Chapter 14: Kitchen Operations: Workflow, Stations, Prep Systems, and the Organized Chaos of Service

"The service was decided at two o'clock. Everything after five is just finding out what you decided." — constructed; the sentence a chef says to a cook who wants to argue about the pass

Overview

Here is the operating question this chapter exists to answer: it is 3:41 on a Friday afternoon, you are one cook short, and you have 142 people coming. What do you change, in what order, and what does each change cost?

That is not a cooking question. Every person in that kitchen can cook. The cost cards are built, the menu is engineered, the walk-in is full. What is about to go wrong has nothing to do with food and everything to do with the fact that a dinner service is a manufacturing run with no inventory buffer, a variable arrival rate you do not control, and a customer sitting fifteen feet away watching the clock. You cannot build ahead, ship late, or substitute — and unlike a factory, your production line reconfigures itself every night depending on who showed up.

So the systems that survive a service are built before four o'clock, and almost all of them are boring: a prep list written against a forecast, a par-cook policy, a station map somebody actually drew, a written ticket-time standard, and an 86 decision made at 4:20 instead of 7:20.

This chapter walks a kitchen day from the first delivery to the last log entry and does the arithmetic operators usually skip. You will compute a station's throughput ceiling and discover that Bellwether's ceiling is not the seats in the dining room and not the cooks on the line — it is the wood fire, and it is smaller than anyone assumed. Then you will spend a Friday night finding out what happens when a ceiling that is fine on average meets a night that arrives all at once.

One warning about that Friday: Bellwether has not opened. Everything below is the plan being stress-tested on paper, which is the only place stress tests are cheap.

In this chapter, you will learn to:

• Lay out a kitchen day as a sequence of blocks, and identify the ones that cannot be recovered.
• Treat mise en place as an auditable production system, with every item tied to a forecast number.
• Compute a station's sustained throughput from cooking surface and dwell time, and find the station that is actually capping your revenue.
• Work a pass: call, fire, hold, quote, and decide who waits when demand exceeds the equipment.
• Measure ticket time correctly and name the specific failure a rising number is describing.
• Decide what to 86 and when, and price that decision in contribution margin against the capacity it returns.
• Write a production sheet with defensible par-cook levels and batch sizes, and state the waste each risks.

Learning Paths

🏗️ Opening — one of the two or three chapters your build-out decisions get graded by. Work §14.3 by hand against your own equipment list; if the arithmetic says your line cannot produce your forecast, the forecast is wrong and no amount of hustle fixes it. 📋 Managing — §14.4, §14.5, and §14.6 are your job description. The ticket-time report in §14.5 is the most useful document here for someone running someone else's kitchen. 🍸 Beverage — §14.4 and §14.6 for the FOH/BOH handshake; a bar that fires food, and a flatbread competing for entrée fire, are both bar problems. Chapter 15 is yours. 🚚 Small Format — §14.3 and §14.7 matter more to you. A truck is one station and a hold cabinet; par cooking and batch size are the entire operating model, and your ceiling is brutal and knowable.

14.1 The day in a kitchen: deliveries, prep, family meal, stretch, service, close

A restaurant day is not one long shift. It is five or six distinct blocks with different work, different people, and — this is the part that gets missed — different reversibility. Some blocks can be recovered if they go badly. Some cannot.

Here is a Friday at Bellwether, drawn as a day.

FIGURE 14.1 — The shape of a Friday at Bellwether              [the Bellwether plan]

  8:00 ▓   chef + prep cook in · temp logs · read last night's waste log and notes
  9:00 ▓▓  produce delivery received, checked against spec and invoice (Ch. 13)
 10:00 ▓▓▓ PREP 1 — butchery, brine, stocks, sauces, salsa verde batch
 12:00 ▓▓▓ protein delivery · counted against the production sheet
  1:00 ▓▓▓ PREP 2 — par-cook: chicken par-roast; pork chop and hanger over wood
  1:30 ★   HEARTH LIT — ninety minutes from cold to working temperature
  3:00 ▓▓▓▓ line cooks in · station mise · plate-up check on anything new
  4:00 ▓▓▓▓ family meal
  4:40 ●●●  STRETCH — ten minutes: the book, 86s and counts, menu, one correction
  4:55 ▓▓▓▓ final set-up · lamps on · fire at working temperature
  5:00 ████ SERVICE ──────────────────────────────────────────────────────
  7:00 █████████ the peak band
  9:30 ███  last seating
 10:30 ▓▓▓  break down · clean · walk-in check · logs · tomorrow's prep list
 11:30 ▓    close

   ▓ prep and set-up   █ service   ● meeting   ★ a ninety-minute lead time

Read the diagram for the thing it is actually about, which is not time but dependency. Almost every block feeds the one after it, and two feed a block that happens tomorrow.

Deliveries (8:00–12:30). Chapter 13 covered receiving. What matters here is scheduling: a delivery landing at 11:00 instead of 9:00 does not cost two hours of waiting, it costs two hours of prep, because the prep cook cannot butcher chicken that is on a truck. A recurring late delivery is a prep-labor problem wearing a logistics costume.

Prep (10:00–3:00). Where the day's real production happens, and where the largest lever in a kitchen lives — par cooking, which §14.7 does in detail. Prep is also the only block with slack. If service goes badly at 7:30 there is no recovering; if prep goes badly at 11:00 you have four hours. Which is precisely why prep failures are expensive: they could have been caught and weren't.

The hearth (lit 1:30). Ninety minutes from cold to working temperature, with no way to hurry it. Put it on the opening checklist, not the grill cook's list. An asset with a long lead time and no substitute should never depend on the punctuality of one hourly employee — a principle Bellwether's plan is about to be very glad it followed.

Family meal (4:00). Every kitchen feeds its staff before service. Not sentiment: it is the difference between a cook who is sharp at nine and one who is not, it costs perhaps \$1.20 to \$1.80 a head in trim and ends that would otherwise be waste, and operators who cut it reliably discover they bought a worse service.

The stretch (4:40). The pre-shift meeting for the kitchen. Ten minutes, standing, everybody.

👨‍🍳 On the Line

What the stretch actually covers, and why it is ten minutes and not thirty.

The pre-shift — in most kitchens called the stretch, because that is roughly how long it takes and because people are usually stretching while it happens — is a short standing meeting immediately before service. In a well-run kitchen it covers exactly five things, in this order:

1. The book. How many covers, what the shape is, and where the pressure will be. "One-twenty on the book, forty of it a private party at six-thirty, so between seven and eight we are going to be underwater and I want you to know that now."
2. The 86s and the counts. What we do not have, and what we have a limited number of. A count is more useful than an 86: "eight ribeyes, that's all there is, tell me when you're at three."
3. Menu changes and specials. Costed, tasted, and described — not announced. If a cook cannot describe the special to a server, the server cannot sell it, and an uncosted special is a Chapter 11 sin that shows up in next week's food cost.
4. One correction. Exactly one. Yesterday something went out wrong; name it, show the fix, move on. More than one and nobody remembers any of them.
5. Assignments and the plan for the crunch. Who is where, and what changes when it gets bad.

It is ten minutes because a twenty-minute stretch has become a meeting, and because those ten minutes are being paid for by four to six people at once. Ten minutes × five people × \$21 an hour is about \$17.50 a night, roughly \$4,400 a year. That is the best money in the kitchen — but it is real money, which is why it should be ten minutes.

The failure mode: the stretch that becomes a monologue. If only the chef talks, the chef is the only person who learns anything, and the chef already knew it. The best version ends with the chef asking each station one question and shutting up.

Service (5:00–9:30, plus the tail). Covered in §14.3 through §14.6.

Close (10:30–11:30). The most-skipped block in the industry, and the subject of §14.8. Two of its outputs — the walk-in check and tomorrow's prep list — are tomorrow's inputs, which is why a kitchen that closes badly starts every day at a deficit it never traces.

14.2 Mise en place as a system, not a virtue

Mise en place — French, "everything in its place" — is the complete set of prepared components, tools, and set-up a station needs to execute its menu at the speed service demands. It is the most famous phrase in professional cooking and the most commonly misunderstood, because it is usually taught as a virtue (be organized, be prepared, be the kind of cook who has their station together) rather than as what it actually is: a production system with a forecast, a bill of materials, a par level, and an audit.

