Chapter 31: Restaurant Accounting: P&L Statements, Cash Flow, Prime Cost, and the Numbers You Must Know Weekly

"Show me the statement. No — show me the one you made, on Monday, for last week." — constructed; what a good operator says when you tell them how the quarter went

Overview

Thirty chapters have each handed you a number. The concept chapter gave you a check average. The lease chapter gave you an occupancy figure and, buried in a clause, an escalation. The costing chapters gave you a food cost target and a signature dish that runs 29.4%. The staffing chapter gave you a roster. The labor chapter gave you a schedule with 453.5 hourly hours on it. The compliance chapter gave you a classification problem worth twenty-seven thousand dollars a year. The technology chapter gave you a fee line that is now the third-largest non-labor expense in the building.

This chapter puts them on one page and adds them up.

That is a less glamorous job than it sounds and a much harder one. Because here is what happens when you actually try: they don't agree. The staffing plan says thirty-one people. The labor model says twenty-four positions. The revenue forecast says \$1,410,760 if you multiply covers by check, and \$1,550,000 if you read the plan's top line. The pro forma holds rent flat for three years while the lease says it steps up in year three. Two different chapters split labor into fixed and variable and got answers \$60,105 apart.

None of that is unusual. Every business plan I have ever read — including several I wrote — contains numbers built by different people, at different times, for different purposes, that were never made to foot against each other. What separates a plan you can operate from a plan you can only present is whether somebody sat down and did the reconciliation, in public, with the disagreements named.

So that is what we are doing. This chapter teaches restaurant accounting — the P&L, the chart of accounts, cost of goods sold, controllable versus non-controllable, EBITDA, the weekly prime-cost calculation Chapter 1 promised you, the flash report, the 13-period calendar, and what comps and voids do to every number above them. And it teaches all of it by using it, on Bellwether, to resolve six specific conflicts that thirty chapters have quietly left on the table.

The reconciliation is the lesson. An operator who can only read a clean statement is not an operator. Statements are never clean.

In this chapter, you will learn to:

• Read every line of a restaurant P&L and say what is in it, what is not, and where the number came from.
• Build a foodservice chart of accounts, and explain why the same expense in two different accounts produces two different — and both defensible — occupancy percentages.
• Compute COGS the honest way, with the inventory adjustment, transfers, employee meals, and comped product at cost, and show why "invoices divided by sales" can swing seven points on a week where nothing changed.
• Compute prime cost weekly, from seven inputs, in under two hours.
• Design the one page you will read every Monday for the rest of your working life.
• Reconcile conflicting figures across a plan and publish the mapping rather than the average.

Learning Paths

🏗️ Opening — this is the chapter where your plan either foots or doesn't. Work §31.1, §31.5, and the Business Plan checkpoint by hand. The six reconciliations are a template for auditing your own plan before somebody else does. 📋 Managing — §31.5 and §31.6 are your job description. If you manage someone else's restaurant, the weekly flash report is the artifact that proves you are managing rather than reacting. Build it in week one. 🍸 Beverage — watch the beverage lines all the way through §31.3: transfers between kitchen and bar are where pour cost quietly becomes food cost and nobody notices. §31.8's comp arithmetic hits the bar hardest. 🚚 Small Format — every structure here scales down. A truck has a P&L, a chart of accounts, a prime cost, and a sales-tax liability. What it does not have is occupancy at 6.1%, which is why §31.5's occupancy caveat matters more to you, not less.

31.1 The restaurant P&L, line by line

A profit and loss (P&L) statement — also called an income statement — reports revenue and expenses over a period and arrives at profit. Every business has one. What makes a restaurant P&L different is the ordering: it is deliberately built so that the costs you can move this week sit at the top, and the costs you agreed to years ago sit at the bottom.

The National Restaurant Association publishes a Uniform System of Accounts for Restaurants (USAR), which exists precisely so that one operator's "other operating expense" means roughly the same thing as another's. You do not have to adopt it line for line. You do have to adopt something consistent, and if you ever intend to compare your numbers against published industry ranges, you should adopt something close to the standard, because a benchmark computed on a different account structure is not a benchmark. It is a coincidence.

Here is the shape.

FIGURE 31.1 — The order of a restaurant P&L                [structure; dollars are illustrative]

   REVENUE                          food + beverage + other, NET of comps and discounts
     less  COST OF SALES            food and beverage product actually consumed
   = GROSS PROFIT
     less  LABOR                    wages, salaries, overtime, taxes, workers' comp, benefits
   ─────────────────────────────────────────────────────────────────────────────────
   ( COST OF SALES + LABOR = PRIME COST — the number Chapter 1 said predicts survival )
   ─────────────────────────────────────────────────────────────────────────────────
     less  OTHER CONTROLLABLE       supplies, marketing, R&M, utilities, direct operating
   = CONTROLLABLE INCOME            what the person running the building is accountable for
     less  OCCUPANCY                base rent, NNN, property tax and insurance
     less  NON-CONTROLLABLE OP.     contracted tech, insurance policies, equipment rental
     less  GENERAL & ADMIN          accounting, legal, licenses, bank fees, office
   = RESTAURANT OPERATING INCOME    also reported as EBITDA
     less  DEPRECIATION & AMORT.    the build-out and the equipment, expensed over time
     less  INTEREST                 the interest portion of debt service only
   = PRE-TAX INCOME
     less  INCOME TAX               depends entirely on your entity (Chapter 8)
   = NET INCOME

Notice three things about that order, because they are the whole design.

