Chapter 13: Purchasing and Inventory: Vendors, Ordering, Receiving, Storage, and Controlling Waste

"The chef decides what the food costs. The receiver decides what you pay for it." — constructed; the sentence a controller says once and an operator remembers for a decade

Overview

Chapter 11 built the Hearth Chicken's cost card and put a number on the plate: chicken \$5.60, roots \$0.95, salsa verde \$1.05, butter and aromatics \$0.42, oil and seasoning \$0.18, garnish \$0.15 — \$8.35 of components, \$8.52 with the waste allowance, priced at \$29.00 for a 29.4% food cost and \$20.48 of contribution margin. Every one of those figures rests on a single assumption that Chapter 11 was entitled to make and this chapter is not: that a 3.5-pound air-chilled bird actually costs \$3.20 a pound, actually weighs 3.5 pounds, actually arrives at 38°F, and actually gets used before it goes bad.

That assumption is the whole chapter. Food cost is not set at the stove. It is set on a Thursday afternoon when somebody decides how many birds to order, at 6:52 on a Friday morning when somebody signs for what shows up, on a shelf in a walk-in where a container either has a date on it or doesn't, and at eleven o'clock on the last night of the period when somebody counts what is left. The cook is downstream of every one of those decisions, and a cook who portions perfectly out of a walk-in that was bought badly, received carelessly, and stored in the dark is producing a beautifully executed 34% food cost.

Chapter 11 also left you with an honest problem. It showed that Bellwether's costed menu, built line by line, lands somewhere between 30.2% and 30.7% against a 30.0% target. Half a point. It then took a \$2,728 variance apart into six named causes and put theft seventh in the investigation order — behind portioning, waste, purchasing, uncosted specials, and mix drift. This chapter is where most of those causes actually live, and where most of that half point is either won or given away.

What follows is the least glamorous material in the book and, dollar for hour, the most productive. There is no technique here that requires talent. There is a scale, a thermometer, a label gun, a clipboard, a written spec, and the discipline to use all six on a morning when you are behind. That is genuinely all of it.

In this chapter, you will learn to:

• Map the vendor landscape — broadline, specialty, direct, cash-and-carry — and read a prime-vendor agreement for what it costs as well as what it saves.
• Write a product specification tight enough that a delivery either meets it or measurably doesn't, and translate a spec failure into a change in plate cost.
• Build an order guide from purchasing programs rather than from dishes, and set par levels from a forecast, a delivery rhythm, and an explicit safety cushion.
• Run receiving as a seven-check procedure, and get a credit memo signed at the door rather than promised on the phone.
• Organize a walk-in so that storage order, labeling, and FIFO rotation serve the health inspector and the food-cost line at the same time — because they are the same discipline.
• Take a physical inventory that produces a number you can trust, apply the usage formula with transfers and credits, and compute inventory turnover and days on hand.
• Run a waste log, quantify what it finds, and be honest about what it misses.

Learning Paths

🏗️ Opening — §13.2 and §13.3 are your build. You will write specs and pars before you have a single sales history, from the forecast and the cross-utilization map, and the Business Plan checkpoint turns them into the section a lender and a landlord both read for evidence of systems. 📋 Managing — §13.4, §13.5, and §13.6. Receiving and the count are the two disciplines a manager personally owns and cannot delegate in the first six months. If you read three sections of this chapter, read those. 🍸 Beverage — the architecture here is identical at the bar and the leaks are not. Note transfers in §13.7 (the citrus and cream that cross between food and beverage and distort both numbers), and note in §13.7 that beverage inventory turns far more slowly than food. Chapter 15 owns the tenths method and the pour-cost variance. 🚚 Small Format — you have no walk-in, and a broadliner's minimum drop may exceed your weekly spend. §13.1's cash-and-carry paragraph and §13.3's pars-at-tiny-scale are yours, and §13.8's over-production discipline matters more to you than to anyone, because you cannot carry anything over.

Before the sections, a legend. This chapter uses four ASCII figures. Bars (█) are proportional within a single figure only and nothing is to scale. An arrow (→ or ▼) is the movement of product or of a decision. Boxes are physical locations in the building.

FIGURE 13.1 — Where a food dollar leaks between the order and the plate  [constructed teaching example]

  $1.00 committed on an order guide
     │
     ├─ ORDERING ....... over-ordering, ordering by feel, taking the rep's suggested
     │                   order, emergency runs to retail at retail prices
     ├─ RECEIVING ...... short counts, light catch-weights, unapproved substitutions,
     │                   price creep against the quote, credits promised and never taken
     ├─ STORAGE ........ temperature, rotation, unlabeled and undated containers,
     │                   product buried behind product, "we'll use it tomorrow"
     ├─ PREP ........... yield below the tested yield, over-production, sloppy trim
     ├─ SERVICE ........ over-portioning, re-fires, drops, mid-service 86 recovery
     ├─ THE CHECK ...... comps, uncosted specials, menu-mix drift
     │
     ▼
  what actually reaches a guest as sold product

  Chapters 11 and 12 own PREP, SERVICE, and THE CHECK. This chapter owns the first
  three — and the first three are where the money is easiest to keep, because nobody
  has to change how they cook in order to fix them.

That figure is the argument of the chapter. Three of the six leak points sit entirely upstream of the kitchen. They are fixed with procedure rather than skill, they can be fixed by one person in one week, and they are the three that almost nobody manages.

13.1 The vendor landscape: broadline, specialty, direct, and the prime-vendor bargain

A purveyor — used interchangeably with vendor or supplier — is anyone who sells you product. You will have between eight and twenty of them. The decision that matters is not which company to like; it is how to distribute your spend across categories of supplier, because each category buys you a different thing and charges you for it differently.

The four categories

A broadline distributor carries a very wide catalog — food, paper, chemicals, smallwares — across every category a restaurant uses, and delivers it on one truck against one invoice with credit terms. The national broadliners are large public companies; there are also strong regional broadliners in most markets. What you buy from a broadliner is consolidation: one order, one delivery window, one account to reconcile, one payables relationship. What you give up is depth and specificity. A broadliner's produce is a produce line; a produce specialist's produce is a business.

A specialty distributor goes deep in one category — produce, seafood, meat and poultry, bread, coffee, cheese, ethnic and imported goods. They are usually smaller, often better, frequently more expensive per unit, and much more useful when you need to talk to somebody who knows what an air-chilled bird is and why you care. Specialty houses are where a chef-driven menu actually gets sourced.

Direct purchasing means buying from the producer: a farm, a rancher, a fisherman's cooperative, a mill, a roaster. Direct is where the story on your menu comes from, and it is operationally the most demanding — irregular delivery, no credit terms sometimes, seasonal availability, no substitutions because there is nothing to substitute. Direct relationships are worth having and they are not worth having for everything.

Cash-and-carry — a warehouse club or a restaurant supply store you drive to — is not a supplier. It is an emergency room. Everything you buy there costs more, you pay for it with somebody's labor hour plus mileage, and the trip itself is a signal that a par level failed. Track cash-and-carry spend as its own line and treat a rising number as a purchasing-system defect, not a purchasing strategy.

⚠️ Where the Money Leaks

Two words on an invoice that should stop you: "market price."

Produce, seafood, and some proteins are frequently sold at "market" — meaning the price is set the week you order and printed on the invoice after it arrives. That is legitimate. What is not legitimate is your not knowing what it was until you pay the bill.

Two specific and very common leaks:

The unquoted market line. Nobody asked what carrots cost this week. The invoice says \$27.50 a case; last week it said \$22.50. Nothing is wrong — carrots moved — but you found out eleven days later when the bookkeeper coded it, and you have already sold the roasted roots at a price built on \$22.50. The fix costs nothing: get the week's market prices in writing before you place the order, on your top ten market lines, every week. A text message is writing.

The rep who writes your order. Every sales representative will happily build your order from your usage history, and many of them are excellent at it. But a suggested order is generated by somebody whose compensation moves with what you buy, and the suggestion is systematically one case heavy on the slow-moving items. Use the rep's expertise on product, availability, and substitution. Do not let the rep set the quantity. The quantity comes off your par sheet, which comes off your forecast, which is §13.3.

Assume, conservatively, that between one and two percent of what you are invoiced for either does not arrive, arrives out of spec, or is priced differently from what you were told. On Bellwether's planned \$334,800 of annual food purchases and \$95,480 of beverage, the midpoint of that range is \$6,454 a year — about 0.45 points of food cost on the food half alone. That is the prize for a clipboard.

The prime-vendor bargain

A prime-vendor agreement is a contract in which you commit a defined share of your purchases — commonly a large majority of what that distributor could supply — to one broadliner, in exchange for better pricing, guaranteed service levels, a dedicated sales representative, sometimes online ordering and inventory tools, and often a contracted price on a defined list of items. Pricing under these agreements is frequently cost-plus: the distributor charges its landed cost plus an agreed markup, either a fixed dollar amount per case or a percentage.

These agreements are real, they are common, and the terms vary enormously and are negotiable. Do not accept anybody's characterization of what is "standard," including mine. What follows is the structure of the bargain and the questions that determine whether a specific one is good.

What to get in writing, every time:

The honest arithmetic. Bellwether plans \$334,800 of food purchases a year. Suppose roughly \$185,000 of that is genuinely undifferentiated — dry goods, dairy commodities, oil, flour, staples that a broadliner and a specialist would deliver as the same thing. A three percent improvement in landed cost on that portion is \$5,550 a year**, which on \$1,116,000 of food sales is 0.50 points of food cost**. That is a real number and it is worth negotiating for.