The virtue framing fails for a specific reason. Virtues do not scale and do not transfer. If your mise depends on a cook being conscientious, your throughput depends on which cook is working, and you will discover this on the night the conscientious one calls out. If your mise depends on a written sheet tied to a number, it survives a personnel change.

The three parts of a mise system

One: the prep list. A prep list is the day's task list for the prep function — the items to be produced, in what quantity, by whom, by when. In a kitchen with any discipline it is not written from memory; it is written from a count of what is on hand against a par, exactly like Chapter 13's order guide.

Two: the production sheet. A production sheet is the prep list plus the arithmetic: for each item, the par level, the amount on hand, the amount to produce, the recipe yield, and the forecast the par is derived from. A prep list says make salsa verde; a production sheet says make 3.75 quarts, because we have one quart, par is 4.75 quarts, par is two days of the forecast, and the forecast is 475 covers this week. A cook can execute a prep list. Only a production sheet can be audited, and only a production sheet tells you why you ran out.

Three: the station set-up standard. What is in each station's low-boy, in what order, in what container, at what quantity, at the start of service. Written down once, with a diagram. It feels bureaucratic and is worth the most, because it converts "did you set up?" into a thirty-second visual check anyone can perform.

🧾 Read the Numbers

```text FIGURE 14.2 — "The Friday production sheet" [the Bellwether plan] THE ARTIFACT The kitchen production sheet for the second Friday in October, printed at 8:05 a.m. Ten of about forty lines shown. THE CONTEXT Bellwether. 142 covers on the book, including a contracted 40-guest private party at 6:30 on a pre-set three-choice menu. Par levels are set to the forecast plus a coverage factor stated in the right-hand column.

 ITEM                     PAR    ON HAND   MAKE   YIELD/BATCH   PAR IS SET TO
 ─────────────────────────────────────────────────────────────────────────────
 chicken, par-roast       46 ea    12 ea   34 ea   2 ea/bird    Fri fcst + party
 pork chop, wood par-cook 30 ea     9 ea   21 ea   —            Fri fcst × 1.15
 hanger, wood par-cook    22 ea     7 ea   15 ea   —            Fri fcst × 1.15
 salsa verde              4.75 qt  1.0 qt  3.75qt  1.25 qt/bt   2 days of fcst
 roasted roots            9 qt      2 qt    7 qt   3 qt/hotel   1.5 days of fcst
 flatbread dough          20 ea     0 ea   20 ea   10 ea/batch  Fri fcst × 1.25
 grain base               6 qt      1.5 qt 4.5 qt  1.5 qt/bt    2 days of fcst
 party first course       40 pl     0 pl   40 pl   —            BEO, plate by 6:15
 brine, chicken           6 gal     2 gal   4 gal  2 gal/bt     rolling
 hearth wood, split       1.5 face  0.6     0.9    —            2 services

WHAT IT SHOWS Every quantity traces to a number, not a feeling. The chicken par of 46 half-birds is the Friday forecast (28) plus the party's contracted 14, plus four of cushion. Pork and hanger carry a 15% coverage factor because they are par-cooked over wood and cannot be replaced mid-service; flatbread dough carries 25% because Friday's bar over-indexes on it. One line — "party first course, plate by 6:15" — is a capacity decision disguised as a prep task, and §14.4 explains why it matters most. WHAT IT DOESN'T It does not say who is doing the work or whether there are hours to do it; that is Chapter 19's schedule, and this sheet assumes a grill cook who is about to be a problem. It carries no cooling times or temperature checkpoints — those belong on the food-safety logs (Chapter 25), and a production sheet that tries to be a HACCP record ends up being neither. It also does not know the à la carte book is 102 covers, not the 80 a normal Friday brings. THE DECISION Make it as written, and flag the two lines with no replacement path — the wood par-cooks — for a physical count at 3:00 rather than a look at 5:30. If either is short the fix costs four hours, and you only have four hours if you find out at three. THE LESSON A prep list tells a cook what to do. A production sheet tells an operator what happened. The column that earns the paper is the right-hand one: every par traced to the forecast that produced it. ```

The mise audit

The operational discipline takes four minutes: at 4:30, before the stretch, the chef walks every station and checks the set-up standard against what is actually there.

Not "are you ready?" — which produces "yeah, chef" from a cook who is not ready and does not yet know it. The audit is physical: open the low-boy, count the containers, check two par-cook counts against the sheet, taste one sauce, confirm the items with no replacement path.

Do it at 4:30 and not 4:55 for a purely arithmetic reason. At 4:30 you have thirty minutes, which is enough to fix a shortfall. At 4:55 you have five, which is enough to discover one and nothing else.

⚠️ Where the Money Leaks

Mise built to last week's number.

The commonest prep failure in an independent restaurant is not laziness. It is a par level that was correct in July and has never been revisited.

Bellwether's plan week is 62 / 78 / 92 / 120 / 123 covers — 475 for the week, and a steep shape: Saturday is almost exactly twice Tuesday. A prep cook who builds Tuesday and Saturday to the same par is doing one of two costly things every week:

• Building to Saturday. Tuesday's kitchen then produces 123 covers' worth of components for a 62-cover night, and roughly half the perishable prep is stretched past its quality window or thrown away. On a kitchen with \$220 a day of perishable prepared components, 40% over-prep on the three slow nights is about \$264 a week — **\$13,700 a year**, a little over five percent of the plan's operating profit before debt service, in food that was cooked and never sold.
• Building to Tuesday. Saturday runs out at 8:15, which is worse, because it turns a food-cost problem into a revenue and guest problem.

What the disciplined operator does: set every par as a stated multiple of a stated forecast — the right-hand column of Figure 14.2 — and re-derive them whenever the forecast or the menu mix moves. Then the par is auditable, and a cook who runs out can be asked a useful question instead of a blaming one: was the par wrong, or was the count wrong?

Most "we ran out" incidents are count errors, not par errors — somebody wrote down what they thought was in the walk-in rather than what was in it. The countermeasure is Chapter 13's: count the thing, in the place, with the sheet in your hand.

14.3 The stations and the line: who does what, and what happens when one is down

Chapter 7 laid out Bellwether's kitchen in physical terms — the line, the stations, the pass, the prep area, dish, and storage. This section takes the same room and asks the operating question: how much can it produce, and what breaks first?

Here is the line as a cook standing in it would draw it.

FIGURE 14.3 — The Bellwether line: four people at peak          [the Bellwether plan]

    ┌───────── WALK-IN + reach-ins ─────────┐   ┌──── DISH PIT ────┐
 ═══╧═══════════════════════════════════════╧═══╧══════════════════╧═══

  [ GARDE MANGER ]   [ SAUTÉ / OVEN / PASTA ]    [ HEARTH ][ PLANCHA ]
   cold · desserts    4 burners · convection      wood fire  flat-top
   bread · party      all pasta · all sauce       fired to   finishes the
   plating            work for the hot line       order      wood par-cooks

   ~13 starters       22% of entrées              44% of     34% of entrées
   ~14 desserts       + 6 starters                entrées    (pork chop,
   (95-cover night)   + finish on 67 plates       + 13       hanger steak)
                                                  flatbreads
        │                     │                      │          │
        └──────────┬──────────┴───────────┬──────────┴──────────┘
              ┌────┴──────────────────────┴────┐
              │       THE PASS / EXPO          │  ← the chef stands here
              │  ticket rail · heat lamps ·    │
              │  86 board · all-day board      │
              └───────────────┬────────────────┘
                    ─── to the dining room ───

   PEAK CREW (4): 1 hearth · 1 sauté · 1 garde manger · chef at the pass
   NOTE: the plancha has NO dedicated cook. The hearth cook works it, two steps
   left. One person therefore owns 78% of every entrée that leaves this kitchen.

That last line is the most important sentence in the chapter. The peak line is four people — one on the hearth, one on sauté, one on garde manger, and the chef expediting at the pass and stepping into whichever station is drowning. The plancha, which finishes the wood-par-cooked pork chop and hanger steak, is not a fifth position. It is the hearth cook's left hand.

Where the entrées actually go

Chapters 7 and 10 settled the mix, and it is worth stating precisely because people garble it. Of Bellwether's entrée orders: 44% go to the hearth, fired to order over live wood; 34% go to the plancha, having been par-cooked over wood during prep — the pork chop and the hanger steak; and 22% go to sauté and oven — the pasta, and the squash-and-grains plate that finishes in the convection oven.

A separate and frequently confused number: peak fire utilization is 67%. That is not a share of entrées. It is the fraction of the hearth's sustained hourly capacity that the busiest hour of a 95-cover plan night consumes — 18.7 fire actions against a ceiling of 28. Two different quantities, two different denominators. Keep them apart or the capacity math collapses.

🧮 Run the Numbers

Where the ceiling comes from, and what it costs when the mix moves.

The hearth's sustained capacity is 28 items an hour, and it is worth deriving, because a number you can derive is one you can re-derive when something changes.

The working grate holds four items at a time in the usable heat zone. Average dwell across the hearth's mix — chicken, fish, vegetables, flatbread — is about eight minutes. Four positions at eight minutes is a theoretical 30 an hour. Subtract the fire itself: feeding wood, raking coals, moving items off the hot zone, and the fact that a cook cannot turn four things at once. Sustained throughput is 28 items an hour, not 30.

Call one item's worth of that capacity a fire slot. One slot is $60 \div 28 = 2.14$ minutes of hearth time — the workhorse conversion for everything that follows.

Now the peak hour of a plan night. 95 covers, 0.90 entrée attachment, so 86 entrées, plus about 43 starters and sides and 14 desserts. The busiest hour carries roughly 30% of the night's food.

Peak-hour demand on the hearth	Items
Entrées: $86 \times 0.30 \times 0.44$	11.35
Hearth flatbreads (13 on the night, ~30% in the peak hour)	4.00
Hearth-finished sides (11 on the night, ~30% in the peak hour)	3.35
Total	18.70

$18.70 \div 28 = 66.8\%$ — the 67% peak fire utilization figure. Comfortable. A third of the fire is spare on the busiest hour of a normal night.

Now solve the ceiling. For a night of $C$ covers with this shape and this mix, peak-hour hearth demand is:

• entrées: $C \times 0.90 \times 0.30 \times 0.44 = 0.1188\,C$
• flatbreads and sides: $(C \div 95) \times 24 \times 0.30 = 0.0758\,C$
• total: $0.1946\,C$

Set that equal to 28: $C = 28 \div 0.1946 = \textbf{144 covers.}$