First, prime cost falls out of the top of the statement. You do not have to compute it separately; it is where the two largest lines meet. A statement that buries labor below occupancy — and plenty of generic accounting software will do exactly that — has hidden the number you manage by.

Second, "controllable income" is a real line with a real purpose. It is what a general manager should be measured on, because it stops before rent, insurance, and the loan — three things a manager did not negotiate and cannot change. Chapter 40 comes back to this when it discusses prime-cost bonuses.

Third, debt service does not appear anywhere as a single line. It splits: the interest portion is an expense, the principal portion is a balance-sheet transaction that never touches the P&L at all. This is the single most common place where a first-time operator's arithmetic falls apart, and §31.9 walks it.

The top line is not one number — it is two, and you must say which

Before we put dollars into that structure, we have to resolve the first conflict, and it sits at the very top.

Bellwether has two revenue bases in circulation, both correct, both used in earlier chapters, and they differ by \$139,240.

The first is the base grid: seats, turns, checks, and operating days, exactly as Chapter 1 introduced the arithmetic and Chapters 22 and 24 refined it. Every cell in it resolves to a whole number of covers at either \$46 or \$24.

Dinner covers total 475 across five services — an average of 95 a night, which is the 68 seats × 1.4 turns figure the plan has carried since Chapter 1. Annualized: 695 × 52 = 36,140 covers and \$27,130 × 52 = **\$1,410,760**.

The second base is the plan revenue line: \$1,550,000, which is what Chapter 4's pro forma carries and what every percentage in this book has been computed against.

They are not in conflict. They are two different things, and the difference has a name.

🧮 Run the Numbers

The revenue bridge, and the phantom check it creates.

The base grid is what the dining room produces on an ordinary week with nobody buying it out and nothing on the patio. The plan's top line is larger because the plan expects revenue from places the grid does not contain:

text Base grid (covers x check, 52 weeks) $1,410,760 + Revenue bridge $139,240 ────────────────────────────────────────────────────────────────── = Plan revenue, Year 1 $1,550,000

Weekly, the same reconciliation:

text Plan revenue / 52 $29,808 (exactly $29,807.69) Base grid, weekly $27,130 ────────────────────────────────────────────── Bridge, weekly $2,678 x 52 = $139,256 ~ $139,240

(The \$16 spread is rounding \$29,807.69 to \$29,808. Round at the end, never in the middle — §31.5 returns to this.)

The bridge is \$139,240, or 9.0% of plan revenue, and it is the accumulated contribution of the channel chapters: the 16-seat patio in season, private events and buyouts (Chapter 29), off-premise (Chapter 28), shoulder-hour and prix-fixe programs (Chapter 24), and gift-card and retail sales (Chapter 27). Each of those chapters contributed a piece. What has never been written down is the single line that adds them up and confirms the pieces total \$139,240. That attribution is owed, and Chapter 32 needs it, because break-even math treats an event deposit and a Tuesday two-top very differently.

Now the trap. Suppose you take the plan's revenue and divide it by covers to get an average check. Here are two ways to do that, both of which a reasonable person would try:

text All-in: $1,550,000 / 36,140 covers = $42.89 Dinner-only: ($1,550,000 - $274,560 brunch) / 24,700 = $51.64

(Brunch at base: \$5,280/week × 52 = \$274,560. Dinner covers: 475 × 52 = 24,700.)

Neither number describes anything. No guest at Bellwether spends \$42.89 — the dinner check is \$46 and the brunch check is \$24, and \$42.89 is simply the weighted average of two populations that never mix. And \$51.64 is worse: it is the dinner check with the entire revenue bridge shoved onto dinner covers, which silently claims that private events and delivery orders are dinner guests sitting in the 68 seats.

The rule: an average check is only meaningful within a single daypart, on a single revenue base, and you must name the base every time. Bellwether's dinner check is \$46 on the base grid. Bellwether's blended plan revenue per base cover is \$42.89 and is useless. If you ever see a \$51 check average in this plan, somebody divided the wrong pair of numbers.

Which base does this chapter use? The plan's \$1,550,000, for every percentage in every statement, because that is what Chapters 4 through 30 computed against and re-basing now would invalidate thirty chapters of arithmetic. The base grid stays alive as the operating forecast — it is what the weekly flash report in §31.6 measures against, because it is the part the dining room controls.