Now the other side. Suppose the commitment forces \$60,000 of protein and produce through the broadliner that a specialist would supply eight percent cheaper, or at the same price but noticeably better. That is \$4,800 — and it eats most of the gain, before you count what happens to the menu when the trout is not the trout you wrote the dish for.

The conclusion, which is the transferable one: a prime-vendor agreement is usually worth real money on the undifferentiated two-thirds of your order guide and usually a bad trade on the lines that define your food. Commit the commodity. Shop the identity. Negotiate the commitment percentage down to the number that covers the commodity and no further, and be prepared to hear no.

One piece of public history worth knowing

In 2015 the Federal Trade Commission sued to block Sysco's proposed acquisition of US Foods — the two largest broadline foodservice distributors in the United States. The companies had proposed divesting a set of distribution centers to Performance Food Group as a remedy. A federal district court granted the FTC a preliminary injunction in June of that year, and the companies abandoned the transaction shortly afterward. US Foods went public the following year.

Whatever you think of the outcome, the case is worth knowing for one structural reason. A regulator went to court on the theory that an independent restaurant's alternatives to a national broadliner are not as good as they look on paper — that regional distributors and self-distribution are not close substitutes for a small account buying across every category. That is roughly the position you are negotiating from. It does not mean you have no leverage. It means your leverage is local, it is about your volume and your payment behavior and your delivery convenience, and it is exercised with a second approved supplier in your phone rather than with a speech about competition. Chapter 13's case study takes this apart properly.

👨‍🍳 On the Line

What actually builds vendor leverage, in order of effectiveness.

1. Paying on time, every time. Nothing else is close. A distributor's credit department and its pricing department talk to each other. An account that pays at fourteen days for a year gets things an account at fifty-one days will never be offered. This is also the single cheapest lever available to an operator who has done the working-capital work in Chapter 33.

2. A second approved source for every line that matters. Chapter 10 was explicit about this: where one delivery reaches four or more menu items, a single source is a single point of failure. The second source does not need to get much volume. It needs to exist, to have your spec on file, and to have delivered to you at least once in the last quarter so that it is not a cold call at 7:00 a.m. on a Friday.

3. Committed volume, stated as a number. "We do about fifty birds a week" is a conversation. "We will commit to fifty-two whole birds a week, two drops, on this spec, for twenty-six weeks" is a negotiation. The second version is worth something to a supplier because it lets them plan; the first is worth nothing because it is a mood.

4. Being easy to deliver to. A restaurant that receives at a scheduled window, has a person at the door, opens the cases, and settles exceptions in four minutes is a stop a driver wants. This is soft and it is not nothing.

5. Talking to the district manager once a year. Not to complain. To be a name.

What does not build leverage: threatening to leave over one bad delivery, arguing with the driver, and asking for a discount without offering a commitment in exchange.

13.2 Specs: writing them, holding suppliers to them, and what happens when you don't

A product specification — a spec — is a written description of exactly what you are buying, precise enough that a delivery either meets it or measurably does not. It is the document that turns "chicken" into a purchase you can hold somebody to.

Most independent restaurants have no written specs at all. What they have is a chef who knows what they want and a distributor who mostly sends it, which works until the chef is off, the regular driver is off, the product is short, or the salesperson changes. Then "chicken" arrives, and it is chicken, and nobody can say it isn't.

What a spec contains

Bellwether writes twelve specs, not fifty-four. Write a spec for anything that is expensive, variable, or load-bearing on the menu, and buy the rest on catalog description. The twelve are the anchor lines of the eight programs: the chicken, the hanger, the ribeye, the grind blend, the trout, the pork rack, the butter, the cream, the eggs, the bread flour, the carrots, and the parsley-and-caper box.

👨‍🍳 On the Line

Writing a spec is a twenty-minute job that nobody does, and here is how to make yourself do it.

Do not sit down to write twelve specs. You will write two, get bored, and stop. Instead:

Write the spec the first time the product is wrong. Something arrives that isn't right — the birds are heavy, the trout are small, the butter is a different brand and behaves differently in the sauce. That is the moment. You are annoyed, you are specific, and you know exactly what you wanted. Write it down then, in the walk-in, on your phone, in eight lines. Clean it up later.

Write it as an acceptance test, not a description. Not "good quality carrots." Rather: "jumbo, 25 lb case, no more than 5% with green shoulder, no soft or split, no case over 24 hours out of refrigeration." A description is an opinion. An acceptance test is something a nineteen-year-old receiver can apply at 6:52 in the morning without calling you.

Then do the thing that makes it real: send it to the supplier. A spec in your binder is a diary entry. A spec on file with the salesperson, acknowledged by email, is the document you point at when the substitution shows up. It costs one email.

And put the spec number on the order guide line. [P-01], [T-01], [D-02]. When the receiver is looking at a case, the invoice line tells them which sheet to check.

What an out-of-spec delivery actually costs

This is where a spec stops being paperwork.

🧮 Run the Numbers

The heavy bird.

The spec is 3.25–3.75 lb, target 3.50. The Friday delivery of 28 birds weighs 109.20 lb — an average of 3.90 lb, comfortably outside the spec ceiling. Poultry is catch-weight, billed on delivered weight at \$3.20/lb.

What you were invoiced: 109.20 lb × \$3.20 = **\$349.44 What 28 birds at spec would have cost: 28 × 3.50 lb = 98.00 lb × \$3.20 = **\$313.60 Overpayment on one delivery: \$35.84

That is annoying. The plate is worse. A 3.90 lb bird halves to 1.95 lb, and a cook who plates what the bird gives them is putting 1.95 lb × \$3.20 = **\$6.24** of chicken on a plate costed at \$5.60.

	On spec	Heavy birds, plated as they come
Chicken	\$5.60 | \$6.24
Roots	\$0.95 | \$0.95
Salsa verde	\$1.05 | \$1.05
Butter and aromatics	\$0.42 | \$0.42
Oil, salt, misc	\$0.18 | \$0.18
Garnish	\$0.15 | \$0.15
Components	\$8.35 | \$8.99
Plate cost (+2% waste allowance)	\$8.52** | **\$9.17
Menu price	\$29.00 | \$29.00
Food cost %	29.4%	31.6%
Contribution margin	\$20.48** | **\$19.83

Sixty-five cents a plate. Bellwether forecasts 96 Hearth Chickens a week; if heavy birds ran all year uncorrected that is 4,992 plates × \$0.65 = **\$3,244.80**, which is 0.29 points of food cost — more than half of the gap Chapter 11 could not close on paper, produced entirely by a weight range nobody enforced.

Now the good answer, which is not "reject the delivery." It is 6:52 a.m. on a Friday with 118 covers on the books. You cannot run Friday and Saturday without chicken. So you do three things: you take the \$35.84 as a credit at the door (§13.4); you send the spec violation to the salesperson in writing before 8:00 a.m.; and — this is the part that saves the plate — you portion to spec anyway. Trim each half back to 1.75 lb. Across 56 halves that is 11.2 lb of chicken meat you did not plate, at \$3.20 = **\$35.84** of product, which does not go in a bin. It goes into the hash and the stock, both of which the cross-utilization map already has outlets for.

The credit you could not get from the distributor, you take out of the bird. That is what a cross-utilized purchasing program is for.

The failure modes of specs

Every method in this book states its limits. Specs have four.

A spec you do not check is worse than no spec, because it produces the confidence of a control without the control. If nobody weighs the birds, the weight range is decoration.

Specs age. Pack sizes change, brands get discontinued, a supplier's "jumbo" is not what it was two years ago. Review the twelve once a year, and immediately whenever a plate cost moves for no reason you can name.

Over-specification costs money. A spec that only one supplier in the market can meet has handed that supplier pricing power. Write specs tight on the dimensions that affect the plate — weight, grade, fabrication, temperature — and loose on the ones that do not, such as brand, wherever "or equal" is genuinely acceptable.

A spec cannot make a market produce something. In February, the root box is what February has. The spec's job then is not to conjure carrots; it is to make sure that when the substitution arrives, somebody makes a decision about it rather than discovering it in a sauce.

13.3 Par levels and the order guide: ordering to a forecast, not a feeling

The order guide

An order guide is the standing list of every item you buy, in the units you buy it in, from the supplier you buy it from, with a place to write what is on hand and what to order. It is the document you order from. If you order by walking through the walk-in with your phone, you do not have an order guide; you have a memory, and memory over-orders the things it can see and forgets the things behind them.

Chapter 10 gave us the map to build it from. Bellwether's menu — twenty-two dinner food items and twelve brunch lines, thirty-four menu lines in all — is fed by eight purchasing programs across roughly fifty-four purchased lines. The order guide is written from the programs, not from the dishes.

Figure 13.2 — Bellwether's food order guide, summarized by program (the Bellwether plan; weekly figures are the plan's steady-state averages at \$1,550,000 of annual revenue, 72% food, 30% food cost)

Read the totals column before anything else. Thirty-eight lines out of fifty-four — seventy percent of the guide — are two-thirds of the money, and they are the eight programs Chapter 10 built. That is what cross-utilization does to a purchasing system: it concentrates your attention. You do not have fifty-four purchasing relationships to manage well. You have eight, plus a produce order.

Note also what \$6,438 a week is. Annualized, it is \$334,800, which is 30.0% of \$1,116,000 of planned food sales — the food-cost target, expressed as a purchasing budget. An order guide is a budget you fill out in cases.