The floor plan allows about 151 covers. The fire allows about 144. The binding constraint on this restaurant's biggest possible night is not the dining room, not the staff, and not the reservation book. It is a wood fire, and it caps the building seven covers below what the seats could hold.

Now move the mix. Suppose the hearth share drifts from 44% to 50% — one popular new hearth item, or the chicken outselling projections. Re-run it: $C \times 0.90 \times 0.30 \times 0.50 = 0.1350\,C$, plus $0.0758\,C$, equals $0.2108\,C$. Set to 28: $C = \textbf{133 covers.}$

Six points of menu mix cost eleven covers of ceiling — about \$506 of sales on every night the room would otherwise have filled. Chapter 12 taught you to move mix toward contribution margin. This chapter adds the constraint: check what the shift does to the fire before you make it.

Here is the same information as a station-by-station load picture — the version you would pin up.

Figure 14.4 — Station load, 95-cover plan night [the Bellwether plan]

Entrées: 86 total = 38 hearth (44%) + 29 plancha (34%) + 19 sauté/oven (22%). Utilization is hands-on minutes ÷ 270 minutes of service — estimates, not stopwatch data.

Three things fall out of that table.

First, the two hot stations run at three-quarters utilization on an ordinary night. That sounds like comfortable headroom and is not. Seventy-five percent sustained over four and a half hours, standing, in heat, with no relief, is a hard shift — and it is an average across a service whose demand is anything but average.

Second, par cooking is what makes a two-person hot line possible. Without it, the pork chop and hanger would each need eleven to fourteen minutes on live fire, and 29 of them would add roughly 350 minutes of hearth time to the 211 already there. The station would need two people and the hearth would need to be twice as big. The par-cook policy in §14.7 is not a convenience; it is the reason this kitchen's labor model works at all — the clearest example in the book of a decision that moves cost between the two halves of prime cost.

Third, garde manger has the slack. Fifty-six percent is the only real cushion in the building, which is why it is the station a chef raids when something goes wrong — and why anything that lands 40 cold plates there in a twelve-minute window turns the only cushion into the first casualty.

👨‍🍳 On the Line

What "a station is down" actually means.

Restaurant books say "cross-train so you can cover a station." The phrase hides the real cost.

When the hearth cook is missing you do not lose a quarter of the kitchen. You lose 78% of entrée production capacity, because that person owns hearth and plancha. You lose the only person who knows what the fire is doing right now — a wood fire is not an appliance with a dial; its hot zones move, it needs feeding on a rhythm, and two hours of standing in front of it cannot be handed over in a sentence. And you lose that station's mise-in-motion: what is resting, what is holding, what is two minutes out.

So "the chef covers the hearth" actually means: the sous moves to the fire and rebuilds the mental model from zero, the chef leaves the pass to absorb the sous's old work, and the pass — the only position with a view of the whole service — goes half-staffed. A station going down does not degrade a kitchen linearly. It removes the coordination layer, and coordination is what a pass is for.

The failure mode nobody plans for: the chef covers the station, does it well, is visibly heroic about it, and never notices that ticket times on the other stations rose six minutes because nobody was calling the order. The heroism is the problem, and it is hard to argue with, because the food that went out was excellent.

🔍 Check Your Understanding

1. Bellwether's hearth does 28 items an hour. On a 95-cover night the peak hour uses 18.7 of them. Why is the restaurant's ceiling 144 covers rather than $28 \div 18.7 \times 95 = 142$ or some much larger number?
2. The chef wants to add a wood-grilled swordfish to replace a sauté item. It would move about four points of entrée mix from sauté to hearth. What happens to the kitchen's ceiling, and what would you need to know before approving it?
3. Why does losing the hearth cook cost more than 25% of a four-person line's output?

(1: Because covers scale both the entrée load and the flatbread/side load on the fire, so peak demand is $0.1946\,C$; setting that to 28 gives 144. 2: Hearth share rises to 48%; peak demand becomes $0.1296\,C + 0.0758\,C = 0.2054\,C$, so the ceiling falls to about 136 — eight covers, roughly \$370 of sales on a full night. You would want the item's contribution margin against the one it replaces, how often the ceiling actually binds, and whether swordfish dwells longer than the eight-minute average the 28-per-hour figure assumes. 3: Because that cook owns hearth plus plancha — 78% of entrées — and covering the station pulls the chef off the pass, removing the coordination layer as well as the capacity.)

14.4 The pass: expediting, calling, firing, and communication under load

Expediting is the coordination function at the pass: receiving tickets, sequencing them, communicating them to the stations, controlling when each component starts cooking, checking every plate before it leaves, and deciding who waits when not everything can be done at once. In a large kitchen the expediter is a dedicated position; in a 68-seat restaurant it is the chef, which is exactly why the chef cannot also be cooking.

The vocabulary is small and precise, and every word exists because a mistake made it necessary.

The whole architecture exists to solve one problem: the components of a table's meal cook at different speeds and must arrive at the same moment. A hearth chicken is eight minutes from par-roast. A pasta is six. A dry-aged ribeye is sixteen on the fire plus six resting. A cold salad is ninety seconds. If four people order those four things and every station simply starts when the ticket lands, three plates sit under a lamp for fourteen minutes while the fourth cooks.

So the expediter works backward from the longest item on the ticket and calls the others accordingly. That is the entire job, and it is why the pass cannot be a person who is also cooking: it requires holding a dozen tickets, each with its own critical path, in working memory, and updating them every time a new ticket lands.

The dry-aged ribeye, and the item you cannot fix in the kitchen

Bellwether's dry-aged ribeye is the clearest case in the book of a menu item that is an operations problem rather than a cost problem. Chapter 12 likes it: high price, high contribution margin, a genuine reason for a certain kind of guest to choose this restaurant. Chapter 14 disagrees.

Sixteen minutes on the fire, six resting. Twenty-two minutes minimum against a fourteen-minute standard. It cannot meet the standard — not on a slow Tuesday, not with a great cook, not ever. And it does not merely take twenty-two minutes for itself: it holds a fire slot for twice the average dwell, so it consumes two fire slots, not one, and it holds the whole table's ticket, so if one person at a four-top orders it the other three plates either sit or arrive late.

There are exactly three responses, and only one works:

1. Cut it from the menu. Solves the operations problem, gives up a differentiating item and a large contribution margin. Sometimes correct; usually an over-correction.
2. Cook it faster. There is no such thing. Every version — thinner cut, hotter fire, finishing in the oven — is a different, worse dish.
3. Fix it at the table. The server quotes it: "the ribeye takes about twenty-five minutes, so I'll bring the rest of your table's food when it's ready." The kitchen carries a separate twenty-four minute standard for any ticket containing one, tracked on its own line so it does not contaminate the main number.

Option three is the professional answer, and note what it is: an operations constraint solved by hospitality. Chapter 22 builds the FOH side properly. The principle generalizes: when the kitchen cannot meet a standard, the fix is usually a conversation at the table, not a shortcut at the stove.

🤝 Hospitality

The held ticket, and why a table would rather wait together.

Under load, an expediter's most useful power is the power to not send food.

Picture a four-top at 7:40 on the Friday. Two plates are ready. Two are nine minutes out because the hearth is backed up. The instinct — strong, because plates are up and the lamp is full and the pass is drowning — is to send the two.

Do not send the two.

A table that waits twenty-six minutes and then eats together has had a slow meal. A table where two people eat while two watch has had a bad meal, and it is bad in a way guests describe to other people. The two who got food eat guiltily and fast; the two who didn't have their evening defined by the wait; and the server, who did nothing wrong, absorbs all of it.

The move is: hold the ticket, and tell the server before they find out from the guest. A guest told at minute twelve that their food is running long is being taken care of. A guest who works it out at minute twenty-four is being ignored — and the two situations are physically identical.

Under load, hospitality is mostly information delivered early. It costs nothing, requires no food, and is the highest-return move available to a kitchen that is behind.

The Friday night

Now we can do the thing this chapter exists for.

It is the second Friday in October. There are 142 covers on the books against a Friday plan of 120. Forty of them are a contracted private party seated at 6:30 on a pre-set three-choice menu, which means they arrive as one ticket. The dining room's design ceiling is about 151; the hearth's, as §14.3 established, is about 144.

At 3:40 p.m. the grill cook — the person who owns hearth and plancha, 78% of entrée production — is forty minutes late and their phone has gone to voicemail twice. The sous is already on a double. The schedule was written to a 30% labor target, which is another way of saying it was written with exactly one person at every station and no depth anywhere.

Notice first what has not gone wrong. The fire was lit at 1:30 by the opening prep cook, because Bellwether's opening checklist owns the hearth rather than the grill cook's shift. Had the fire been the grill cook's first task, service would start ninety minutes late and nothing below would matter.

Second, notice the collision on paper. 142 covers is below the hearth's 144-cover ceiling. On the arithmetic this night fits. It does not fit, and understanding why is the point.

🧮 Run the Numbers

The hearth ledger, and what an average hides.

Four no-shows on the à la carte side; the party is contracted and guaranteed at 40. The night lands at 138 covers: 98 à la carte, 40 party.

À la carte entrées: $98 \times 0.90 = 88$. Party entrées: 40, all pre-selected — 14 hearth chicken, 18 pork chop (plancha), 8 squash and grains (oven). À la carte starters and sides: about 44, of which 16 are flatbreads and 11 hearth-finished. Nine dry-aged ribeyes are forecast, each consuming two fire slots rather than one.

Ledger A — the night as booked, with no intervention.

Hour	Ceiling	À la carte entrées	Flatbread	Sides	Party	Ribeye extra	Demand	Util.
5–6	28	3.5	2	1	—	0	6.5	23%
6–7	28	7.5	4	3	—	1	15.5	55%
7–8	28	11.0	5	4	14	4	38.0	136%
8–9	28	10.1	4	2	—	3	19.1	68%
9–9:30	14	6.6	1	1	—	1	9.6	69%
Total	126	38.7	16	11	14	9	88.7	70%

