────────────────────────────────┘
      Bottom shelf sits at least 6 in. off the floor. Nothing on the floor, ever.
      Nothing under a condensate line. Nothing blocking the fan or the coil.
      Six inches of clearance from the walls so air can actually move.

Read that diagram as a drip diagram. Gravity is the only mechanism it defends against, and it is a mechanism that operates whether or not anyone is watching. The running inspection scenario in this book — raw chicken stored above ready-to-eat greens — is the single most common critical violation an inspector writes, and it is caused not by ignorance but by a Saturday night when someone needed a shelf and took the nearest one.

The countermeasure is labeling the shelves, not training the staff. Training decays. A label that says POULTRY ONLY on the bottom rack is still there in March.

FIFO in practice rather than in theory

FIFO — first in, first out — means the oldest product is used first. Everyone knows this. Almost nobody executes it, for a simple physical reason: the new case is in your hands and the old case is behind it, and putting the new one behind the old one takes forty extra seconds.

What actually makes FIFO happen:

• Date everything on receipt, in the same place, in the same format. A label that requires interpretation will be ignored.
• Load from the back, and design the shelving so that is physically easy — open shelving, not deep bins, with enough space to slide a case.
• Assign the walk-in to a person on each shift. Not "everyone." A person. Rotation happens because it is somebody's job at a specific hour, usually the start of PM prep.
• Make the oldest product the most accessible. This is the actual trick. If the front of the shelf is where cooks reach, FIFO stops being a discipline and becomes the path of least resistance.
• Walk the walk-in daily. Five minutes, by a manager, looking for anything undated, anything unrotated, and anything quietly about to become waste. It is the highest-yield five minutes in a kitchen manager's day.

Labeling

Every prepped item, every opened package, every transferred container gets a label with:

Chemicals

Cleaning chemicals, sanitizers, pesticides, and any other toxic material are stored separately from food, food equipment, utensils, linens, and single-use items — in a designated area, below or apart from food storage, never above, and never in an unlabeled container. Never in a food container, and never decanted into anything that looks like something you would drink. This is not a quality rule; it is a rule about not poisoning somebody, and it is one of the violations that will close a restaurant. The detail is in Appendix E and in your local code.

The bar

The bar is a storage area that most operators do not treat as one. It should have a locked liquor storeroom with one person's access per shift and a documented transfer between shifts; a par-stocked back bar (G.9) so the gap between par and actual is a number rather than a feeling; wine at a stable temperature, on its side for cork-finished bottles, away from the dish machine and the fryer exhaust; kegs on a first-tapped, first-emptied rotation with the tap date written on the collar; and beer rotated hard, because beer is a perishable product and a dusty case of it is a quality problem, not a cellar.

G.6 Shelf life

The safety limit and the quality limit are different numbers

This distinction is the most useful idea in this section, so it goes first.

• The food-safety limit is the point past which the product may be unsafe. For ready-to-eat, time/temperature-controlled food held refrigerated, the Food Code's date-marking rule caps this at seven days at 41°F or below, counting the day of preparation as day one. Local adoption varies — verify. This limit is not a matter of your judgment.
• The quality limit is the point past which the product is no longer good enough to serve. It is usually much shorter.

Most shelf-life decisions in a good kitchen are quality decisions made well before the safety limit. A stock is safe on day six and tastes tired on day four. A fish fillet is safe on day three and should have been cooked on day one. A cut herb is safe on day five and is compost on day three.

The consequence for a manager is precise: you may never extend past the safety limit for any reason, and you should routinely discard well before it. If your kitchen's shelf lives are set at the safety maximum, you are serving food at the edge of what is legal and nowhere near the edge of what is good, and your guests will notice before your inspector does.

The table

All figures below are planning ranges only, assuming proper receipt, proper temperature, and proper handling. Verify against your own product, your own equipment, and your local code. Never exceed the safety limits in Appendix E, whatever this table says.

Fresh proteins (refrigerated at 35–38°F)

Cooked proteins and prepared items (refrigerated, properly cooled)

Dairy (refrigerated)

Produce (refrigerated unless noted)

Dry goods (50–70°F, 50–60% humidity, sealed, off the floor)

Opened bar product

⚠️ Where the Money Leaks

The shelf life nobody set.

An item with no stated shelf life gets one of two treatments, and both cost money.

One: it never gets thrown away. The quart of sauce migrates to the back of the shelf, gets used at the bottom of a busy Saturday, and goes onto a plate at eleven days old. You did not save money. You spent a guest.

Two: it gets thrown away every time anyone is unsure. A cook looks at an undated container, cannot tell how old it is, and does the responsible thing. You threw away four good days because you would not spend four seconds with a marker.

What the disciplined operator does: every prep recipe carries its shelf life in the recipe, so the use-by date is computed when the label is written, by the person writing it. Not looked up. Not remembered. Printed on the card, next to the yield.

G.7 Yield

What a yield test is

A yield test measures how much usable product you actually get from what you bought.

As-purchased (AP) weight is what you paid for. Edible portion (EP) weight is what ends up on plates. The yield percentage is EP ÷ AP. The gap between them is bone, skin, fat, peel, trim, seeds, tops, cores, and the small tragedies of a dull knife.