Par levels

A par level is the quantity of an item you want on hand at a defined point in the ordering cycle — enough to cover forecast usage until the next delivery lands, plus a deliberate cushion for the forecast being wrong.

$$\text{Par} = \big(\text{forecast usage per service} \times \text{services until the next delivery}\big) + \text{safety stock}$$

Three things about that formula deserve emphasis before we use it.

"Services," not "days." Bellwether is closed Monday and serves no dinner on Sunday. A par built on calendar days will over-order every week and the over-order will sit in the walk-in for two days at whatever temperature the walk-in happens to be running.

Coverage includes lead time. If you order Thursday for Friday delivery, the Friday par must cover Friday and Saturday. If your fish supplier needs 48 hours, the coverage window is longer and the safety stock has to absorb two days of being wrong instead of one.

Safety stock is a decision, not a rounding error. It is the answer to "what does it cost me to run out?" For a garnish, running out is an inconvenience. For the signature dish on a Saturday at 8:40 p.m., running out is a guest who ordered the reason they came. Put the cushion where the volatility and the consequence are, not evenly across the guide.

Building the poultry par

This is the chapter's worked example, and it is worth doing slowly because it is the model for every other line.

Chapter 10 sized the poultry program at fifty-two whole 3.5 lb air-chilled birds a week on two deliveries — one hundred and four halves — and split it evenly, twenty-six and twenty-six, because it was sizing a program. Chapter 13 writes the order, and the order is not even, because the week is not even.

Start with the forecast. The plan's 475 dinner covers a week are not 95 a night; they are the shape of a Midwestern neighborhood restaurant's week:

Chapter 10 sized the program at "roughly one Hearth Chicken for every four entrées." Held against an actual forecast, "roughly one in four" is a range: at one in four exactly you sell 107 halves a week, and at the conservative end you sell 96. A par level is where you decide which end of a range you are buying, and the two ends produce very different weeks. At 107, the program is short and Sunday brunch has no hash. At 96, eight halves have to go somewhere.

Bellwether buys the conservative end and gives the surplus a designed destination.

🧮 Run the Numbers

The poultry par, derived.

Deliveries land Tuesday and Friday, before 7:30 a.m., ordered the previous afternoon.

The Tuesday delivery covers Tuesday, Wednesday, and Thursday dinner. Forecast usage: 12 + 15 + 18 = 45 halves Safety stock: 3 halves — one cushion plate per service, on the three quiet nights Par: 48 halves = 24 whole birds

The Friday delivery covers Friday and Saturday dinner. Forecast usage: 24 + 27 = 51 halves Safety stock: 5 halves — the weekend carries more cushion, because an 86 on the signature dish at 8:40 on a Saturday costs a great deal more than one at 7:15 on a Wednesday Par: 56 halves = 28 whole birds

Weekly order: 24 + 28 = 52 birds. The program Chapter 10 sized is unchanged. The split is not.

What it costs. 52 birds × 3.5 lb = 182 lb × \$3.20/lb = **\$582.40 a week, or \$30,284.80 a year** — 9.0% of the food order guide, from one line.

Tuesday's drop: 24 × 3.5 = 84 lb × \$3.20 = **\$268.80 Friday's drop: 28 × 3.5 = 98 lb × \$3.20 = **\$313.60 Sum: \$582.40 ✓

Now watch the week close, which is the test of any par:

FIGURE 13.3 — One week of poultry, in halves on the shelf        [the Bellwether plan]

                              halves on hand after each event
  TUE 7:00a   receive 24 birds   +48   ████████████████████████████████  48
  TUE close   sold 12            -12   ████████████████████████          36
  WED close   sold 15            -15   ██████████████                    21
  THU close   sold 18            -18   ██                                 3
  FRI 7:00a   receive 28 birds   +56   ███████████████████████████████████████  59
  FRI close   sold 24            -24   ███████████████████████           35
  SAT close   sold 27            -27   █████                              8
  SAT night   the 8 remaining halves are roasted in the cooling hearth, picked, and held
  SUN brunch  chicken hash        -8   │                                   0
  MON         closed                   │                                   0
  TUE 7:00a   receive 24 birds   +48   ████████████████████████████████  48   ← identical
                                                                                 to last
                                                                                 Tuesday
  Bars proportional within this figure only. Nothing is to scale.

The week returns to exactly where it started. That is what a par level is supposed to do, and it is the only proof that a par is right: run it forward through a full cycle and see whether you land on the same shelf.

Two things in that trace are worth naming.

Poultry is a standing order, not an order-to-par. For most lines the ordering rule is order quantity = par − on hand. For the chicken, Bellwether places a fixed order of 24 and 28 every week regardless of what is on the shelf, for three reasons: the supplier needs a committed number to reserve air-chilled birds, a standing order is the volume commitment that earns anything in a negotiation (§13.1), and — decisively — the overage has a guaranteed downstream use. The eight halves left at Saturday close are not a forecasting failure. They are Sunday's hash, and Chapter 10 put them on the menu on purpose.

Cross-utilization is what converts safety stock from a spoilage risk into a menu item. On a line with no second outlet, safety stock is money you might throw away, and the rational operator keeps it thin and runs out sometimes. On a cross-utilized line, safety stock is money that gets sold at full margin two days later. That is not a small difference. It is the reason the poultry program can carry an eight-half cushion into a Saturday night without anybody being nervous, and the reason Bellwether can hold the last seating on a Saturday without either 86ing the chicken or throwing money in a bin.

Sunday's yield, for completeness: 8 halves, roasted and picked, give roughly 9 ounces of meat each — about 72 ounces, or 18 four-ounce hash portions at \$16 on a brunch that does about 110 covers. Say \$288 of revenue from product that a restaurant without a map would have discarded.

The par sheet

Here is the artifact, for a representative slice of the guide.

Figure 13.4 — Par sheet, extract (the Bellwether plan)

Four things this sheet teaches that the formula does not.

The order unit rounds your par. Carrots come in 25 lb cases. A par of 41 lb is two cases, which is 50 lb, which is nine pounds of over-order every cycle. That is not a failure — it is the cost of the case pack, it is knowable in advance, and the answer is either to accept it and make sure the ninth pound has a use, or to negotiate a split case at a higher per-pound price and do the arithmetic on which is cheaper.

Safety stock varies inversely with spoilage risk. Flour carries a full bag of cushion because a bag of flour costs \$26.40 and does not go bad. Trout carries two fish because trout has no second outlet on the map — the frames get discarded, Chapter 10 said so — and a whole fish nobody sells is a whole fish in a bin.

High-value slow movers get counted, not cushioned. The ribeye is the plan's most expensive line per unit and its slowest mover. Six units of cushion is a lot of money sitting still, and it is correct anyway, because a restaurant that 86s its most expensive entrée on a Saturday has trained a table to order down.

Some things have no par at all. The salsa verde herbs are made twice a week to order; there is no cushion, because the cushion is the delivery rhythm.

⚠️ Where the Money Leaks

Ordering by feel, quantified.

An operator without pars orders the way everyone orders without pars: they walk the walk-in, look at what seems low, and add a little because running out is embarrassing and over-ordering is invisible.

"A little" is remarkably consistent — call it a ten percent standing over-order across the perishable half of the guide. Bellwether's perishable lines run about \$4,600 of the \$6,438 weekly guide. Ten percent is \$460 a week of product bought early.

Most of it gets used, and that is why this leak is invisible: nothing is thrown away, so nothing looks wrong. What actually happens is two things. First, average inventory rises — that \$460 sits on a shelf as cash that is not in the account, week after week, which is Chapter 33's problem and it is real. Second, a fraction of it — call it fifteen percent of the excess — ages past use. \$69 a week is **\$3,588 a year, or 0.32 points of food cost**, produced by nothing but a reluctance to write a number down before walking into the box.

The countermeasure costs one sheet of paper: fill in the on-hand column before you look at the order column, and let subtraction do the deciding.

13.4 Receiving: the single highest-leverage twenty minutes of the day

Everything in this chapter has been preparation for this section. Receiving is the only moment in the entire cycle when you can still say no.

Once a case is inside and the driver is gone, your options collapse. The product is yours, the price is yours, the substitution is now a menu decision, and the short case is a phone call you will make three times and probably lose. At the door, for about four minutes per delivery, you hold every card.

⚠️ Where the Money Leaks

The twenty minutes, and what they are worth.

Bellwether takes five delivery mornings a week. Doing receiving properly — a person at the door, cases opened, catch-weights on the scale, a thermometer in the cold load, the invoice checked against the order guide and the quoted price, exceptions written and signed — costs about twenty minutes a morning. One hundred minutes a week. 86.7 hours a year. At a fully loaded manager rate of \$22 an hour, that is **\$1,907 of labor.**

Against it: \$334,800 of food and \$95,480 of beverage purchased annually, of which a conservative one and a half percent is short, out of spec, or mispriced — \$6,454.

Net: \$4,547 recovered for \$1,907 spent. A 3.4× return, and it recurs every year.

There is nothing else in this book with that return on that little skill. Not menu engineering, not a price increase, not a labor initiative. Twenty minutes with a scale.

And notice where it lands on the P&L. The \$5,022 that is the food share of that recovery is 0.45 points of food cost — most of the half point Chapter 11 could not close between the honest build at 30.2–30.7% and the 30.0% target. Chapter 11 said the menu's arithmetic was tight but achievable. This is how it becomes achievable.