Read the bottom row and the bold row together; the gap between them is the lesson.

Across the whole service the hearth runs at 70% of capacity — barely above the 67% peak figure from a normal night. By any average measure, this kitchen has room.

In the hour from seven to eight it runs at 136%. Demand exceeds the ceiling by 10.0 fire slots. At 2.14 minutes a slot, that is 21 minutes of hearth work that physically cannot happen inside that hour. It lands as delay on whoever is at the back of the queue: a fourteen-minute standard becomes a thirty-five-minute ticket.

The party alone is fourteen hearth items — exactly thirty minutes of the hearth's entire sustained capacity, arriving as a single ticket.

This is why 142 covers broke a kitchen rated for 144. A capacity ceiling computed against an average arrival shape says nothing about a night that arrives all at once. Every capacity number in this book carries that caveat. Write it next to the number.

So: it is 3:41. What does a chef actually do?

The honest answer starts with what the chef does not know. Whether the grill cook is hurt, quit, or stuck on a bus. How many of the 142 will show. Whether the party will bring extra guests. Every option below is chosen against incomplete information, on a clock, and the options expire. At 3:41 all are available. At 7:20, exactly one is.

A — call someone in. The off line cook has Friday as their only night off in a nine-day stretch. Costs overtime, costs a person you need tomorrow at 123 covers, and — the part that decides it — they cannot be in the building before 5:15. Held in reserve.

B — run short. Sous to hearth and plancha; chef at the pass, stepping into sauté when it drowns. Costs nothing on the schedule; costs ticket time and the chef's ability to hold the pass. The default, and it happens whether or not you choose it.

C — cut the menu. 86 the flatbread through the crunch; cap the ribeye at a count. Costs contribution margin and costs some guests the thing they came for. Cheap, reversible, and the only option that returns actual fire capacity.

D — re-engineer the party's fire. Split the 14 chickens into two waves; confirm the first course is cold and can be plated ahead at 6:15 rather than cooked at 6:50. Costs fifteen minutes of planning and one conversation with the FOH partner. Do this first, because it is free.

E — comp the party. No. Nobody has had a bad experience yet, and pre-emptive comping buys nothing except a smaller check.

The chef takes D, then C, then B, holding A in reserve. Here is the night.

FIGURE 14.5 — The second Friday in October, 3:40 p.m. to close
                                             [the Bellwether plan — a MODELED scenario, not a report]

 TIME    EVENT                                       HEARTH LOAD   WHAT IS BREAKING
 ────────────────────────────────────────────────────────────────────────────────────
  3:40   grill cook 40 min late · voicemail twice       idle       nothing yet
  3:41   ▲ THE DECISION WINDOW OPENS                    idle       all five options live
  3:50   sous → hearth + plancha · chef → pass          idle
  4:05   party rewritten: 14 chickens in TWO waves;     idle
         first course confirmed cold, plate at 6:15
  4:20   flatbread 86'd 6:00–8:45 · ribeye capped at 6  idle
  4:30   mise audit — both wood par-cooks counted       idle
  4:40   STRETCH — six minutes, not ten                 idle
  5:00   service opens                                ▓░░░░░  23%   nothing
  6:00   à la carte builds                            ▓▓░░░░  41%   nothing
  6:30   ▲ PARTY SEATS — 40 covers, one ticket        ▓▓░░░░  41%   garde manger
  6:52   party first course lands (plated ahead)      ▓▓▓░░░  50%   garde manger
  7:00   ▲ PARTY WAVE 1 — 7 chickens fire             ██████ 100%   THE HEARTH
  7:12   ▲ PARTY WAVE 2 — 7 chickens fire             ██████ 100%   THE HEARTH
         à la carte hearth queue now 8 slots deep
  7:18   the night's worst ticket enters the rail      ██████ 100%   the pass
  7:30   party entrées all away · queue starts down    ██████ 100%
  7:49   that 7:18 ticket leaves the pass — 31 MIN     ▓▓▓▓▓░  85%   the pass
  8:06   queue clear · ticket times back under 20      ▓▓▓▓░░  70%
  8:20   sauté slowing — four hours, no relief         ▓▓▓░░░  58%   SAUTÉ (fatigue)
  8:45   flatbread back on                             ▓▓▓░░░  54%
  9:30   last seating                                  ▓▓░░░░  32%
 10:40   close · walk-in check · logs · a.m. list        out        the sous (15-hr day)
 ────────────────────────────────────────────────────────────────────────────────────
   HEARTH LOAD = fire slots demanded ÷ 28-per-hour sustained ceiling, by band.
   Bellwether has not opened. This is a plan being stress-tested on paper.

What broke, in what order

Read the right-hand column, because the sequence is not the one people expect.

Nothing breaks for ninety minutes, and that is the trap. From 5:00 to 6:30 the kitchen runs at 23% and then 41% of hearth capacity with three cooks. It feels fine. A chef who reads "fine" at 5:45 and relaxes has just wasted the last window in which Option A was purchasable.

Garde manger breaks first, at 6:30 — not the hearth. Forty plated first courses in a twelve-minute window is roughly 120 minutes of plating compressed into twelve: arithmetically impossible. Which is why the 4:05 decision — confirm the first course is cold, plate it at 6:15, hold it — was the highest-value fifteen minutes of the day. Plated ahead and held at or below 41°F, forty first courses become a delivery problem rather than a production problem. Had the event order specified a hot first course, that conversation needed to happen at 3:45. (Chapter 29 builds the event order; Chapter 25 owns the cold-holding rules that make plating ahead legitimate rather than reckless.)

The hearth breaks at 7:00 and stays broken for thirty minutes. Fourteen chickens at 28 an hour is exactly thirty minutes of fire. Splitting them into two waves does not create capacity — worth being blunt about, because wave-splitting gets sold as a fix and is not. It moves the pain: the party eats within a twelve-minute spread instead of twenty-five, and the à la carte queue absorbs the difference.

The pass breaks at 7:18, when the night's worst ticket enters the rail, cannot start until 7:39, and leaves at 7:49 — 31 minutes against a 14-minute standard. Eleven entrée tickets exceed the 22-minute not-to-exceed, all between 7:02 and 7:58.

Sauté breaks at 8:20, from fatigue rather than volume. One person has covered pasta, the grain plate, and every sauce for the hot line since 5:00 with no relief, because the chef who normally steps in has been at the pass all night holding a queue together. This is the compounding failure, and the one that never appears in a post-mortem, because the volume numbers for 8:00 to 9:00 look mild.

🧮 Run the Numbers

What the 3:41 decisions actually bought, and what they cost.

Ledger B — the same night, after the decisions. Flatbread 86'd from 6:00 to 8:45 (12 of the night's 16 removed). Ribeye capped at 6 (three fewer, three fewer double slots). Party split into two waves — which, note, changes nothing at the hour level.

Hour	Ceiling	Demand as booked	After the cuts	Util.
5–6	28	6.5	6.5	23%
6–7	28	15.5	11.5	41%
7–8	28	38.0	31.0	111%
8–9	28	19.1	15.1	54%
9–9:30	14	9.6	9.6	69%
Total	126	88.7	73.7	59%

The peak hour falls from 136% to 111%; the overrun falls from 10.0 slots to 3.0 — from 21 minutes of impossible work to about 6.

And the night still produced a 31-minute ticket. Both are true, and the gap between them is the most useful thing in this chapter: an hour is the wrong unit for a station that behaves like a queue. Inside the 7:00–7:30 half-hour the hearth was 100% committed to the party, so à la carte items ordered in that window had a capacity of zero, not 111%. Averages smooth. Queues do not.

Now price the decisions. Contribution margins from the Chapter 11 and 12 cost cards: flatbread \$16.00 menu / \$4.35 plate / CM \$11.65 · ribeye \$54.00 / \$21.60 / CM \$32.40 · hanger \$34.00 / \$10.20 / CM \$23.80 · cold starter \$14.00 / \$3.60 / CM \$10.40.

Line	Amount
Flatbread 86: 12 units lost; 6 tables substituted a cold starter	\$77.40
Ribeye cap: 3 units lost; all three substituted the hanger	\$25.80
Three comped courses (\$84 of menu price; the cost is the food) | \$27.00
Sous overtime: 2.5 hrs at time-and-a-half (\$26 base → \$39)	\$97.50
Prep cook called in three hours early Saturday, at \$19 loaded | \$57.00
Grill cook's 8-hour shift, not worked and not paid, at \$21 | (\$168.00)
Net hard cost of the night	\$116.70

Sales for the night: 98 à la carte covers at a \$46 check is \$4,508, plus 40 party covers at a contracted \$58 a head is \$2,320. \$6,828.

The night that broke the kitchen cost \$116.70 — one and seven-tenths percent of its own sales — and the restaurant came out ahead on wages. That is not a comforting result; it is the point. Everything that actually mattered is invisible on that table: a sous who worked fifteen hours, eleven tables who waited past the standard and will decide over the next three weeks whether to come back, a sauté cook alone for four hours, and a schedule with no depth at any station.

The financial statement will record a good Friday. Chapter 19 will show you what the schedule cost. Chapter 21 will show you what the sous cost.

The transferable lesson is not "have a backup cook." Most independents cannot afford one, and pretending otherwise makes a book useless. The lesson is narrower and actually usable:

A kitchen's resilience is decided by which decisions are still available and for how long. At 3:41 the chef held five options. At 6:30, three. At 7:20, one — and it was the one that costs the guest. Every system in this chapter — the production sheet, the 4:30 mise audit, the 3:00 par-cook count, the written ticket-time standard, the 86 board — exists to move the moment of discovery earlier, which is the same thing as keeping options alive.

14.5 Ticket times: measuring them, and what a rising number is telling you

Ticket time is the elapsed interval from a ticket reaching the kitchen to the last plate of that course leaving the pass. It is the kitchen's most useful operating metric and the one most commonly measured wrong. Three definitional traps, each of which makes your number a fiction:

Trap one: measuring from order rather than from arrival at the pass. If a server takes an order at 7:04 and rings it in at 7:09, the five-minute delay is somebody else's. Measure the kitchen from when the kitchen could have started.

Trap two: measuring to "plate up" rather than to "sold." A plate sitting under a lamp for six minutes waiting for a runner is a service failure that a badly configured kitchen display system records as a triumph. Measure to the moment food leaves the pass.

Trap three: mixing courses. Starters and entrées have different standards and different failure modes; a single blended number tells you nothing you can act on.

The standard

Write it down, post it, and make it a number rather than an adjective.

Where does fourteen minutes come from? Not from a book — from the menu. The Hearth Chicken is eight minutes from par-roast, the longest routine hearth item about ten, plating and checking two to three, and the pass needs a minute of slack to hold a table together. Fourteen is what this menu on this equipment does repeatably. A standard that does not derive from your own menu is decoration.

🧾 Read the Numbers