Yield matters because your cost card is built on EP cost and your invoice is denominated in AP cost, and if you cost a recipe using the invoice price you will understate your food cost by whatever the yield loss is. On a whole chicken, that error is on the order of forty percent. It is not a rounding difference. It is the difference between a business and a hobby.

How to run one, step by step

1. Pick the item and the specific preparation. "Carrots" is not a yield test. "Carrots, peeled and cut into 1-inch oblique" is. The same carrot has three yields for three cuts.
2. Use a real sample. At least 5–10 units for proteins, at least a full case for produce. One chicken tells you about one chicken.
3. Use your own cook, your own knife, your own station, at a normal pace. A test run carefully by the chef on a slow Tuesday produces a number your kitchen will never hit again.
4. Weigh the AP quantity and record the invoice price and unit.
5. Fabricate it exactly as the recipe requires.
6. Weigh every output stream separately: primary EP product, usable secondary product (bones, trim for stock, ends for a purée), and true waste.
7. Yield percentage = EP weight ÷ AP weight.
8. EP cost per unit = AP cost per unit ÷ yield percentage. (Divide. See the warning below.)
9. Portion cost = EP cost × portion size.
10. Record the date and re-run it when the spec, supplier, season, or the cook on the station changes — twice a year at minimum on your top-spend items.

The blank yield test worksheet

The arithmetic, shown once, clearly

This is the calculation that makes every table in this section mean anything.

FIGURE G.3 — Cost per usable unit                            [constructed teaching example]

  STEP 1   AP cost, from the invoice ......................  $3.20 per lb
  STEP 2   Yield, AP to EP, from YOUR test ................  56%   (0.56)
  STEP 3   EP cost = AP cost / yield ......................  $3.20 / 0.56 = $5.71 per usable lb
  STEP 4   Portion cost = EP cost x portion weight ........  $5.71 x 0.375 lb (6 oz) = $2.14

  DIVIDE by the yield. Do not multiply.
     Wrong:  $3.20 x 0.56 = $1.79 per lb   <-- understates by $3.92 on every pound
     Right:  $3.20 / 0.56 = $5.71 per lb

  Sanity check, always available: the EP cost must be HIGHER than the AP cost,
  because you threw part of it away. If your EP cost came out lower, you multiplied.

⚠️ Where the Money Leaks

Multiplying instead of dividing.

The error above is not hypothetical and it is not a beginner's error. It shows up in cost cards built by experienced chefs, in spreadsheets inherited from a previous manager, and in software configured by somebody who did not read the field label.

The reason it survives is that it produces a food cost percentage that looks wonderful, and nobody audits a number that flatters them. A kitchen costing every fabricated protein at AP × yield instead of AP ÷ yield will report a theoretical food cost in the low twenties and run an actual in the mid-thirties, and it will spend a year hunting for theft that does not exist.

What the disciplined operator does: the sanity check in the diagram. EP cost is always higher than AP cost. Always. Put that sentence at the top of your cost-card template.

A fully worked example: the whole chicken

🧮 Run the Numbers

Yield test — whole air-chilled chicken, fabricated to boneless meat. (constructed illustrative figures, mid-size Midwestern market)

The test. Ten whole air-chilled birds to the spec in G.2, fabricated by the PM prep cook at normal pace on a Wednesday.

Line	Figure
AP weight, 10 birds	35.0 lb (3.5 lb average)
AP cost	35.0 lb × \$3.20/lb = **\$112.00**
EP — boneless skin-on meat	19.6 lb
Frames, bones, wingtips, fat, skin trim	15.4 lb
Check	19.6 + 15.4 = 35.0 lb ✓

Yield. 19.6 ÷ 35.0 = 56.0%

EP cost. \$3.20 ÷ 0.56 = **\$5.71 per usable pound**

Cross-check the other way: \$112.00 ÷ 19.6 lb = \$5.71. Same number, which is how you know you did it right.

Portion cost. A 6 oz portion is 0.375 lb: \$5.71 × 0.375 = **\$2.14**

---

Now the secondary product, and the honest version of it.

The 15.4 lb of frames make roughly four gallons of stock. You currently buy stock at \$6.00/gal, so not buying it saves \$24.00.

Credit that against the test: \$112.00 − \$24.00 = \$88.00 ÷ 19.6 lb = **\$4.49 per usable pound**, and the 6 oz portion drops to \$1.68.

That is a legitimate calculation and it is also the most abused calculation in kitchen costing. Two conditions must both hold before you take the credit:

1. You would otherwise have bought the thing. If you would not have bought stock, you have not saved \$24.00 — you have made \$24.00 of stock you did not need, with labor you did pay for.
2. You counted the labor. Four gallons of stock is a stockpot, a burner, a strain, a cool-down, and a container. Chapter 1's argument applies directly: the make-or-buy decision moves cost from one half of prime cost to the other, and the only way to know whether the trade was good is to watch the total.

The conservative default is not to take the credit on your primary cost card. A cost card that quietly assumes perfect byproduct utilization describes a kitchen that does not exist.

The Hearth Chicken, and why its card looks different

Bellwether's signature dish is a half bird, roasted bone-in on the hearth [the Bellwether plan]. That preparation puts essentially the whole half bird on the plate, so the cost card correctly uses the as-purchased figure:

½ × 3.5 lb × \$3.20/lb = \$5.60.