What makes people skip it: the truck comes at 6:45 and prep starts at 8:00 and there is bread to shape. Every operator who has skipped receiving has skipped it for a good reason. That is why it has to be on the schedule as a position, not as a hope — a named person, a named window, on the printed schedule, the way the fish station is on the schedule.

The seven checks

FIGURE 13.5 — The receiving lane: seven checks in twenty minutes   [constructed teaching example]

  TRUCK ARRIVES
      │
      ▼
  [1] TIME ─── Is a trained receiver here? Is it mid-service? ─── no ──► reschedule the
      │ yes                                                              delivery window;
      ▼                                                                  this is negotiable
  [2] TEMP ─── probe or IR the cold and frozen loads before ───── out ──► REJECT the
      │ in spec   anything comes off the truck                            affected lines
      ▼
  [3] COUNT ── open the case. count the pieces. ───────────────── short ─► CREDIT MEMO,
      │ matches   a sealed case is an unverified case                      at the door
      ▼
  [4] WEIGH ── every catch-weight line on the scale ──────────── light ─► CREDIT MEMO,
      │ matches                                                            at the door
      ▼
  [5] SPEC ─── grade, size, pack, brand, fabrication, ─────────── off ──► REJECT, or
      │ matches   pack date, country of origin                             ACCEPT UNDER
      ▼                                                                    PROTEST + credit
  [6] PRICE ── invoice line vs. the quote or contract price ───── off ──► note on the
      │ matches                                                            invoice, call
      ▼                                                                    the rep before 8
  [7] SIGN ─── the driver signs every exception. You keep a copy with the exception on it.
      │
      ▼
  LABEL AND DATE ──► ASSIGNED SHELF ──► invoice into the book the same day

Check 1, time. Deliveries arrive when you tell suppliers they may. A restaurant that accepts whenever the route happens to reach it has surrendered the control. Set a window — Bellwether's is 6:30 to 8:30 a.m. — put it in the vendor file, and enforce it. This is negotiable with almost every supplier and almost nobody negotiates it.

Check 2, temperature. Probe the cold load before it comes off the truck. Refrigerated product at 41°F or below, frozen product frozen solid with no evidence of thaw-and-refreeze. This is a food safety control and Chapter 25 owns the reasoning; it is also a cost control, because product received warm has already spent part of its shelf life and you will pay for that on a Sunday. Product received out of temperature is rejected, not discounted. There is no price at which it is a good buy.

Check 3, count. Open the case. This is the single most-skipped step in American restaurants, because a sealed case looks like a fact. Count the pieces against the invoice and against the order guide — those are two different comparisons and both matter, because a case can be full and still not be what you ordered.

Check 4, weight. Anything sold catch-weight — poultry, whole fish, primals, cheese — goes on a scale. Buy a floor scale or a good bench scale; it will pay for itself in a month. This is where the heavy-bird arithmetic in §13.2 either gets caught on Friday morning or gets discovered in a variance report in five weeks.

Check 5, spec. The receiver holds the spec sheet, not a memory of a conversation. Grade, size, pack, fabrication, pack date. This is why the spec number lives on the order-guide line.

Check 6, price. Compare the invoice line to what you were quoted. Price creep on market lines is the most common invoice defect there is and the least likely to be challenged, because by the time anybody looks the product is eaten.

Check 7, signature. Every exception is written on the invoice — both copies — and the driver signs it. A driver who will not sign is a driver whose company will not honor the credit.