```text FIGURE 14.6 — "The ticket-time report, week 14" [the Bellwether plan] THE ARTIFACT Weekly kitchen ticket-time summary, entrée course only, pulled from the POS/KDS. Five dinner services. Standard 14 min, not-to-exceed 22 min. Ribeye tickets excluded and reported on a separate line. THE CONTEXT Bellwether, week 14. The week ran 497 covers against a plan of 475 — 4.6% over — and the entire overage landed in one service, which included a contracted 40-guest party seated at 6:30 as a single ticket.

 SERVICE   COVERS   ENTRÉE    MEDIAN   75th   95th   OVER 22   HEARTH UTIL.
                    TICKETS                          MIN       (peak hour)
 ──────────────────────────────────────────────────────────────────────────
 Tue         62       24       12      15     18       0          38%
 Wed         78       30       13      16     20       1          49%
 Thu         92       35       14      17     23       2          63%
 Fri        138       35       16      24     31      11         111%
 Sat        123       47       15      19     26       5          82%
 ──────────────────────────────────────────────────────────────────────────
 WEEK       493      171       14      18     26      19
 Ribeye tickets (all services): 22 tickets, median 25 min, 2 over the 30-min NTE.

WHAT IT SHOWS Four of five services are healthy: median at or under the standard, 95th percentile inside or barely outside the not-to-exceed. Friday is not a degree worse; it is a different animal. Note the column that explains it — Friday carried the MOST covers and among the FEWEST tickets, because 40 covers arrived as one. A kitchen is loaded by tickets and simultaneity, not by covers. The 95th percentile is what moved (18→31 across the week's range) while the median barely did (12→16): a shape failure, not a capacity failure. Saturday, at 123 covers on a normal arrival curve, handled more food than Friday's à la carte side with half the pain. WHAT IT DOESN'T It does not say WHY Friday broke — a report identifies the night, never the cause. It cannot see that a station was covered by the sous, that the flatbread was 86'd for nearly three hours, or that the ribeye was capped; all three moved these numbers and none appears here. It does not capture what the guest experienced: a 31-minute ticket quoted at the table is a different event from an unquoted one. And it says nothing about quality — a kitchen can drive this report green by sending food that is not ready. THE DECISION Do not change the standard and do not add labor on one service. Do two things: (1) cap simultaneous party seating so no single ticket exceeds ten hearth items — which means seating a 40-top in two waves or moving its menu off the fire; (2) pull the eleven over-22 tickets against the reservation log. If they cluster in one 40-minute band it is shape; if they are spread it is staffing, and that is Chapter 19's problem. THE LESSON Watch the 95th percentile, not the average. The average tells you whether the kitchen is adequate. The tail tells you how many guests had a bad night, and guests do not experience averages. ```

Reading a rising number

A ticket-time report is a diagnostic instrument, and it is nearly useless unless you know which disease each pattern indicates. The five patterns that matter:

The median rises across every service. Something structural changed: a menu change added a longer item; a station lost an experienced person; portioning drifted; a piece of equipment is underperforming. Structural causes need structural fixes and do not go away on a slow night.