Against roasted roots \$0.95, salsa verde \$1.05, butter and aromatics \$0.42, oil and seasoning \$0.18, garnish \$0.15, the components total \$8.35. A 2% waste and spillage allowance brings the plate to \$8.52**. At a **\$29.00 menu price that is a 29.4% food cost and a contribution margin of \$20.48.

This is the correct treatment, and it is the exception that makes the rule legible: the yield question is always "what fraction of what I bought reaches the plate?" For a bone-in half-bird preparation the answer is essentially all of it, so AP and EP cost converge. Change the preparation to a boneless breast and the same bird's usable-pound cost jumps from \$3.20 to \$5.71 — a 78% increase, driven entirely by fabrication.

🧮 Run the Numbers

What a 0.3-pound drift in average bird weight costs. [the Bellwether plan]

The spec in G.2 says 3.25–3.75 lb, case average not to exceed 3.60 lb. Suppose it slips and the birds average 3.8 lb for a quarter — still within the individual range, over on the case average, and something nobody would notice without a scale.

• New chicken component: ½ × 3.8 lb × \$3.20 = **\$6.08** (baseline: \$5.60)
• New component total: \$8.35 − \$5.60 + \$6.08 = **\$8.83**
• Plus the same 2% allowance: \$8.83 × 1.02 = **\$9.01** (baseline: \$8.52)
• At the unchanged \$29.00 price: \$9.01 ÷ \$29.00 = 31.1% food cost (baseline: 29.4%)
• Contribution margin: \$29.00 − \$9.01 = **\$19.99** (baseline: \$20.48)

\$0.49 a plate. 1.7 points of food cost on the item.

Now annualize it. At the volume used in G.9 — 106 orders a week — that is \$0.49 × 106 × 52 = **\$2,701 a year**, on one menu item, from a weight drift that no cook, no server, and no guest would ever notice.

This is what Chapter 1 means by cost drift, expressed as a specific mechanism. It is also the entire argument for putting a weight range in a spec and a scale at the receiving door.

The planning-yield table

Planning ranges only. Run your own tests. Yields depend on grade, size, season, supplier, the exact cut you are making, and the skill of the person making it. The bottom of each range is a realistic outcome for a rushed line; the top is a careful, skilled fabrication.

Poultry

Beef and pork

Note what happens on those last few rows. Cooking loss is a separate loss from fabrication loss, and a braised or roasted item incurs both. If your recipe portions a cooked weight, your cost card needs the cooked yield, which is a different and usually much lower number.

Seafood

Produce

Citrus and dairy

👨‍🍳 On the Line

Why bar costing goes wrong on citrus.

A bar runs on juice, and juice is where beverage cost cards drift the hardest, because nobody weighs a lemon — they count them.

Citrus is packed by count per case, and a lower count number means a larger fruit. That is the convention across citrus and most tree fruit, and it is backwards from what people expect. Confirm the specific counts your supplier carries; they vary by region and season.

Take a 40 lb case of lemons at a 115 count: 640 oz ÷ 115 = 5.6 oz per lemon. At a 35% juice yield, roughly 2 fl oz of juice per fruit.

Now a 40 lb case at a 165 count: 640 ÷ 165 = 3.9 oz per lemon, and about 1.4 fl oz of juice.

Note carefully what does and does not change. The case weighs the same, so the total juice in it is about the same, and your cost per ounce of juice barely moves. What moves — by roughly 30% — is the juice per fruit. And that is the number every bar in America actually works from.

So: a prep list that says "juice 30 lemons for service" delivers 60 ounces from one case and 42 from the other, and the bar runs dry at nine on a Friday. A cost card that says "1 lemon = 2 oz juice" is 30% wrong the week the count changes. A garnish spec written for 115s produces wheels the wrong size on 165s.

This is why the spec (G.2) carries a size-or-count field, and why "lemons" on an order guide is not a purchasing instruction. It is a hope.

G.8 Conversions

The three systems that collide in a kitchen

Restaurants measure in weight (pounds, ounces, grams), volume (gallons, quarts, cups, fluid ounces, liters), and count (each, dozen, case, per-pound counts). Every costing error of any size starts where two of these meet.

The compact conversion table

Weight

Volume (US)

Bar and beverage

(Keg sizes are the common US standards; the sixth-barrel in particular is sometimes stated slightly differently and the 20-liter European "sixtel" is a different volume. Confirm with your distributor.)

Useful density anchors

The conversions that cause errors

1. Volume to weight for anything that is not water.

⚠️ Where the Money Leaks

Costing flour, greens, or chopped product by volume.

"A cup of flour" is not a quantity. It is a range.

Spooned into the cup and leveled, all-purpose flour runs around 4.25 oz. Dipped and packed straight from the bin — which is what a busy cook does — it can run 5.5 oz. That is a 29% error, in the same kitchen, on the same day, from the same bag.

Chopped and leafy product is worse. A quart container of chopped parsley can weigh four ounces or ten, depending on how finely it was chopped and how hard it was pressed in, and both of those containers look identical on a shelf. A "quart of mirepoix" is a hopeful description of a volume of air containing some vegetables.

The consequences run in two directions and both cost you. A recipe costed low understates food cost and produces a menu price that does not cover the plate. A recipe executed to a volume measure over-portions the expensive ingredient, and the dish costs more than the card says forever.