🧾 Read the Numbers

```text FIGURE 13.6 — "The Friday invoice" [constructed teaching example] THE ARTIFACT One protein-and-dairy delivery invoice, received Friday 6:52 a.m., as it appeared on the clipboard before anything was signed. The distributor is deliberately unnamed. THE CONTEXT Bellwether, first winter, week 34 of operation. 118 covers on the books for Friday dinner, 133 for Saturday. The sous is receiving; the chef is in at 8:00. Ordered Thursday at 3:10 p.m. off the par sheet.

INVOICE 44118                                           TERMS: NET 21
LN  ITEM (spec)                       ORD   SHP   BILLED     PRICE      EXTENDED
 1  Chicken, whole, AIR-CHILLED,      28ea  28ea  109.20 lb  $3.20/lb   $  349.44
    3.25-3.75 lb, giblets in [P-01]
    *** shipped IMMERSION-CHILLED, label "4% retained water", 3.9 lb avg ***
 2  Carrots, jumbo, 25 lb cs [R-02]    4cs   3cs    4 cs      $22.50/cs $   90.00
 3  Butter, unsalted 1 lb, 36/cs[D-01] 1cs   1cs    1 cs     $138.60/cs $  138.60
 4  Cream, heavy 40%, qt [D-02]       12qt  12qt   12 qt      $4.95/qt  $   59.40
 5  Parsley/caper box [H-01]           2bx   2bx    2 bx     $34.00/bx  $   68.00
 6  Trout, whole, 1.25-1.5 lb [T-01]  34ea  34ea   44.20 lb   $9.85/lb  $  435.37
 7  Flour, bread, 50 lb [B-08]         3bg   3bg    3 bg     $26.40/bg  $   79.20
                                                   SUBTOTAL             $1,220.01
                                                   FUEL SURCHARGE       $   18.00
                                                   TOTAL DUE            $1,238.01

WHAT IT SHOWS Two defects, both catchable in under four minutes and neither visible after the driver leaves. (a) SUBSTITUTION, line 1. The spec says air-chilled and forbids substitution without approval. What arrived is immersion- chilled with 4% declared retained water, at 3.90 lb average — above the 3.75 lb spec ceiling. Billed 109.20 lb; at spec weight the line is 98.00 lb x $3.20 = $313.60. Overcharge against spec: $35.84. Of the delivered weight, roughly 4.37 lb is declared water, or about $13.98 of the invoice. (b) SHORT DELIVERY, line 2. Four cases of carrots invoiced; three cases on the dolly. $22.50 billed and not delivered — and the weekend's root par is now one case short. Total recoverable at the door: $58.34. WHAT IT DOESN'T It does not show whether the substitution was a supply failure or a fill decision. It does not show what the water-chilled skin will do over the hearth, which is a guest-experience cost that never appears on any invoice. It does not tell you whether line 6's trout met the size spec, because nobody wrote a weight range per fish. And it says nothing about the $18.00 fuel surcharge, which is a contract question (13.1), not a receiving question. THE DECISION Accept line 1 under protest -- you cannot run Friday and Saturday without chicken -- write the exception on both copies, get the driver's signature, and demand a $35.84 credit plus a written spec restoration for Tuesday. Then portion the halves back to 1.75 lb and route the 11.2 lb of trim to the hash and the stock. Take a signed credit memo for $22.50 on line 2 at the door. Call the salesperson before 8:00 a.m. -- not to complain, to restate the spec and ask what Tuesday will be. THE LESSON At the door you have three moves -- reject, accept-under-protest with a written credit, or accept -- and only one of them is "argue." An invoice defect caught at 6:52 costs four minutes. The same defect caught in a variance report costs five weeks and cannot be recovered at all. ```

Invoice reconciliation and the credit memo

Invoice reconciliation is the three-way match: what you ordered (the order guide), what the invoice says, and what physically arrived. Most restaurants do a one-way check — they look at the invoice — which catches nothing, because the invoice is the supplier's account of events.

A credit memo is the supplier's written acknowledgment that they owe you money or product for something short, rejected, or wrong. Four rules, and they are absolute:

1. Get it at the door, in writing, signed by the driver. A verbal "I'll take care of it" is worth nothing and everyone in the transaction knows it.
2. Keep the credit memo with the invoice. They travel as a pair to the bookkeeper.
3. Verify that the credit actually appears on the next statement. Credits issued and never applied are among the most common quiet losses in restaurant purchasing, and they are invisible because nobody is looking for money they already decided they were owed.
4. Log it. A one-line credit log — date, vendor, amount, reason, applied Y/N — takes ten seconds an entry and is the only way you will ever be able to say "your service level on my account has cost me \$1,400 this year" with a straight face in a negotiation.

If a single invoice defect of \$58.34 happens once a week and is never caught, that is **\$3,033.68 a year — 0.27 points of food cost** from one clipboard nobody was holding.

🔍 Check Your Understanding

1. Why is "open the case" a financial control and not merely a quality control?
2. A delivery of hanger steak arrives at 47°F. The supplier offers a 20% discount to keep it. What is the correct answer and why is the discount irrelevant?
3. Your spec says 3.25–3.75 lb birds. A delivery averages 3.10 lb. Is this good news? Compute the effect on the Hearth Chicken's plate cost and say what it does to the guest.

(1: Because a sealed case is an unverified assertion about count and contents; the invoice bills you for the assertion. 2: Reject the affected lines. Temperature is not a quality dimension you can discount your way past — it is a food-safety control, and there is no price at which product received above 41°F is a good buy. Chapter 25 explains the pathogen reasoning. 3: No. A 3.10 lb bird halves to 1.55 lb; 1.55 × \$3.20 = \$4.96 of chicken against a costed \$5.60, so components fall to \$7.71 and the plate to \$7.86 — a 27.1% food cost that looks like a win. It is not: the guest gets an obviously smaller portion of a \$29 signature dish, which is a price increase they did not agree to. Under-spec is a spec violation in exactly the same way over-spec is.)

13.5 Storage, rotation, and FIFO — and the walk-in as a financial statement

Chapter 11 handed this chapter the physical side of inventory control — receiving, storage, and rotation — and this is where it gets paid off. The argument of this section is a single sentence: the walk-in is a financial statement, and it is the only one you can read without a computer.

Stand in the door of any restaurant's walk-in for ninety seconds and you can tell, with something close to certainty, what its food cost is doing. Not because you can see the number, but because everything that produces the number is physically present: whether product is labeled, whether it is dated, whether the oldest is in front, whether raw is below ready-to-eat, whether there is a thermometer on the wall and what it says, whether things are on the floor, whether anything is buried behind anything.

The surprise inspection

A Tuesday in Bellwether's first winter. Ten-forty in the morning, mid-prep. The health inspector walks in unannounced — which is how they come — and by 11:25 has written five things:

• the walk-in at 46°F
• raw chicken stored above ready-to-eat greens
• a sanitizer bucket at 50 ppm
• no thermometer in the reach-in
• one employee without a food-handler card

The visit ends with a re-inspection in ten days, not a closure. That is the good outcome, and it is worth saying plainly that it was not guaranteed: a sewage backup or a loss of hot water would have closed the doors by noon, and no amount of good intentions in the walk-in would have changed it.

Chapter 25 owns this inspection. It owns HACCP, the Food Code, the pathogens, critical versus non-critical violations, the scoring, the re-inspection, and what actually gets a restaurant closed. Do not look here for the food-safety law; look here for something Chapter 25 will not spend time on, because it is a cost book's job rather than a safety book's:

Every one of the first four findings is simultaneously a safety violation and a food-cost problem, and they are produced by the same neglect.

That is the connection this chapter exists to make. The same twenty minutes of neglect writes the violation and the variance. An operator who fixes the walk-in because an inspector told them to has also, without meaning to, fixed a food-cost problem — and an operator who fixes it before an inspector arrives gets the money and never gets the report.

🧮 Run the Numbers

What 46°F costs, before anyone from the health department is involved.

Bellwether's average food inventory runs about \$6,025 (§13.7). Roughly 70% of it — call it \$4,200 — is in the walk-in at any moment: proteins, dairy, cut produce, prepped sauces.

A box running at 46°F rather than 38°F does not spoil everything; it shortens shelf life across the board. Suppose the effect is an additional two percent of walk-in inventory lost each week to accelerated spoilage, trim-back, and product that has to be re-made a day early.

\$4,200 × 2% = **\$84 a week = \$4,368 a year On \$1,116,000 of planned food sales, that is 0.39 points of food cost.**

Hold that against Chapter 11's finding that Bellwether's honest costed menu lands at 30.2–30.7% against a 30.0% target. A warm walk-in is, by itself, roughly the whole gap — and unlike the gap in the cost cards, you do not close it with a price increase or a re-costed recipe. You close it with a service call.

What the fix costs. A walk-in running warm is usually one of four things, and three of them are maintenance rather than capital: a failed door gasket, condenser coils packed with grease and dust, airflow blocked by product stacked against the evaporator, or low refrigerant. Call a coil cleaning and a gasket \$400 and an afternoon.

\$400 ÷ \$84 a week = payback in under five weeks, and then it is \$84 a week of pure margin forever. There is no menu change in this book with that return.

The limit of this arithmetic: two percent is an assumption, not a measurement. The honest version is that nobody can tell you what a warm box costs your restaurant without a waste log (§13.8) and a variance (Chapter 11). What is not an assumption is the direction and the order of magnitude.

Storage order and the walk-in map

FIGURE 13.7 — The walk-in, as shelved and as counted            [the Bellwether plan]

  ┌────────────────────────────────────────────────────────────────────────┐
  │ ● THERMOMETER — door end, eye height, read and logged a.m. and p.m.    │
  │   Target 36–38°F. Anything at 41°F or above is a work order, today.    │
  │                                                                        │
  │  ┌── A ─ DAIRY & EGGS ──────────┐      ┌── D ─ PREPPED / READY-TO-EAT ─┐│
  │  │  butter · cream · buttermilk │      │  salsa verde · dressings ·    ││
  │  │  ricotta · eggs              │      │  cures · stock · sauces       ││
  │  └──────────────────────────────┘      │  ── every container labeled   ││
  │                                        │     and dated, no exceptions  ││
  │  ┌── B ─ PRODUCE ───────────────┐      └───────────────────────────────┘│
  │  │  greens, herbs      [top]    │      ┌── E ─ RAW PROTEIN ────────────┐│
  │  │  cut vegetables     [mid]    │      │  [1] whole fish — trout   TOP ││
  │  │  root box       [floor rack] │      │  [2] whole cuts — beef, pork  ││
  │  └──────────────────────────────┘      │  [3] ground blend             ││
  │                                        │  [4] POULTRY           BOTTOM ││
  │  ┌── C ─ BEVERAGE / OVERFLOW ───┐      └───────────────────────────────┘│
  │  │  wine · beer · juice · dairy │                                       │
  │  │  for the bar                 │      COUNT ROUTE: A → B → C → D → E,  │
  │  └──────────────────────────────┘      always, in that order, every time │
  │                                                                        │
  │  FIFO: new product goes BEHIND. The oldest thing is always the thing   │
  │  closest to the aisle. Nothing on the floor. Nothing unlabeled.        │
  └────────────────────────────────────────────────────────────────────────┘

  Storage order within section E runs top-to-bottom by required minimum
  cooking temperature — ready-to-eat highest, poultry lowest — so that
  anything that drips, drips onto something that will be cooked hotter.
  Chapter 25 owns the pathogen reasoning and the temperatures.

Three design decisions in that map are worth calling out, because they each do double duty.

Sections are lettered and the count route is fixed. A, B, C, D, E, every single time. This is a food-cost decision disguised as tidiness: the count sheet in §13.6 is printed in that order, and a count sheet that matches the physical route is the difference between a 75-minute count and a two-hour count with three lines missed.

Raw protein is a single section, ordered by cook temperature, at the bottom. That is the Food Code principle and it is also the only arrangement in which a receiver who has never worked in a restaurant before can put the chicken away correctly on their second shift. Rules that require judgment get broken on busy mornings. Rules that are physical — chicken goes on the bottom shelf of section E — survive.

Prepped and ready-to-eat is its own section with an absolute labeling rule. This is where money actually disappears in most kitchens. Unlabeled prepped product is not inventory; it is a guess, and the guess is resolved by throwing it out. Chapter 14 builds the prep system that fills this section; this chapter's only demand is that everything in it carries three dates: made, and use-by, plus the initials of the person who made it.

FIFO — first in, first out — is the rotation discipline that makes all of it work: the oldest stock of any item is always used first, which in practice means new product is loaded behind old product, never in front of it. It is the simplest control in the building and the most consistently broken, for the most human reason: at 7:00 a.m. with a dolly of cases, putting the new box in front takes four seconds and putting it behind takes forty. Forty seconds, eight lines, ten deliveries a week is about an hour a week — call it \$22 of labor — against the alternative, which is that the back of every shelf becomes an archaeological site.

⚖️ Code and Compliance

What storage owes the law, and where the law lives.

The storage practices in this section are simultaneously cost controls and regulatory obligations. The regulatory framing, in the American context:

• Cold holding at or below 41°F, hot holding at or above 135°F, with the range between them being the temperature danger zone — Chapter 25 defines it and explains the pathogen behavior.
• Storage order by required cooking temperature, ready-to-eat above raw, poultry at the bottom.
• Date marking on prepared ready-to-eat foods held beyond 24 hours.
• Nothing stored on the floor; product on shelving at a specified clearance.
• Chemical sanitizer at the concentration on the label — quaternary ammonium products are commonly used at roughly 200–400 ppm — with a test strip, not a guess.
• Thermometers in every cold-holding unit, and a temperature log somebody actually signs.

All of this varies by jurisdiction. States adopt versions of the FDA Food Code on their own schedules, counties and cities amend them, and the specific requirements — logging intervals, date marking, required certifications, inspection frequency — differ from one county to the next. Get your local code, read the actual document, and when something is consequential, ask the health department rather than the internet. They will tell you; it is literally their job, and an operator who calls before opening is a more pleasant file than one who calls after a re-inspection.

Chapter 25 is the full treatment. This chapter's claim is narrower and it is a cost claim: every control on that list also protects inventory value, which means the compliance case and the business case point in exactly the same direction. That is unusual and you should exploit it.

13.6 Taking inventory: the count, the sheet, the discipline, the frequency

A physical inventory is an actual count of everything on hand at a specific moment, valued at what you paid for it. Not an estimate, not a system's on-hand figure, not what the invoices imply. A count.

It is the least popular two hours in restaurant management and it is non-negotiable, because — as Chapter 1 said and Chapter 11 proved — without an ending count you do not have the numerator of your food cost. You have a division problem with a missing term, and every operator who says "my food cost is about thirty" is telling you they have not done this.

The count sheet

Figure 13.8 — Count sheet, section E extract (the Bellwether plan)

Check the extensions: 3 × 11.20 = 33.60; 6 × 5.60 = 33.60; 4 × 18.75 = 75.00; 12.5 × 11.50 = 143.75; 11 × 14.85 = 163.35; 9.0 × 7.40 = 66.60; 7 × 12.55 = 87.85. Sum: 33.60 + 33.60 + 75.00 + 143.75 + 163.35 + 66.60 + 87.85 = \$603.75. Every column on every sheet you produce should be checkable like that by anyone who picks it up.

Five rules for the sheet itself:

Print it in shelf order, not alphabetical order. This is the single highest-value formatting decision in inventory control and it is astonishing how many restaurants get it wrong. A sheet in shelf order is walked once. An alphabetical sheet is walked eleven times, and on the eighth pass somebody stops caring.

Count in the unit you buy in, and extend at the price you paid. Not the list price. Not last year's price. If chicken was \$3.20/lb this period, a whole bird is \$11.20. Pull the prices from the last invoice of the period; most inventory software does it automatically and you should spot-check it anyway.

Count prepped items. Two ounces of made salsa verde is money exactly as much as the parsley was. The practical method is a standard batch value: cost the batch once (Chapter 11 built the cards), write the batch value on the recipe, and count in batches and fractions — "two full and a half." Restaurants that skip prepped inventory systematically understate ending inventory, which systematically overstates food cost, which produces a variance investigation into a problem that does not exist.

Leave the "counted by" and the time on the sheet. Not to blame anyone. To diagnose: when the same section is wrong twice and it is the same section both times, the problem is the sheet or the shelf, not the person.

Never modify last period's sheet. Ending inventory for one period is beginning inventory for the next; if you change one you have silently changed both, and the variance you produce next period will be entirely fictional.

👨‍🍳 On the Line

How a count actually goes wrong, at eleven at night.

Two people, always. One counts and calls; one writes and repeats it back. It is twice the labor and less than half the errors, and it removes the single largest source of bad counts, which is a tired person holding a clipboard in one hand and a case in the other.

Count after close and before the next delivery. Never mid-service, never on a delivery morning. Bellwether counts Monday night — the restaurant is closed, nothing has moved since Sunday brunch, and the Tuesday truck has not arrived. If your period ends on a delivery day, count before the truck and note the time on the sheet, because an inventory taken at 9:00 a.m. and a truck received at 7:15 are a \$1,238 error waiting to happen.

Same day, same time, same people, every period. Consistency beats accuracy here, and that is a genuinely counter-intuitive claim, so here is why. Suppose you systematically miss the two speed racks in the prep alcove — call it \$180 of product. Your ending inventory is \$180 low every period, which makes usage \$180 high every period, which makes food cost about 0.8 points high and constant. You can still see change. Now suppose you count them some periods and not others. Your food cost swings 1.6 points at random, and you spend the year investigating variances that are counting artifacts. A consistent bias is a rounding error. An inconsistent bias is noise, and noise destroys the whole instrument.

Three things that go wrong every time, in order of frequency: product on the floor of the dry store that nobody counts because it isn't on a shelf; the reach-in behind the line that belongs to nobody's section; and opened cases counted as full ones. Fix them by putting them on the sheet with a location code, which is the only thing that has ever worked.

Frequency: the three cadences

The key-item count is the highest-yield counting discipline in the building and almost nobody does it. Bellwether's twelve key lines — the chicken, the hanger, the ribeye, the grind blend, the trout, the pork rack, the butter, the cream, the eggs, the bread flour, the carrots, and the potatoes — are about \$3,550 of the \$6,438 weekly order guide: 55% of the food money in twelve lines out of fifty-four. Twelve minutes, six days a week, tells you within a day when one of them starts moving faster than it should.

Add it up: 12 minutes × 6 days + one 75-minute weekly count = about two and a half hours a week to control \$6,438 of purchasing. Chapter 1 said the survivors measure weekly and the casualties say their food cost "runs about thirty." This is the two and a half hours that separates them.

The limits, which are real. A count is a snapshot and it inherits every error in the sheet. It cannot tell you why a number moved — that is the variance in Chapter 11, and even the variance only tells you how much, never who or what. It is defeated entirely by product that isn't on the sheet. And it is expensive in the one currency a restaurant is shortest of, which is a manager's attention late at night. Counting more often is not automatically better. Count the key items daily, the food weekly, and everything monthly, and resist the software vendor who wants you to count 54 lines every night.

13.7 The usage formula and what your food cost really was

Chapter 11 did the arithmetic here and it does not need repeating. What it does need is a name, a procedure, and the two adjustments that make it honest.

The usage formula is the identity that converts a count into a cost:

$$\text{Food used} = \text{beginning inventory} + \text{purchases} - \text{ending inventory}$$

It is the only way food cost is ever real. Every shortcut — invoices divided by sales, purchases divided by sales, the POS's theoretical number on its own — is an estimate of this quantity, and each one is wrong in a direction that depends on which way your inventory moved.

The two adjustments

Transfers. Product crosses between food and beverage constantly. Citrus, cream, herbs, and simple syrup go from the kitchen to the bar; wine for a braise and beer for a batter come back the other way. If you do not record transfers, food cost is overstated and pour cost is understated by exactly the same dollars, and you will spend a quarter congratulating your bar manager and interrogating your chef. A transfer sheet is a clipboard on the wall between the two: date, item, quantity, cost, direction, initials.

Credit memos. A credit reduces purchases for the period in which it is applied. Credits chased in week 2 and applied in week 5 move money between periods and make both periods lie a little. Apply them in the period of the invoice wherever your bookkeeping permits.

🧾 Read the Numbers