Only the 95th percentile rises. A shape problem. Capacity is fine; the arrival pattern is not. Look at the reservation book, not the line — a party, a compressed seating window, a host stand seating four tables in six minutes. The fix is almost always in the dining room. Chapter 22 owns it.

One station's items rise and the others don't. That station: under-staffed, out of mise, or a cook struggling. The easiest pattern to act on and the one operators most often miss, because they look at the ticket total instead of the item detail.

Times rise while covers are flat. A prep failure, almost always. A par-cook ran out and the kitchen is cooking from raw — at Bellwether, a Hearth Chicken from raw is a thirty-five-minute proposition against an eight-minute one. Check the production sheet against what was actually made.

Times fall while comps and remakes rise. The most dangerous pattern on the list. The kitchen is meeting the number by sending food that is not ready — under-rested steaks, under-cooked centers, plates going before the last component is right. This is what happens when a ticket-time standard becomes a performance target rather than a diagnostic, and Case Study 2 works it out properly.

That last pattern is the honest limit of everything in this section. Ticket time measures speed, and speed is a proxy for the thing you care about, which is a guest having a good meal. Proxies degrade when you push on them. A kitchen judged only on ticket time will find ways to make ticket time look good, and every one of those ways costs something that does not appear on the report. Read it next to comps, remakes, and the review score, or do not read it at all.

14.6 86ing, all-day, and managing scarcity mid-service

Two pieces of kitchen vocabulary do the work of an entire inventory system during service, and both are management tools rather than jargon.

86ing is declaring an item unavailable — removing it from what the kitchen will accept. The origin of the term is disputed; what matters is that it is a decision, made by a person, at a moment, with consequences in both directions. All day is the running total of an item across every open ticket: if three tickets call for chicken — one for two and two for one each — the chef calls "four chicken all day," and every cook knows the true demand rather than the demand on the ticket in front of them.

All-day exists because a station cook can only see their own tickets. It converts a stack of independent orders into a production queue, and it makes batching possible mid-service: four chickens all day can go on the fire as a group; four chickens called separately cannot.

FIGURE 14.7 — The board at 7:22 p.m., second Friday in October       [the Bellwether plan]

   ┌─ 86 BOARD ──────────────────────┐   ┌─ COUNTS ────────────────────┐
   │  FLATBREAD    (back at 8:45)    │   │  RIBEYE ............  2 left │
   │  SOFT-SHELL   (out of season)   │   │  HALIBUT ...........  5 left │
   │  ─ do not sell. tell the guest  │   │  CHICKEN ...........  9 left │
   │    what we DO have. ─           │   │  ─ call it at 3 ─           │
   └─────────────────────────────────┘   └─────────────────────────────┘

   ┌─ ALL DAY ───────────────────────────────────────────────────────────┐
   │  CHICKEN ......... 5      HANGER .......... 3      PASTA ........ 4 │
   │  PORK ............ 4      SQUASH .......... 2      RIBEYE ....... 1 │
   │  ── hearth all-day: 5 chicken + 1 ribeye(×2 slots) = 7 fire slots ──│
   │  ── plancha all-day: 4 pork + 3 hanger = 7 finishes ───────────────│
   └─────────────────────────────────────────────────────────────────────┘

   Seven fire slots is fifteen minutes of hearth at the sustained ceiling. Any
   ticket landing now is at least fifteen minutes out before it starts cooking.
   That is the number the pass tells the servers, and it is why the board is on
   the wall where servers can see it rather than in the chef's head.

The board is the whole scarcity-management system, and it is four square feet of whiteboard. Note the element most kitchens omit: the counts. An 86 is binary and final. A count — "nine chicken left, call it at three" — is information, and information lets the dining room sell differently for the next forty minutes instead of finding out at 8:40 that the signature dish is gone.

The 86 decision

86ing is usually treated as something that happens to a kitchen. Under load it is a lever, and it should be pulled deliberately with the arithmetic in front of you. There are exactly three reasons:

1. You have run out. No decision involved; the only question is how fast the information reaches the floor.
2. Quality has failed. The product is not right. This should be non-negotiable and the chef should never have to defend it.
3. Capacity. The item is consuming production you need elsewhere. This is the interesting one, and it is the one operators get wrong, because it feels like giving up revenue and it is usually buying more than it costs.

⚠️ Where the Money Leaks

The 86 that cost more than it saved — and the one that didn't.

Two versions of the same decision, on the Friday.

The good 86: the flatbread. A starter, priced at \$16, competing for entrée fire in the exact window when entrée fire is the scarcest thing in the building. 86ing it from 6:00 to 8:45 removed twelve units and about five fire slots from the peak hour.

• Cost: twelve units of \$11.65 contribution margin, six of which substituted a cold starter at \$10.40 — a real loss of $6 \times \$11.65 + 6 \times \$1.25 = \$77.40$.
• Bought: five fire slots in the peak hour — 10.7 minutes of hearth time — applied to the ticket queue at its worst moment; in the modeled night, the difference between a peak-hour overrun of 21 minutes and one of about 6.
• The trade: \$77.40 against roughly fifteen minutes of ticket time on the tables waiting longest. Take that trade every time.

The bad 86: 86ing the chicken at 8:40 because you ran out. The par-roast par was set to a 120-cover Friday and the night ran 138. From 8:40 the kitchen is selling a menu without its signature item — a Star on the Chapter 12 matrix, the largest contribution-margin item at \$20.48 — and servers are apologizing at every table.

• Cost: eight or nine unsold chickens' contribution margin, call it \$175, plus an unknown and larger number in second visits that will not happen.
• Bought: nothing. It was not a decision; it was the absence of one, made at 8:40 by arithmetic that should have happened at 8:05 that morning.

The distinction that matters: a capacity 86 is a decision made in advance with a number attached. A stock-out 86 is a production-sheet failure disguised as a decision. The first is management. The second is what management prevents.

Announcing scarcity well

An 86 reaches the guest through a server. "Sorry, we're out of the flatbread" describes a restaurant that has failed. "The kitchen's holding the flatbread until about a quarter to nine — but the whipped ricotta with the hearth bread is what I'd order anyway" describes a restaurant running its night on purpose. Identical fact, opposite impression, zero cost difference.

So the 86 board is only half the system. The other half is the twenty seconds at the stretch where the chef tells the floor what to sell instead. A kitchen that 86s an item without naming its replacement has done the hard operational work and thrown away the easy hospitality work.

14.7 Production planning: prep lists, par cooking, batch sizes, and the waste tradeoff

Everything hard about a service is decided in prep, and prep has exactly three decisions in it: what to make, how far to cook it, and how much to make at once.

Par cooking

Par cooking is partially cooking a component during prep so it can be finished quickly to order during service. At Bellwether it appears in three forms:

Read the right-hand column, because it is the entire argument. A Hearth Chicken from raw is a thirty-five-minute proposition against a fourteen-minute standard. If the par-roast runs out at 8:10 the chicken is not slow — it is effectively 86'd, because a kitchen cannot produce a thirty-five-minute item inside a fourteen-minute standard without destroying every other ticket on the rail.

This is what §14.3 meant by calling the par-roast a dependency. It is not a shortcut or a convenience; it is a structural precondition for the menu, the ticket-time standard, and the two-person hot line all being simultaneously true. Which makes the par-roast count the most consequential number on the production sheet, and is why Figure 14.2 flags a physical count at 3:00 p.m. rather than a glance at 5:30.

Par cooking buys: speed (a 2–8 minute finish instead of 11–35); consistency (a controlled batch varies less than 38 pieces cooked under load); and labor structure — it moves work out of the most expensive, least flexible hours into cheap, flexible prep hours, which is a prime-cost trade inside the same total.