What the disciplined operator does: weigh everything on a cost card. Write recipes in weight. Buy a \$40 digital scale for every station and treat a broken one as an emergency. Volume is acceptable for liquids that behave like water — stock, water, milk, oil at the margin — and for nothing else.

2. A "case" is not a fixed quantity.

There is no standard case. A case of tomatoes is 25 lb. A case of another vendor's tomatoes is 20 lb. A case of chicken is 6 birds here and 8 birds there. A case of #10 cans is usually 6, and "usually" is doing work in that sentence.

The consequence: you cannot compare two vendors' case prices. You can only compare a price per usable unit, which means converting both to a common denominator before you look at them. The bid comparison form in G.14 has a column for exactly this, and it is the only column that matters.

Practical notation, which you should insist your order guide uses:

A #10 can holds roughly 6 lb 9 oz to 7 lb 5 oz of product depending on what is in it, and roughly a bit over 3 quarts by volume. Confirm the drained weight with the label, because the drained weight is what you actually cook with and it can be dramatically less than the net weight on the same can.

3. Count sizes on shellfish and produce.

• Shrimp and scallops are sold by count per pound. 16/20 means 16 to 20 shrimp per pound — so a lower number is a larger shrimp. U/15 means "under 15 per pound," i.e., very large. This convention is standard and reliable.
• Produce is sold by count per case, and here too a lower number means a larger item — a 90-count russet case and a 60-count russet case both weigh 50 lb, but the 90s are about 8.9 oz each (800 oz ÷ 90) and the 60s are about 13.3 oz each. If your baked-potato spec says "one potato," you have just specified a portion that can vary by half.
• Oysters are sold by count per box or bushel and by regional size grades whose names differ by coast and by species. Do not guess. Ask your purveyor what the grade names mean in their house and write it into the spec.

G.9 Par levels

What a par is

A par level is the quantity of an item you want to have on hand at a defined moment — typically just before a delivery, or at open, or at the start of a shift. The order quantity is the par minus what you actually have.

$$\text{Order quantity} = \text{Par} - \text{On hand}$$

A par is not a maximum and it is not a minimum. It is a target, chosen to make the answer to "how much do I order?" a subtraction instead of an argument.

Setting one from usage rather than from intuition

The formula:

FIGURE G.4 — The par level formula

  PAR  =  ( average usage per period  x  periods of coverage )  +  safety stock

  where:
    average usage per period ..  from the POS or from counts, over at least 4 periods
    periods of coverage ......  order cycle + lead time, measured in the same units
    safety stock .............  a cushion, expressed as a % of coverage or as days

  then:  ROUND UP to the vendor's pack size, and record the rounded figure as the par

Three of those inputs get set wrong routinely:

• Average usage. It must come from data — POS item counts, or counted usage from G.11 — over at least four comparable periods. A number the chef recalls is not usage.
• Periods of coverage. This is the order cycle plus the lead time, not the order cycle alone. If you order Monday for a Tuesday delivery and the next delivery is Friday, the Tuesday delivery must cover Tuesday, Wednesday, and Thursday — three service days, not one.
• Safety stock. For a shelf-stable item, 20–30% is cheap insurance. For a highly perishable item, safety stock is not free — it is spoilage waiting to happen — and 10% or less may be correct. Safety stock is a deliberate trade between the cost of running out and the cost of throwing away, and it should be set item by item, not globally.

A worked example

🧮 Run the Numbers

Setting the par for whole chickens. [the Bellwether plan] (constructed illustrative figures)

Step 1 — usage from data. Hearth Chicken orders from the POS, last four weeks:

104, 98, 115, 107 → total 424 orders ÷ 4 = 106 orders per week.

Each order is half a bird, so weekly bird usage = 106 ÷ 2 = 53 birds per week.

Step 2 — distribute across the week. From the same POS report, the four-week average share of Hearth Chicken orders by service day (dinner Tuesday through Saturday; the item does not run at brunch):

Tue	Wed	Thu	Fri	Sat	Total
14%	16%	18%	25%	27%	100%

Step 3 — periods of coverage. Poultry delivers Tuesday and Friday mornings, ordered by 2 p.m. the previous day.

• The Tuesday delivery must cover Tue + Wed + Thu = 14 + 16 + 18 = 48% of the week.
• The Friday delivery must cover Fri + Sat = 25 + 27 = 52% of the week.

Step 4 — the arithmetic.

	Tuesday par	Friday par
Coverage share	48%	52%
Birds needed (53 × share)	25.44	27.56
Safety stock at 20% (× 1.20)	30.53	33.07
Pack size (6 per case)	5.1 cases	5.5 cases
Par, rounded to the pack	30 birds (5 cases)	36 birds (6 cases)

Step 5 — the rounding decision, made deliberately. Tuesday rounds down to 30 against a padded figure of 30.5, which still leaves 4.6 birds of cushion over the raw forecast of 25.4. Friday rounds up to 36 against 33.1, because Saturday is the night you cannot be short and because chicken at 41°F holds two days comfortably.

Step 6 — the order. Tuesday morning's count shows 7 birds on hand.

Order = par − on hand = 30 − 7 = 23 birds → round to the pack → 4 cases (24 birds), arriving at 31 on hand.

What this buys you. The order is now a thirty-second subtraction that a sous chef can do correctly, in the same way, every week, whether or not the chef is in the building. That is the entire point of a par: it converts a judgment call made 156 times a year into a calculation.