```text FIGURE 13.9 — "The week that was 30.4%, 30.7%, and 29.8%" [the Bellwether plan] THE ARTIFACT One week's food-usage worksheet, built from two Monday-night counts, the week's invoices, the transfer sheet, and the credit log. THE CONTEXT Bellwether, first winter, a full seven-service week: five dinners and two brunches. Nothing unusual happened. No equipment failed, nobody quit, and the walk-in held 38F all week.

Beginning food inventory (Mon 11:10 p.m.)              $  6,120
+ Food purchases, gross invoices                       $  6,720
- Credit memos applied                                 $   (210)
  = Net food purchases                                 $  6,510
- Transfers OUT to bar (citrus, cream, herbs)          $    (95)
+ Transfers IN from bar (wine for braise, beer)        $     40
- Ending food inventory (following Mon 11:05 p.m.)     $ (5,930)
────────────────────────────────────────────────────────────────
= FOOD USED                                            $  6,645

Food sales for the week                                $ 21,860
TRUE FOOD COST  =  6,645 / 21,860                      =   30.4%

The same week, computed the two common wrong ways:
  gross invoices / sales   =  6,720 / 21,860           =   30.7%
  net purchases  / sales   =  6,510 / 21,860           =   29.8%

WHAT IT SHOWS Three defensible-looking answers from one week of real documents, and only one of them is what the restaurant actually used. The spread is 0.9 points -- about $197 on this week, roughly $10,200 a year. The shortcut is not even well defined: whether you subtract credits changes the answer by nearly a full point, and nothing in the phrase "invoices over sales" tells you which version somebody means. At 30.4% the week landed inside the 30.2-30.7% band Chapter 11's honest costed menu predicted. That is the system working: the cards said 30.4 and the walk-in said 30.4. WHAT IT DOESN'T It does not say where the 0.4 points above target went. Usage is a total, not a diagnosis. Chapter 11's ideal-versus-actual variance is what converts this number into causes, and even that only says how much. It also does not include employee meals, which Bellwether books to a separate line -- so this is food sold, not food consumed, and the chart of accounts in Chapter 31 has to be consistent about which. THE DECISION Post 30.4% to the weekly flash report, not 29.8%. Then pull the variance and the waste log for the same week and read all three together. Do not touch a menu price on one week of data. THE LESSON You cannot compute food cost without counting, and the shortcut does not fail in a consistent direction -- it flatters you in a week you ran the walk-in down and punishes you in a week you bought ahead. A number that is wrong in a predictable direction is a bias. This one is just noise wearing a percentage sign. ```

The arithmetic in that figure, checked: 6,120 + 6,510 = 12,630; less 95 transfers out = 12,535; plus 40 transfers in = 12,575; less 5,930 ending = \$6,645. And 6,645 ÷ 21,860 = 30.40%. The two shortcuts: 6,720 ÷ 21,860 = 30.74%; 6,510 ÷ 21,860 = 29.78%.

Note the identity buried in there, because it is the thing to remember when you are standing in front of a number: the difference between the shortcut and the truth is exactly the change in inventory, divided by sales. Inventory fell \$190 this week; \$190 ÷ \$21,860 = 0.87 points, which is precisely the gap between 29.78% and 30.65% before the transfer adjustments. Buy heavy at the end of a period and the shortcut punishes you for money you still own. Run the walk-in down and it congratulates you for eating your own inventory.

Inventory turnover

Inventory turnover is how many times you sell through your average inventory in a period. It is the number that tells you whether your pars are right in aggregate, and it is the bridge between food cost and cash.

$$\text{Inventory turnover} = \frac{\text{cost of food used in the period}}{\text{average food inventory for the period}}$$

Average inventory is (beginning + ending) ÷ 2. For the week in Figure 13.9:

• Average food inventory = (\$6,120 + \$5,930) ÷ 2 = \$6,025
• Weekly turnover = \$6,645 ÷ \$6,025 = 1.10 turns a week
• Annualized = 1.10 × 52 = about 57 turns a year
• Days of inventory on hand = 7 ÷ 1.10 = 6.4 days

"Days on hand" is the version to carry in your head, because it is physical: this restaurant has six and a half days of food in the building. Industry guidance generally puts a well-run full-service food inventory somewhere in the range of four to eight days on hand — roughly 45 to 90 turns a year — with produce-heavy operations at the fast end and operations carrying a lot of dry goods, aged product, or a deep wine list at the slow end. Bellwether at 6.4 days sits comfortably inside it. Treat those as orientation, not as a target for your specific business.