It charges: committed inventory (a par-roasted chicken is a decision you cannot un-make); a shorter quality window; a shorter safe holding window with real regulatory teeth; and the risk of a shortfall that cannot be recovered mid-service.

⚖️ Code and Compliance

Par cooking is a food-safety control point, not just a production technique.

Every par-cook creates a cooked, cooled, held, and reheated product — precisely the sequence where temperature failures cause illness. The FDA Food Code framework, which states, counties, and cities adopt in their own versions (so verify what your jurisdiction has actually adopted), covers this directly:

• Cooked food must be cooled through the temperature danger zone within specified time limits, in two stages, and the Food Code sets those limits explicitly. Product depth, container choice, and whether you use an ice bath or a blast chiller decide whether you meet them.
• Cold holding is at or below 41°F; hot holding at or above 135°F; poultry is cooked to 165°F. A par-roasted chicken finished on the hearth must reach 165°F every time, verified with a calibrated probe — not judged by eye.
• Reheating for hot holding has its own requirement, and it is not the same as cooking.
• The forty plated first courses held from 6:15 to 6:52 in the Friday scenario are legitimate only because they are cold and held at or below 41°F. The same plan with a hot first course is not a scheduling question; it is a violation waiting for an inspector.

There is also a records dimension: a par-cook program without a cooling log has no evidence it worked, and in an inspection an undocumented control is treated as an absent one.

Chapter 25 builds all of this properly, including HACCP and the inspection itself. The operations principle belongs here: a production decision that changes a food's temperature history is a food-safety decision and must be designed with both hats on at once. Verify your local code and use a certified manager to design the program.

Batch size

Batch cooking is producing a component in quantity rather than to order. Every batch decision trades labor against waste, and the trade is small enough per item that nobody bothers to compute it — which is exactly why it is worth computing once.

🧮 Run the Numbers

Two production questions, worked.

Question one: the salsa verde batch cadence.

Bellwether's week is 475 covers. At 0.90 attachment that is 427 entrées; the Hearth Chicken runs 22.4% of entrées, so 96 portions of salsa verde a week — \$100.80 of product at \$1.05 a portion. (That 96 is the same 96 the purchasing chapter orders birds against and the costing chapter builds its extract on; when four chapters agree on a unit count, the operating model is real.) A batch yields 1.25 quarts (20 portions) and takes about 25 minutes of hands-on labor. Salsa verde is bright for two days and dull by four. Prep labor is \$19 an hour, fully loaded.

Cadence	Labor per week	Est. quality waste	Total cost
Once weekly (4.8 batches at once)	120 min → \$38.00 | ~6% → \$6.05	\$44.05
Twice weekly (2.4 batches × 2)	120 min → \$38.00 | ~2% → \$2.02	\$40.02
Daily (0.96 batches × 5)	5 × 15 min = 75 min → \$23.75 | ~1% → \$1.01	\$24.76

Daily wins, which is counterintuitive and worth seeing why: the work that dominates this task — picking and washing herbs, portioning, cleaning the processor — scales with volume, not with the number of times you start. Making a week's worth at once does not make the herbs pick themselves. Batching is not automatically cheaper. Compute it for the item rather than assuming. Daily beats weekly by \$19.29 a week — **\$1,003 a year on one sauce.**

The real point: there are perhaps forty of these decisions on a production sheet. If the average one is worth \$300 a year and half are made wrong, that is **\$6,000 a year decided by people making twenty-second choices about container size**, against a plan operating profit of about \$261,020 before debt service. This is Chapter 1's cost-drift mechanism observed at its source.

Question two: what the Hearth Chicken really costs to produce.

The frozen Chapter 11 cost card is unchanged: 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, plus a 2% waste allowance = \$8.52**, priced at **\$29.00 for a 29.4% food cost and a \$20.48 contribution margin.

A cost card prices ingredients. It does not price production. Overlay the labor: butchering, brining, par-roasting, and chilling runs about 4.5 minutes of hands-on prep per whole bird in batches. At \$19 an hour — \$0.3167 a minute — that is \$1.43 a bird, and one bird yields two portions: \$0.71 a plate.

Labor-loaded plate cost: $\$8.52 + \$0.71 = \$9.23$.** Against \$29.00 that is a 31.8% cost and a \$19.77 contribution margin** — not 29.4% and \$20.48.

This does not change the cost card. The card is right and the price stays at \$29. What it changes is how you compare items to each other: a menu-engineering decision made on cost cards alone is systematically biased toward labor-intensive dishes. Chapter 12's matrix is still the right tool; this is the correction to carry into it.

Sizing the par: the honest version

Return to the production sheet. The chicken par was 46 half-birds — the Friday forecast of 28, plus the party's contracted 14, plus four of cushion. Actual need was 38: 24 à la carte plus the party's 14. So the kitchen over-produced by eight half-birds, four whole birds at \$11.20 each: **\$44.80 of committed product.**

Is that a failure? It depends entirely on what happens next, and this is the part the textbooks skip:

• Friday into Saturday, where Saturday needs about 29 chickens, the eight carry over and sell. The over-par costs approximately nothing.
• Saturday into Tuesday, with two dark days and a 62-cover Tuesday, the same eight half-birds are outside their quality and safe-holding window before they can be sold. That is \$44.80 of waste plus the labor already spent on it.

The same par is prudent on Friday and expensive on Saturday, and any par-setting rule that ignores the shape of the week — 62 / 78 / 92 / 120 / 123 — is wrong two nights out of five. Write the day into the coverage factor: 15% cushion into an up-day, 5% into a down-day, zero into the last service before a dark day.

🔍 Check Your Understanding

1. Why does a Hearth Chicken from raw effectively 86 itself during service, even though the kitchen has raw chickens in the walk-in?
2. The chef proposes moving the pork chop from a wood par-cook to fully à la minute on the hearth, arguing it will taste better. Using Figure 14.4, what happens to the hearth station and what would you need to see before saying yes?
3. A cook sets the same salsa verde par on Tuesday and Saturday. Name both ways that is wrong and say which is more expensive.

(1: 35 minutes cannot be produced inside a 14-minute standard without wrecking every other ticket on the rail; availability of raw product is not availability of the dish. 2: Twenty-nine plancha finishes at ~2.5 min each become 29 live-fire cooks at 11–14 minutes, adding roughly 350 minutes to a station already at 211 — a second cook, and the 28-per-hour ceiling blown at far fewer covers. You would want a blind tasting proving the quality difference is real, the new cover ceiling, and Chapter 19's cost for the second position. 3: Building to Saturday over-produces on Tuesday; building to Tuesday runs out on Saturday. Running out is more expensive, because it converts a food-cost error into a revenue and guest error — but over-prep is the one that hides, because nobody complains about it.)

14.8 Closing: cleaning, logs, the walk-in check, and setting up tomorrow

The close is the most-skipped hour in the industry and the cheapest to fix, because everything in it is a checklist and none of it requires judgment. Four jobs, in priority order.

One: break down and clean. Surfaces cleaned and sanitized at the correct concentration and contact time, hood filters on schedule, floors done, the fire safely dead down. Chapter 25 owns the sanitation standards; the operations point is that a kitchen cleaned at 11:00 p.m. by tired people is only as good as its checklist and sign-off, and a closing checklist without a name on it is a wish.

Two: the logs. Temperature logs for every refrigerated unit, the cooling log for anything par-cooked, the waste log, the closing notes — the only artifact proving a control was operating, and the first thing an inspector asks for.

Three: the walk-in check. Every night, by a person, with their eyes open.

Four: set up tomorrow. Count what is on hand, write the prep list, leave it on the board.

👨‍🍳 On the Line

The walk-in check, and the Tuesday morning it would have prevented.

Four minutes, five things:

1. Read the thermometer and write the number down. Not "it looked fine" — the number, on the log, with a time and initials. A unit reading 41°F tonight, 43°F tomorrow, and 45°F the night after is a compressor telling you it is dying, in a language only a written log can speak.
2. Look at the storage order. Ready-to-eat on top; raw proteins below, ordered by required cooking temperature, whole poultry on the bottom. Thirty seconds, and among the most common serious findings in an inspection.
3. Check that everything is covered, labeled, and dated. An unlabeled container is unusable product — tomorrow you will not know what it is or how old it is, and neither will an inspector.
4. Pull anything past its window onto the waste log tonight, while you still know what it was.
5. Look at the floor and the door gasket. Standing water, a gasket that no longer seals, and a propped door are the three physical causes behind most temperature drift.