How pars change

A par set in October is wrong in July. Four things move it, and all four are predictable:

• Seasonality. Patio seats open, brunch grows, winter thins Tuesday and thickens Saturday. Revisit pars at each menu change and each seasonal inflection.
• Day of week. Handled above by splitting coverage rather than using a flat weekly average. A flat average on a business with a 27%-Saturday / 14%-Tuesday shape guarantees you are short on Saturday and long on Tuesday, every week.
• Menu change. A new dish using the same protein raises the par; a dish coming off lowers it — and the item you forget to lower is the one that spoils.
• A known event on the books. A 40-top private party, a holiday, a street festival, a concert two blocks away. The reservation book is the best forecasting instrument in the building and almost nobody uses it for purchasing. A manager who reads next week's book on Sunday and adjusts three pars has done more for food cost than a month of policing portions.

The failure modes

Pars set once and never revisited. The most common. A par sheet with no revision date on it is describing a restaurant that no longer exists. Put a revision date on every par sheet, review quarterly, and review every par attached to an item whose menu placement changed.

Pars set to the busiest week. The instinct is protective and the cost is real. A par set so you never run out on your best Saturday means you carry that inventory for fifty-one other weeks. The money shows up as inventory value (which is cash you cannot spend — see Chapter 33), as spoilage, and as a walk-in so full that FIFO becomes physically impossible. Set the par to normal and manage the exception by hand. The busiest week gets a special order, not a permanent par increase.

Pars for items that should not have one. Some items should be bought to the forecast every single time, not to a standing quantity:

• Highly perishable center-of-plate items with short shelf lives — fresh fish especially. You order fish for Thursday's service on Wednesday, against Thursday's book. A par on fresh fish is a spoilage schedule.
• Items whose minimum order exceeds a period's usage. If the smallest unit you can buy is three weeks' worth, you do not have a par, you have a purchase decision.
• Event and special items, by definition.
• Anything on the menu for fewer than four weeks, because you do not yet have usage data and a par built on guesses is a guess with a number next to it.

G.10 The physical count

Why the count is the whole thing

Every number in G.11 and G.12 depends on an accurate ending inventory. Food cost is not a number your accountant produces; it is a number your count produces and your accountant reports. A sloppy count does not produce a sloppy answer — it produces a confidently wrong answer, which is worse, because you will act on it.

How to run one

Tools: count sheets organized by location, a clipboard or tablet, a scale (a bench scale for cases and a smaller one for partials), a flashlight for the back of the walk-in, a pen, and two people.

Timing: at the same point relative to deliveries, every single period. If you count Sunday night after the last delivery of the week, you count Sunday night after the last delivery of the week — always. Counting after a Tuesday delivery in one period and before it in the next produces a swing that has nothing to do with your operation and will send you hunting for a problem that does not exist.

Count after service and after the last delivery, with the kitchen closed. Counting during prep means counting product that is moving, which means counting it twice or not at all.

The order:

1. Dry storage
2. Walk-in cooler
3. Reach-ins and prep tables (yes, including the pans on the line — that product is inventory)
4. Freezer
5. Bar: back bar, speed rails, well, then the liquor storeroom
6. Beer cooler and keg box
7. Wine storage
8. Paper, disposables, and chemicals (if you track them, and you should)

Organize the count sheet by storage location, not alphabetically

This is the single most consequential formatting decision in inventory, and it sounds trivial until you have done it both ways.

An alphabetical count sheet forces the counter to walk the building in alphabetical order. They find anchovies in dry storage, then apples in the walk-in, then arborio rice back in dry storage. They cross the kitchen forty times, lose their place, and miss the second location where an item also lives. It takes two hours and it is full of holes.

A location-organized sheet lists items in the order you physically encounter them — shelf by shelf, left to right, top to bottom, one storage area at a time. You walk the walk-in once. In my experience it roughly halves the time, and, far more important, it sharply reduces the two errors that actually corrupt an inventory: missed items and double-counted items.

Two rules follow: re-sequence the sheet whenever you rearrange a shelf, because a sheet that no longer matches the layout teaches the counter to skip around; and list an item in every location where it lives, with the location named, so the counter adds them rather than choosing one.

The two-person rule

One person counts and calls. One person records.

This is faster than one person doing both — the caller never puts down the product, the recorder never puts down the pen — and substantially more accurate, because the recorder hears the number and the unit and will question anything that sounds wrong. It also means no single person's count of the most valuable inventory in the building goes unverified.

It matters most in the liquor storeroom and among the walk-in proteins: the two highest-value-per-cubic-foot areas in a restaurant, and the two where an error costs the most.

Partial cases, partial bottles, and the tenths problem

• Partial cases: weigh them or count the units. Do not eyeball "about half a case." A 50 lb sack that is "about half" is somewhere between 18 and 32 lb, and on an expensive item that spread is material.
• Partial containers of prepped or bulk product: weigh, using a tare. Every count-sheet line for a bulk item should have a unit a scale can produce.
• Partial bottles at the bar: estimate to the tenth. A bottle judged by eye as "half" is routinely 0.4 or 0.6, and across sixty open bottles that is real money. Three methods, in increasing order of accuracy and cost: 1. Visual tenths, using a marked bottle or a printed template behind the bottle. Free, fast, adequate if everyone uses the same method every time. 2. Weighing, with the empty bottle weight tared out. Considerably more accurate. Slower. 3. A metered pour or inventory system. Most accurate, real capital cost, and worth it above a certain beverage volume.