What a turn is worth, in cash. Suppose disciplined pars pull average food inventory from \$6,025 to \$5,000 — from 6.4 days on hand to 5.3. That is not profit. Nothing on the P&L changes. What changes is that \$1,025 that was sitting on a shelf is now in the bank account, permanently, and it stays there as long as the discipline holds. Against a working-capital reserve of \$45,000 (Chapter 33 will test whether that is enough), a thousand dollars released from the walk-in is not a rounding error. This is the connection between the least glamorous section of this book and the theme that closes restaurants: cash is not profit, and inventory is cash you decided to store as food.

And the limit, which matters more than the technique. Turnover has an optimum, not a maximum. Push it too high and you are 86ing items at 8:30 on a Saturday, sending a line cook to a cash-and-carry at retail prices, ordering below minimum drops and paying small-order fees, and losing every volume commitment you negotiated in §13.1. A restaurant running two days on hand is not efficient; it is one late truck away from a menu it cannot serve. Turnover is a range you manage into, and the bottom of the range is expensive in ways that never appear on the inventory report.

Shrinkage

Shrinkage is the gap between the inventory you should have — beginning plus purchases minus theoretical usage from the POS — and the inventory you actually counted, valued at cost. It is the physical-inventory face of the same thing Chapter 11 measured as the ideal-versus-actual variance.

Chapter 11 took a \$2,728 variance apart into six named causes and put theft seventh in the investigation order, behind portioning, waste, purchasing, uncosted specials, and mix drift. Nothing in this chapter changes that ordering, and you should treat it as a professional obligation rather than a preference. An operator whose first instinct on a bad variance is to suspect their staff will be wrong most of the time, will damage a team they need, and — this is the practical objection — will stop looking before they find the actual cause, which is usually sitting in plain sight on a receiving dock.

What this chapter adds is that the first several of those causes are physical rather than human. Receiving errors, temperature, rotation, labeling, and par discipline are all measurable with a scale, a thermometer, a label gun, and a shelf plan. You can attack most of a variance without a single conversation about anybody's honesty. Chapter 34 handles controls, surprise counts, and what to do in the rare case where the answer really is a person — including how to do it fairly. It is a later chapter for a reason.

13.8 Waste: pre-consumer, post-consumer, spoilage, over-production, and the waste log

Everything thrown away was bought. That sentence is the whole section, and it is worth sitting with, because kitchens are extraordinarily good at not seeing it. A bin is not a cost center in anyone's mental model. It is where things stop existing.

Chapter 38 defines pre-consumer and post-consumer waste properly, builds the waste audit, and makes the environmental case — which is a real case and a good one. Here we want something narrower: what waste is doing to this week's food cost, and which kinds of waste have different fixes.

Four categories, in descending order of how much money they usually represent:

Over-production. Prep made and never sold. This is the largest and the most invisible, because it does not look like waste — it looks like being ready. Three quarts of salsa verde made on Tuesday for a week that needed two.

Spoilage. Product that aged out on a shelf. This is a par problem, a rotation problem, or a temperature problem, and §13.3 and §13.5 are the fixes.

Trim above spec. Yield below what the yield test said. Chapter 11 owns the measurement; the fix is training and a knife.

Service waste. Re-fires, drops, plates sent back, mid-service recovery. Smaller than people think in dollars and larger than people think in ticket times. Chapter 14 owns it.

The waste log

A waste log is a clipboard at the bin. Date, item, quantity, unit cost, extended cost, reason code, initials. Six reason codes are enough: SPOIL, OVERPREP, TRIM, DROP, REFIRE, 86-REC.

It works for exactly one reason, and it is not the data: writing something down at the moment you throw it away makes people throw away less. The measurement is the intervention. Every operator who has ever put a log on a bin has watched waste drop in week one, before anybody analyzed anything.

Figure 13.10 — Waste log summary, one week (the Bellwether plan)

\$216.25 against \$21,860 of food sales is 0.99% — almost exactly one point of food cost. Annualized: **\$11,245 a year.** One point of food cost on \$1,116,000 of food sales is \$11,160, so that check ties.

Read the log, not just the total. Three of the four lines name their own fix:

• The greens on the back of a top shelf are a FIFO failure, not a purchasing failure. The par was right; the rotation was not. Fix: §13.5's loading rule and a Sunday "front-face" pass.
• Three quarts of salsa verde for a two-quart week is a production-planning failure, and it is Chapter 14's material. But note what cross-utilization does here: salsa verde reaches four menu items across two dayparts, so an over-produced batch has three other places to go before it dies. Compare the greens, which reach one dish.
• Two re-fired chickens on a Saturday is \$17.04 of product and about nine minutes of a hearth station, and it is a service-pressure signal (Chapter 14), not a cost problem.
• Hanger trim from a new prep cook in their first week is training, it is expected, and it should disappear by week three. If it does not, it is a different problem.

⚠️ Where the Money Leaks

Over-production, and why it hides better than spoilage.

Spoilage announces itself: something smells, somebody throws it out, everyone remembers. Over- production is silent, because it usually gets eaten — the next day, or as family meal, or as "let's run it as a special." Nothing hits a bin, so nothing hits a log, so the number is zero and the money is still gone.

The tell is not in the waste log. It is in the count. Prepped inventory that is systematically high, period after period, is over-production sitting in section D of the walk-in. If your prepped line is \$900 when your production sheet implies \$600, you are carrying \$300 of labor and product that the week did not ask for — and you are carrying it at 38°F for two or three days, which is how over-production quietly converts itself into spoilage.

Three fixes, in order of effect:

1. Prep to a forecast, not to a par, for anything with a short life. Pars are right for purchased goods with predictable shelf life. Prepped items with a three-day life should be produced against the covers forecast, which is why the forecast in §13.3 goes on the prep list and not just on the order guide. Chapter 14 builds this properly.

2. Over-produce only what cross-utilizes. Make the extra quart of salsa verde, which has four homes. Do not make the extra quart of anything that has one.

3. Put the batch size on the recipe card and the yield on the prep list. "Make salsa verde" is an instruction to make as much as fits in the robot coupe. "Salsa verde — 2 qt" is a decision somebody made on purpose.

🤝 Hospitality

The 86 is a guest problem before it is a cost problem.

Everything in this chapter pushes toward buying less and holding less, and there is a point past which that discipline starts costing you the thing the restaurant actually sells.

Consider the table at 8:40 on a Saturday. Four people, a reservation made eleven days ago, and one of them came for the chicken because someone at work told them about it. "We're out of the chicken" is not a small sentence. It is the moment a guest learns that the restaurant they chose could not be counted on to have the thing it is known for — and the second visit, which Chapter 23 shows is where the profitability actually lives, is decided by exactly that kind of moment.

The arithmetic is lopsided and worth internalizing. Over-producing eight halves costs \$44.80 of chicken, and Chapter 10's map converts most of it into Sunday's hash at full margin. 86ing the signature dish on a Saturday costs a party of four, their next visit, and whatever they say about it on Monday. Those are not comparable numbers.

This is why Bellwether's poultry safety stock is five halves on the weekend and three midweek rather than a flat two, and why the hash exists at all. A par level is a hospitality decision that happens to be denominated in cases. Set the cushion where the guest would notice, and run tight everywhere else.

🔍 Check Your Understanding

1. Bellwether's waste log shows \$216.25 for the week and Chapter 11's variance for the same week is \$540. Which number is closer to the truth about how much product was lost, and why can neither one be trusted alone?
2. Why does an over-produced batch of salsa verde cost less than an over-produced batch of something that appears on one dish, even when the two batches cost the same to make?
3. Your prepped-inventory line has been \$280 above the production sheet's implied value for four consecutive weeks. Is this a waste problem, a counting problem, or a production problem, and how would you find out?

(1: Neither. The log is a floor — people record what they put in a bin and not what quietly aged in a walk-in — and the variance is a ceiling, because it also contains portioning, receiving errors, uncosted specials, and mix drift. The truth is between them, and the gap between the two numbers is itself the most useful thing on the page. 2: Because cross-utilization gives it three more outlets before it dies; the same dollars of product have four chances to be sold instead of one. 3: Most likely production. Test it by counting the prepped line twice on the same night with two different people — if the number is stable, it is not a counting problem — and then compare the production sheet's batch sizes to actual sales for the same four weeks. Four consecutive weeks in the same direction is a system, not an accident.)

🍽️ The Business Plan

Checkpoint 13 of 40 — the Purchasing & Inventory section.

Chapter 10 gave the plan a menu and a cross-utilization map. Chapter 11 costed it. Chapter 12 ranked it. This checkpoint turns all three into the operating system that has to deliver a 30% food cost on \$1,116,000 of food sales — and it is the section a landlord, a lender, and an insurer all read for the same reason, which is that it is the cheapest available evidence that these two partners will run a business rather than a kitchen.

What goes into the plan.

1. Vendor structure. One broadliner carrying the undifferentiated commodity lines — dry goods, oil, chemicals, paper, dairy staples — approached for a cost-plus arrangement on a commitment covering roughly the \$185,000 of genuinely undifferentiated annual spend, and no further. A produce specialist, a meat-and-poultry specialist, and a fish specialist for the eight programs. Direct farm relationships seasonally for the root box and the herb box. A named second approved source, with the spec on file, for every program that reaches four or more menu items — which Chapter 10's map says is seven of the eight. Cash-and-carry tracked as a defect line, not a supplier.