Now recall the surprise inspection this book keeps returning to — a Tuesday, 10:40 a.m., mid-prep, and the inspector finds 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, and one employee without a food-handler card. It ends in a re-inspection rather than a closure, which is lucky rather than skillful.

Set that against the five above. Items one, two, and three of that inspection are items one, two, and four of this check. A 46°F walk-in at 10:40 on Tuesday was not 41°F at midnight on Saturday; a refrigeration unit does not fail by four degrees between breakfast and lunch without something else being wrong. The failure was visible on Saturday night and nobody was looking, because the close is the shift where everyone is tired and it feels like the day is already over.

The day is not over. The close is where tomorrow's food cost, tomorrow's ticket times, and your next inspection are decided — by people paid \$19 an hour at 11:15 at night, who will do the check properly if and only if there is a checklist, a signature line, and a manager who reads it in the morning.

FIGURE 14.8 — The closing kitchen log, one night                   [the Bellwether plan]

  TEMPERATURES     walk-in cooler 10:52 __°F · walk-in freezer 10:52 __°F
                   low-boy hearth __°F · low-boy sauté __°F · garde manger __°F
  COOLING LOG      chicken par-roast 34 ea · pork 21 ea · hanger 15 ea ... y / n
  WALK-IN VISUAL   storage order · covered, labeled, dated · floor and gasket ... y / n
  WASTE LOG        items pulled tonight, and why ......................... attached
  COUNTS FOR A.M.  chicken par-roast __ ea · pork / hanger __ / __ ea · salsa __ qt
  CLOSING NOTES    what broke tonight, what the a.m. needs to know
                   ______________________________________________________________

  CLOSED BY ______________   READ BY (a.m.) ______________   time ______

  The last line makes the rest of the page real. A log nobody reads in the morning
  is a log that gets filled in from memory by the third week.

That final signature line is the difference between a system and a ritual. A closing log is a control only if somebody reads it the next morning. The moment the crew learns nobody reads it, the numbers become approximately correct, then habitually correct, then fiction — and the fiction is worse than no log at all, because it produces false confidence and looks like compliance to everyone including you.

🍽️ The Business Plan

Checkpoint 14 of 40 — the Kitchen Operations section.

The plan has a menu (Chapter 10), cost cards (Chapter 11), a margin ranking (Chapter 12), and a purchasing system (Chapter 13). This chapter contributes the section proving the restaurant can actually produce what it has promised to sell.

What goes into the plan.

The finding this section contributes, which the plan did not have before: the binding constraint on Bellwether's revenue is the wood fire, not the dining room. The floor plan allows about 151 covers. The hearth allows about 144 at the current menu mix, and about 133 if hearth items drift to half the entrée mix. The plan's ceiling is therefore a menu-mix-dependent number and should be stated that way rather than as a seat count — because a seat count implies a fixed capacity this restaurant does not have.

What this checkpoint does not settle.

• What this line costs. The above is a production model, not a labor model. The 32.3% target, the hours behind these four positions, and whether the schedule can carry any depth are Chapter 19's — and the Friday scenario is a direct question posed to it.
• Whether the room can pace to it. A 28-an-hour fire requires a dining room that does not seat forty covers into one twelve-minute window. Table management and pacing are Chapter 22's, and the Friday is a bill sent to that chapter.
• Whether the party is good business. A 40-guest event is high-margin revenue on paper and a capacity event in practice. Chapter 29 prices it; this chapter only showed what it does to the fire.
• Whether the par-cook food-safety program is adequate. Chapter 25 builds it. This section states the requirement and the logs; it does not validate the program.

Open questions carried forward.

1. Can a four-person peak line hold a 14-minute entrée standard at 120+ covers with no depth at any station, and what does depth cost? (Chapter 19)
2. What booking rule protects a 28-an-hour fire — a cap on simultaneous party size, staggered seating, or a party menu specced off the hearth? (Chapters 22, 24, 29)
3. Is the dry-aged ribeye worth its two fire slots and its separate standard? (Chapters 12, 24)
4. Does the hearth's 144-cover ceiling constrain the \$1.55M year-one forecast, which assumes about 95 covers a night? (Chapters 24, 32)
5. What does the par-cook program require in cooling capacity, logs, and certification before it is defensible to an inspector? (Chapter 25)

Conclusion

A service looks like chaos and is not. It is a production system with a hard hourly ceiling, a demand curve nobody controls, and a set of decisions that are cheap in the afternoon and unavailable by seven o'clock.

This chapter did three things with that. It made the ceiling computable — four positions on a grate, eight minutes of dwell, 28 items an hour, and from there a 144-cover cap the floor plan never mentioned. It made the prep system auditable, by insisting every par trace to a forecast rather than a habit. And it walked a Friday in which 142 covers, against a kitchen rated for 144, produced a 31-minute ticket — not because the kitchen lacked capacity, but because forty of those covers arrived in a single instant. A ceiling computed against an average arrival shape says nothing about a night that arrives all at once.

The Friday's other lesson is smaller and more useful than "hire more people." At 3:41 the chef held five options and used the three that were free or cheap. By 7:20 only the expensive one remained. The systems here are not about doing more work; they are about learning things earlier, because knowing earlier is the same as having choices.

And the night's hard cost was \$116.70 against \$6,828 of sales, which should worry you rather than reassure you. The expensive parts — a sous on a fifteen-hour day, eleven tables past the standard, a sauté cook alone for four hours — appear on no report in this chapter.

Chapter 15 moves fifteen feet left, to the bar. The kitchen is where the product is made and most of the cost lives; the bar is where the margin lives and the leaks are largest, because the product is liquid, portable, popular, and easy to give away. The discipline is the same — count the thing, write the standard, read the variance — but the arithmetic is better and the temptations are worse.

Key Terms

Mise en place — "everything in its place": the complete set of prepared components, tools, and station set-up a cook needs to execute the menu at service speed. Best understood as an auditable production system tied to a forecast, not as a personal virtue. (Ch. 14)

Prep list — the day's task list for the prep function: which items to produce, in what quantity, by whom, by when. (Ch. 14)

Production sheet — a prep list plus its arithmetic: par level, amount on hand, amount to make, recipe yield, and the forecast the par was derived from. A prep list can be executed; only a production sheet can be audited. (Ch. 14)

Ticket time — the interval from a ticket reaching the kitchen to the last plate of that course leaving the pass. Measured by course, from arrival at the pass to "sold," reported by median and 95th percentile rather than average. (Ch. 14)

Expediting — the coordination function at the pass: receiving and sequencing tickets, calling them to the stations, controlling when each component fires, checking every plate, and deciding who waits when demand exceeds capacity. (Ch. 14)

86ing — declaring an item unavailable. Legitimate for three reasons — a stock-out, a quality failure, or a deliberate capacity decision — and only the third is management. (Ch. 14)

All day — the running total of an item across every open ticket. Converts a stack of independent orders into a single production queue and makes mid-service batching possible. (Ch. 14)

Batch cooking — producing a component in quantity rather than to order. Every batch-size choice trades prep labor against quality waste, and the trade must be computed per item, not assumed. (Ch. 14)

Pre-shift / stretch — the short standing meeting before service covering the book, the 86s and counts, menu changes, one correction, and assignments. Ten minutes, because the whole crew is being paid for it at once. (Ch. 14)

Spaced Review

1. Bellwether's hearth sustains 28 items an hour. Derive the cover ceiling from the entrée mix and the starter load, and explain in one sentence why 142 covers broke a kitchen rated for 144.
2. A ticket-time report shows a flat median, a 95th percentile risen from 19 to 29 minutes over three weeks, and unchanged covers. Most likely cause, and which chapter fixes it?
3. From Chapter 11: the Hearth Chicken's cost card is \$8.52 against a \$29.00 price; this chapter added \$0.71 of production labor. Does the cost card change? What does change, and what decision should now be made differently?
4. From Chapter 13: you run out of par-roasted chicken at 8:10 on a Saturday. Name two upstream controls — one from Chapter 13, one from this chapter — and the hour each would have caught it.
5. The recurring question: an operator moves the pork chop from a wood par-cook to fully à la minute over live fire because it tastes better. Trace the consequence through station load, the hourly ceiling, the cover ceiling, ticket times, and prime cost. Which half of prime cost goes up, and what would you measure to know whether the trade was worth it?