The critical rule is consistency: an estimate that is biased the same way every period still produces a correct trend, and the trend is what you are actually managing. An estimate whose method changes between periods produces noise indistinguishable from theft.

Counting the bar

The bar deserves its own paragraph because it is where inventory discipline most often stops.

Count the back bar and speed rails (all those open partial bottles), the well, the storeroom, the beer cooler, the keg box (a tapped keg is a partial — estimate it or weigh it), the wine, and the prepped bar product: juices, syrups, batches, garnishes. Prepped bar product is inventory and it is almost universally uncounted, which is one reason reported pour cost and actual pour cost so often disagree.

👨‍🍳 On the Line

What a real count night looks like.

Two people, clipboard and scale, starting when the last table leaves. Dry storage first because it is easy and it warms you up. Walk-in second, while you still have the patience for it. Freezer fourth, because the freezer is miserable and you want it behind you before you are tired.

Ninety minutes to two hours for a restaurant this size, if the sheets are sequenced correctly and the storage is organized. Four hours if they are not — and the last hour of a four-hour count is worthless, because the counter is estimating to get out of the building, and estimating at midnight is how a two-point variance appears out of nowhere.

The compounding effect nobody mentions: an organized walk-in is faster to count and faster to work out of and rotates better and wastes less. The count is not a separate discipline from the kitchen's organization. It is a monthly audit of it, and it will tell you the truth whether or not you wanted to hear it.

G.11 The usage formula and variance

The formula

$$\text{Usage} = \text{Beginning inventory} + \text{Purchases} - \text{Ending inventory}$$

This is the only honest way food cost is ever computed, and Chapter 1 makes the point in its opening pages: the operator who divides invoices by sales is not measuring usage, they are measuring buying, and the two differ by whatever the inventory level did.

A worked example

🧮 Run the Numbers

One four-week period at Bellwether. [the Bellwether plan] (constructed illustrative figures)

Line	Amount
Beginning food inventory	\$8,900
+ Food purchases	\$26,500
− Ending food inventory	\$9,400
= Food usage	\$26,000
Food sales for the period	\$85,800
Actual food cost	\$26,000 ÷ \$85,800 = 30.3%

(Check: \$8,900 + \$26,500 = \$35,400; \$35,400 − \$9,400 = \$26,000. ✓)

Now the shortcut, for comparison. Invoices ÷ sales = \$26,500 ÷ \$85,800 = 30.9%.

That is six-tenths of a point wrong, in this period, in the pessimistic direction — because inventory rose \$500, meaning \$500 of what you bought is still sitting on the shelf and has not been used. Buy heavy in the last week of a period and the shortcut punishes you for nothing. Run the walk-in down and the shortcut congratulates you for nothing.

Neither number has anything to do with what you actually used, which is the only figure you can manage.

Theoretical usage, from the POS item counts multiplied by the cost cards: \$24,900, or \$24,900 ÷ \$85,800 = 29.0%.

Variance = \$26,000 − \$24,900 = \$1,100**, which is \$1,100 ÷ \$85,800 = 1.3 points of food cost**.

Annualized over 13 four-week periods: \$1,100 × 13 = \$14,300 a year.

Say that in the book's other currency: \$14,300 ÷ \$20.48 of contribution margin = 698 Hearth Chickens. That is roughly seven weeks of your best-selling entrée, sold entirely to cover a variance nobody has diagnosed.

Theoretical versus actual

• Theoretical usage is what you should have used: every item the POS says you sold, multiplied by the cost card for that item.
• Actual usage is what you did use: the count formula above.
• The variance is the gap, and the gap is the waste, the over-portioning, the receiving errors, the uncosted specials, the comps, and — sometimes — the theft.

Two honesty notes before you chase one:

1. A theoretical number is only as good as the cost cards behind it. If half your specials are uncosted and your cost cards are eighteen months old, your theoretical is fiction and your variance is measuring your bookkeeping, not your kitchen.
2. Some variance is normal. A hand-counted inventory in a working restaurant has real measurement error in it. A variance under about half a point is inside the noise of the instrument.

What sizes of variance mean

That last row deserves emphasis, because a negative variance is the one people accept without question. If your actual usage is meaningfully below your theoretical usage, you have a counting or costing error, and next period it will reverse and appear as a large positive variance you will then misdiagnose.

The diagnostic order

When you have a variance, chase it in this order. Not the order that is most interesting. This order.

FIGURE G.5 — Chasing a variance, in order of likelihood

  1. ARITHMETIC AND COUNTING
       Recount the ten highest-value lines. Check extensions and unit prices on the
       count sheet. Check that the beginning inventory equals last period's ending.
       Check that purchases include every invoice and no invoice twice.
       Check that transfers between kitchen and bar were recorded in both directions.
       >> The single most common cause of a large variance is a data error.

  2. RECEIVING
       Pull the period's invoices against the orders. Short shipments, unweighed catch
       weights, price increases nobody approved, credits requested and never received.
       >> See G.4. This is the second most common cause and the cheapest to fix.

  3. PORTIONING AND PRODUCTION
       Are the cost cards current? Are the specials costed? Is the line using the
       specified scoop, ladle, and scale? Has a yield changed because a spec changed?
       >> See G.7. This is where a slow, permanent leak usually lives.