2. Twelve written specs, covering the anchor line of each program: chicken [P-01], hanger [B-01], ribeye [B-03], grind blend [B-05], trout [T-01], pork rack [K-01], butter [D-01], cream [D-02], eggs [D-05], bread flour [B-08], carrots [R-02], parsley-and-caper box [H-01]. Each spec on file with the supplier and acknowledged in writing.

3. The order guide: 54 lines, eight programs, \$6,438 a week. Thirty-eight of the fifty-four lines and 66.9% of the money sit inside the eight programs. Annualized, \$334,800 — which is the 30% food cost target, expressed as a purchasing budget rather than a percentage.

4. Par levels and the delivery rhythm. Two protein drops (Tuesday and Friday), three produce drops, one dry-goods drop, receiving window 6:30–8:30 a.m. with a named receiver on the printed schedule. The poultry par is the model: 24 birds Tuesday, 28 birds Friday, 52 a week, \$582.40 — covering a 96-half forecast with an eight-half cushion that Chapter 10's map has already sold as Sunday's hash.

5. Opening inventory.

6. Control cadence. Daily key-item count (12 lines, 55% of food spend, ~12 minutes). Weekly full food count Monday night, feeding the weekly flash report Chapter 31 designs. Monthly full food and beverage count. Waste log at every bin from day one. Transfer sheet between kitchen and bar. Credit log with an "applied Y/N" column.

7. Targets this section commits to. Food cost 30.0%. Inventory turnover ~57 times a year, 6.4 days on hand. Receiving exceptions logged on 100% of deliveries. Waste log entries at or below 1.0% of food sales.

What this checkpoint does not settle — and it is a longer list than it looks.

• Nobody has quoted this restaurant anything. Every price in the order guide, including the \$3.20/lb chicken that the entire cost structure rests on, is a planning assumption. Two purveyors need to see the spec sheet and the volume commitment before any of it is real.
• Whether fifty-two birds a week buys anything. It is a small commitment. It may be worth a better price, a guaranteed spec, or nothing at all, and the plan should not assume the first two.
• Whether one prep cook can absorb the butchery. Chapter 10 put it at an hour and a quarter twice a week. Chapter 19 has to find those hours in the staffing guide, and if it cannot, the poultry program changes shape and the \$5.60 changes with it.
• **The \$20,700 of opening inventory has to be reconciled** against the \$35,000 pre-opening budget from Chapter 9, which also has to carry pre-opening labor, training, and licensing. Chapter 33's working-capital work has to resolve it, and the honest reading today is that the pre-opening line is thin.
• The walk-in. Chapter 7 specified refrigeration for a 2,800 sq ft space with 200 sq ft of storage. Nothing in the plan yet commits to a preventive maintenance schedule for it, and §13.5 just showed that eight degrees is worth \$4,368 a year.

Open questions carried forward:

1. What does a real quote against these twelve specs look like, and does the plan's food cost survive it? (Chapter 31)
2. Can the staffing guide fund the butchery hours the cross-utilization map assumes? (Chapter 19)
3. How much working capital is actually tied up in inventory across a full year, including the February trough? (Chapter 33)
4. What happens to purchasing when the restaurant adds off-premise volume with a different pack and a different waste profile? (Chapter 28)
5. What does the food-safety plan — logs, certifications, temperature discipline, inspection readiness — need to look like so that the Tuesday in §13.5 never gets written up? (Chapter 25)

Conclusion

Food cost is decided in four places and the stove is not one of them.

It is decided on the order guide, where a par level is either a calculation from a forecast or a feeling about how the walk-in looked; where eight purchasing programs concentrate seventy percent of your lines and two-thirds of your money into a small number of decisions you can actually manage.

It is decided at the door, in the twenty minutes a day when you can still say no — where a scale, a thermometer, and a spec sheet turn a 3.90 lb bird from a sixty-five-cent-a-plate leak into a \$35.84 credit and eleven pounds of hash meat, and where a returned 3.4× on a manager's hour is the best recurring investment in this book.

It is decided in the walk-in, which is a financial statement you can read from the doorway. The same neglect that puts raw chicken over ready-to-eat greens and lets a box drift to 46°F produces both a health inspector's report and a food-cost variance, and the arithmetic in §13.5 says the temperature alone is worth roughly the whole gap between Chapter 11's honest costed menu and its target. Chapter 25 owns the inspection. This chapter owns the shelf.

And it is decided on the count sheet, at eleven at night, on the same day of the week, by the same two people, in shelf order — because until you have counted, you do not have a food cost. You have an opinion with a percent sign after it.

None of this requires talent. It requires a written spec, a par derived from a forecast, a person at the door, a labeled shelf, a clipboard at the bin, and the willingness to do the same unremarkable things on a Friday when you are behind. That is the entire discipline, and it is worth three to four points of food cost — which on Bellwether's plan is between thirty-three and forty-five thousand dollars a year, against an operating profit the plan projects at about sixteen points before debt service.

Chapter 14 takes the delivery you just received and turns it into service. The birds are broken down, the sauce is made, the prep list is written, and at five-thirty the tickets start. Everything this chapter bought now has to survive contact with a Saturday.

Key Terms

Purveyor (also vendor, supplier) — any business that sells product to a restaurant. The operative decision is not which purveyor to like but how to distribute spend across categories of purveyor. (Ch. 13)

Broadline distributor — a supplier carrying a very wide catalog across food, paper, chemicals, and smallwares, delivered on one truck against one invoice with credit terms. You buy consolidation and give up depth. (Ch. 13)

Specialty distributor — a supplier that goes deep in a single category — produce, seafood, meat, bread, coffee, cheese. Usually smaller, often better, frequently more expensive per unit, and where a chef-driven menu actually gets sourced. (Ch. 13)

Product specification (spec) — a written description of exactly what you are buying — grade, size or weight range, fabrication, pack, temperature at receipt, substitution rule — precise enough that a delivery either meets it or measurably does not. Write one for anything expensive, variable, or load-bearing on the menu. (Ch. 13)

Par level — the quantity of an item you want on hand at a defined point in the ordering cycle: forecast usage until the next delivery, plus a deliberate safety stock sized to what running out would cost. (Ch. 13)

Order guide — the standing list of every purchased item, in the unit you buy it in, from the supplier you buy it from, with columns for on-hand and order quantity. The document you order from; a walk through the walk-in is not one. (Ch. 13)

Receiving — the procedure performed when a delivery arrives: check the time window, temperature, count, weight, spec, and price, then sign every exception. The only moment in the purchasing cycle when you can still say no. (Ch. 13)

Invoice reconciliation — the three-way match of what was ordered (the order guide), what the invoice says, and what physically arrived. Checking the invoice alone catches nothing, because the invoice is the supplier's account of events. (Ch. 13)

Credit memo — a supplier's written acknowledgment that they owe you money or product for something short, rejected, or out of spec. Get it at the door, signed by the driver, and verify it appears on the statement. (Ch. 13)

Prime-vendor agreement — a contract committing a defined share of purchases to one broadliner in exchange for better pricing (often cost-plus), service-level commitments, and a contracted item list. Terms vary widely and are negotiable; the definition of "cost" is frequently worth more than the markup. (Ch. 13)

FIFO (first in, first out) — the rotation discipline in which the oldest stock of any item is always used first, which in practice means new product is loaded behind old product, never in front of it. (Ch. 13)

Physical inventory — an actual count of everything on hand at a specific moment, valued at what you paid for it. Taken in shelf order, by two people, at the same time on the same day every period. (Ch. 13)

The usage formula — the identity that converts a count into a cost: beginning inventory + purchases − ending inventory = product used, adjusted for transfers between food and beverage and for credits applied. Chapter 11 worked the arithmetic; this is its name and its procedure. (Ch. 13)

Inventory turnover — cost of product used in a period divided by average inventory for that period, often expressed as days of inventory on hand. It is the bridge between food cost and cash: inventory is cash you decided to store as food. (Ch. 13)

Shrinkage — the gap between the inventory you should have and the inventory you counted, valued at cost. The physical-inventory face of the ideal-versus-actual variance, and — per Chapter 11's ordering — investigated through portioning, waste, purchasing, uncosted specials, and mix drift long before anyone suspects a person. (Ch. 13)

Spaced Review

1. From this chapter: Bellwether's Friday poultry par is 28 birds and the Tuesday par is 24. Explain where each number comes from, and why the split is uneven when Chapter 10 sized the program as twenty-six and twenty-six.
2. From Chapter 11: a delivery of birds averages 3.90 lb against a 3.25–3.75 lb spec. Without looking back, compute the new plate cost of the Hearth Chicken if the halves are plated as they come, and state the new food cost percentage and contribution margin against the frozen \$8.52 / \$29.00 / 29.4% / \$20.48 baseline.
3. From Chapter 1: an operator tells you their food cost is 28% and they have never counted their walk-in. Using the identity in §13.7, describe the two circumstances under which that claim would be badly wrong in opposite directions.
4. From Chapter 10: why does cross-utilization change the correct size of a safety stock, and not merely the cost of getting it wrong?
5. The recurring question: you have twenty minutes a day and one clipboard. You may spend them at the receiving door, on a daily key-item count, or on a waste log at the bin. Rank the three for a restaurant that has never done any of them, justify the ranking in dollars, and say what would change your ranking.