  4. WASTE, SPOILAGE, AND COMPS
       Read the waste log and the comp report. Is production forecast to the book?
       Is FIFO actually happening? Are comps running above their normal rate?
       >> See G.13. Often the variance is fully explained here and nobody had looked.

  5. THEFT
       Only now. And when you get here, you are looking for a PATTERN in the data
       first -- a specific item, a specific shift, a specific door -- before you are
       looking at a person.

⚠️ Where the Money Leaks

Starting at theft.

When a manager sees a two-point variance, the first thought is very often that somebody is stealing. I want to be direct about why that instinct is both usually wrong and genuinely corrosive.

Usually wrong, because the base rates are not close. In every variance I have personally chased to a conclusion, the great majority resolved in the first three steps of the list above — a transposed count, an unrecorded transfer, an invoice entered twice, a case never delivered, a cost card that predated a price increase. Theft happens and it is real, and it is still not where the money usually is.

Corrosive, because the investigation is not free even when you are right, and it is very expensive when you are wrong. A kitchen that learns the owner's first response to a bad number is suspicion will stop telling the owner about bad numbers. Your waste log goes quiet. Your breakage stops getting reported. The prep cook who spilled six quarts of stock cleans it up and says nothing. You have just destroyed the reporting system you needed to find the actual problem, and you have done it in the name of finding it.

What the disciplined operator does: work the list in order, in front of the team, out loud. "We're 1.3 points off. Let's start with my arithmetic." A manager who begins with their own numbers has earned the right to reach step five if the data takes them there — and by then it will not be an accusation, it will be a specific item, on a specific shift, with a specific pattern behind it.

G.12 Inventory turnover

The formula

$$\text{Inventory turnover} = \frac{\text{Usage (COGS) for the period}}{\text{Average inventory at cost}}$$

where average inventory = (beginning + ending) ÷ 2.

The more intuitive companion figure is days of inventory on hand:

$$\text{Days on hand} = \frac{\text{Days in the period}}{\text{Turnover for the period}}$$

🧮 Run the Numbers

Bellwether's food turnover, same period as G.11. [the Bellwether plan]

• Average inventory = (\$8,900 + \$9,400) ÷ 2 = \$9,150
• Turnover for the 28-day period = \$26,000 ÷ \$9,150 = 2.84 turns
• Annualized: 2.84 × 13 periods = ~37 turns a year
• Days on hand = 28 ÷ 2.84 = 9.9 days

The interpretation: roughly ten days of food sitting in the building at any moment. For a fresh-driven, short-menu restaurant that is on the heavy side of reasonable — not alarming, but worth a look at which categories are carrying it. Almost certainly the dry storage and the freezer, not the walk-in.

And the cash reading, which is the one operators skip: \$9,150 of average inventory is \$9,150 of cash sitting on shelves. Get days on hand from 9.9 to 7.5 and you release roughly \$2,200 of working capital — permanently, one time — without changing a single thing about how you cook.

What a healthy turn looks like

Planning ranges only. These vary enormously by concept, menu length, delivery frequency, and how much of your menu is fresh versus dry or frozen.

Notice the wine row. Wine turns slowly on purpose. A cellar with real depth is a deliberate investment in inventory, and judging it against a food turnover benchmark will lead you to a catastrophically wrong conclusion. Wine is the one category where a slow turn may be exactly the strategy — and it is also the category where a slow turn can hide dead stock that will never sell. Distinguishing the two requires a bottle-level velocity report, not a category turnover number.

Why a very high turn is not automatically good

Every operator's instinct is that faster is better: fresher product, less cash tied up, less spoilage. All true, up to a point, and then it inverts.

A very high turnover may simply mean you are running out of things, and that cost does not appear anywhere on a P&L. The 86'd item is a sale you did not make and often a guest who ordered their second choice and enjoyed it less. The emergency cash-and-carry run is a manager off the floor for ninety minutes during prep. The small order under the minimum is the \$35 fee from G.3, and the \$1,820 a year behind it. The substitution made in a hurry is a plate that does not match the card.

The honest framing: turnover is a diagnostic, not a target. A number outside the range is a question — "why is this so slow?" or "why is this so fast?" — and the answer is where the management happens. An operator who manages to a turnover target will either strangle their inventory or bloat it, because the number knows nothing about your menu.

🤝 Hospitality

What running out actually costs, in the currency that matters.

A guest orders the Hearth Chicken. It is the dish their friend told them about. It is the reason they picked this restaurant on a Tuesday when there were eleven other options within a ten-minute walk. The server comes back and says they are out.

The financial cost of that moment is small and calculable — one \$20.48 contribution margin, less whatever they ordered instead.

The real cost is the second visit that does not happen, and Chapter 1 already did that arithmetic: a first visit barely covers the cost of acquiring it, and the second visit is where the business actually lives. You did not lose \$20. You lost a customer relationship you had already paid for.

This is why par levels and safety stock are a hospitality decision as much as a cost decision. An operator who tightens every par to the last unit in pursuit of a beautiful turnover number is optimizing a metric against the revenue model. The correct posture is to be tight on the items nobody orders by name and generous on the ones that bring people through the door.

G.13 Waste tracking

The log

The categories are what make a waste log useful. Without them, "waste" is a number you cannot act on.

The blank waste log

Why the log is worth more than its data

Here is the thing about a waste log that surprises people the first time they run one properly.

The data is useful. The visibility is transformative.

A cook who has to write down "3 qt demi — over-production — prepped to the old prep list" is performing an act of attention: stopping, naming the thing, assigning a cost to it, signing their name. That single act changes behavior more than any policy, because it makes an invisible loss visible to the person who created it, at the moment they created it.

This is the entire mechanism, and it is why the design details matter more than they seem:

• Put the log where the waste happens — at the trash can, the dish station, the walk-in. A log in the office does not get filled out.
• Make it fast. Nine columns, no narrative. A form that takes ninety seconds is a form nobody completes on a Saturday.
• Cost it. "3 quarts of demi" is abstract. "\$14.40" is not. Pre-print the costs of your twenty most-wasted items so the cook does not have to look anything up.
• Never punish an entry. This is the one that decides whether the system works at all. The moment a cook is disciplined for writing something down, the log stops being a record and becomes fiction — and you have lost both the data and the visibility.
• Read it out loud in pre-shift, weekly, without naming anyone. "We threw away \$310 last week; \$180 of it was over-produced sauces on Tuesday and Wednesday. Let's cut the Tuesday batch in half and see what happens." That is a team solving a problem, and it works.

The connection to G.11 is direct: a good waste log frequently explains most of a variance, which means it stops an investigation at step four and prevents it from arriving, wrongly, at step five.

G.14 Blank forms

Copy these into a spreadsheet and adapt them. Every one of them should carry a revision date, because a form nobody has revised is a form describing a restaurant that has changed without it.

Order guide

Sequence this by storage location, in the same order as the count sheet, so a single walk produces both.

Receiving log

The last two columns are the ones that make this form pay for itself. An unclaimed credit is free money left with a vendor who is not going to remind you.

Count sheet

Two count columns because some items live in two places. Sequenced by location, shelf by shelf.

Yield test worksheet

(The full version is in G.7. Compact form:)

| Item | Prep tested | Date | AP wt | AP \$/unit | AP total \$ | EP wt | Yield % | EP \$/unit** | Portion oz | **Portion \$ | |---|---|---|---|---|---|---|---|---|---|---| | | | | | \$ | \$ | | % | \$ | | \$ |

Reminder printed on the form: EP cost = AP cost ÷ yield. Divide. EP cost is always higher than AP cost.

Waste log

Categories: spoilage · over-production · error · trim · comp · staff meal

Par sheet

The revision-date column is not optional. A par sheet without one is an artifact, not a tool.

Vendor bid comparison

The row that decides the bid is cost per usable unit, not case price. The rows below it decide whether the winner of that row is actually the winner.

G.15 What inventory control cannot do

I want to close this appendix the way the book closes every method: by saying plainly what it cannot do, because a technique sold without its limits has already misled you.

It cannot make a bad concept work. If nobody wants what you are selling, a beautifully sequenced count sheet will tell you precisely, to the dollar, how much of it you are not selling. That is genuinely useful information and it will not save you. The count is a measuring instrument; it measures the business you have.

It cannot price your menu. A cost card tells you what a plate costs. What you charge is a market question, a positioning question, and a psychology question, and Chapters 11 and 12 spend real time on the fact that plate cost ÷ target percentage is where pricing starts and never where it ends. The most profitable item in this book runs a 29.4% food cost, which a naive target would call a failure.

It cannot tell you why. This is the limitation operators most often misunderstand. A variance is a magnitude, not a cause. The formula in G.11 will tell you, precisely, that you used \$1,100 more than you should have. It will not tell you whether that was a case that never arrived, a cook plating seven ounces instead of six, four days of over-produced sauce, or something worse. Only the diagnostic walk in G.11 finds that, and it is done with your feet and your eyes, in the building, not at a desk.

It cannot substitute for standards. A par level does not portion anything. A count sheet does not teach a cook to use the scale. Inventory control measures the gap between your standards and your execution — so if you have no written standards, the measurement has nothing to compare against and your theoretical usage is a guess with decimal places.

It cannot be done occasionally. A single count produces one number and no information. Inventory control is a time series: this period against the last four, this variance against the trend, this turnover against where it was in March. One count a quarter is an expensive way to feel diligent.

And it cannot be fully delegated, at least not at first. For your first six months, count the walk-in yourself — not because you distrust anyone, but because the count is the richest information stream in the building. Walking every shelf with a clipboard teaches you things no report will surface: which cook's station is organized, what is quietly rotting behind the milk crates, how much of last month's special is still on the freezer shelf, and whether your kitchen believes any of this matters.

Here is what it can do, and it is a great deal.

It converts an opinion into a number. It moves the discovery of a problem from month sixteen to week one. It makes waste visible to the person creating it. It gives you a defensible basis for a price, an honest food cost you can hand to a partner or an accountant, and a variance you can chase to a specific cause on a specific shift. It turns purchasing into a market without destroying the relationship. And it does all of that weekly, for about two hours of a trained manager's time.

Chapter 1's argument, in one sentence: the restaurants that die in year two and year three are almost never killed by their food, and almost always killed by three or four points of cost that ran quietly underneath a dining room that looked busy. This appendix is the instrument that finds those points. It is not the hand that fixes them.

That is still your job, on Monday morning, with a clipboard.