Bellwether, Year 1, with dollars in it

🧾 Read the Numbers

```text FIGURE 31.2 — "Year One, on plan" [the Bellwether plan] THE ARTIFACT The Year-1 pro forma profit-and-loss statement for Bellwether, assembled from every operating chapter in this book. Twelve months, first Tuesday in April through the following March. THE CONTEXT 68 seats (56 dining room, 12 bar) plus a 16-seat seasonal patio, Rivermill District. Dinner Tuesday-Saturday plus weekend brunch. Full bar, roughly 40-bottle list. Second-generation space, 2,800 sq ft.

               REVENUE
                 Food sales                    $1,116,000    72.0%
                 Beverage sales                  $434,000    28.0%
                 TOTAL REVENUE                 $1,550,000   100.0%

               COST OF SALES
                 Food cost      (30.0% of food sales)          $334,800    21.6%
                 Beverage cost  (22.0% of beverage sales)       $95,480     6.2%
                 TOTAL COST OF SALES                           $430,280    27.8%

               LABOR
                 Wages and salaries                            $415,000    26.8%
                 Employer payroll taxes (9.25% of wages)        $38,388     2.5%
                 Workers' compensation (2.90% of wages)         $12,035     0.8%
                 Benefits, meals, uniforms, training             $34,577     2.2%
                 TOTAL LABOR                                   $500,000    32.3%
               ───────────────────────────────────────────────────────────────────
                 PRIME COST                                    $930,280    60.0%
               ───────────────────────────────────────────────────────────────────

               OTHER OPERATING
                 Technology and payment processing              $73,273     4.7%
                 Utilities                                      $40,000     2.6%
                 Insurance (GL, liquor, property, EPLI)         $29,300     1.9%
                 Supplies, smallwares, linen, chemicals         $27,900     1.8%
                 Marketing                                      $23,250     1.5%
                 Repairs and maintenance                        $15,500     1.0%
                 Waste, grease, pest, hood, music license        $7,777     0.5%
                 TOTAL OTHER OPERATING                         $217,000    14.0%

               OCCUPANCY
                 Base rent ($28.00/sq ft x 2,800)               $78,400
                 NNN ($6.00/sq ft x 2,800)                      $16,800
                 TOTAL OCCUPANCY                                $95,200     6.1%

               GENERAL AND ADMINISTRATIVE                       $46,500     3.0%
               ───────────────────────────────────────────────────────────────────
               OPERATING PROFIT (EBITDA)                       $261,020    16.8%
               ───────────────────────────────────────────────────────────────────
                 Depreciation and amortization (illustrative)   $58,500     3.8%
                 Interest portion of debt service               $39,700     2.6%
               PRE-TAX INCOME                                  $162,820    10.5%

               MEMO — CASH VIEW
                 Operating profit (EBITDA)                     $261,020
                 less total debt service (principal + interest) $69,500
                 = cash before taxes, capex, distributions     $191,520    12.4%

WHAT IT SHOWS A plan that works, on its own terms, with room in it. Prime cost lands exactly on the 60.0% full-service benchmark. Occupancy at 6.1% is unusually low and is the reason the rest of the statement breathes. Operating profit of 16.8% is well above the 3-10% range Chapter 1 gave for full service, which should make you suspicious rather than pleased -- a plan that beats the industry by ten points is usually a plan with an optimistic line in it. Section 31.5 finds which one. WHAT IT DOESN'T It does not show cash by week, so it cannot tell you which Thursday in February the account gets thin (Chapter 33). It does not show the ramp: this is an annual average, and Chapter 9 established that the first quarter runs nowhere near these numbers. It does not show comps, voids, or discounts, which are netted invisibly inside "revenue" (§31.8). And the depreciation figure is illustrative -- your accountant sets the schedule, and whether the equipment lease is capitalized changes it. THE DECISION Adopt this as the plan of record, publish the chart of accounts in §31.2 that produces it, and build the weekly flash report in §31.6 that will tell you within eight days whether you are on it. Then go find the optimistic line. THE LESSON An annual P&L is a scoreboard, not an instrument. It tells you what happened after you can no longer do anything about it. Everything useful in this chapter happens weekly. ```

Every line in that statement foots. Cost of sales plus labor plus other operating plus occupancy plus G&A is \$1,288,980, and \$1,550,000 minus \$1,288,980 is \$261,020. Check it yourself; you should check every statement anyone hands you, including this one, and you should do it before you read the commentary, because commentary is where a bad number gets explained rather than found.

FIGURE 31.3 — Where a Bellwether dollar goes, Year 1 on plan        [the Bellwether plan]

  cost of sales           ██████████████             27.8¢   ┐
  labor, all-in           ████████████████           32.3¢   ┘  PRIME COST = 60.0¢
  other operating         ███████                    14.0¢
  occupancy               ███                         6.1¢
  general & admin         ██                          3.0¢
  ────────────────────────────────────────────────────────────
  = operating profit      ████████                   16.8¢    (EBITDA)
      of which: depreciation & amortization           3.8¢    non-cash
                interest                              2.6¢
                = pre-tax income                     10.5¢

  Compare Figure 1.2, the industry-typical full-service dollar: prime 64¢, occupancy 8¢,
  operating profit 7¢. Bellwether's plan claims four points better on prime and two points
  better on rent. The rent is real and negotiated (Chapter 6). The prime is the argument.

The year-three correction

One more thing has to be fixed before this statement can carry three years, and it is a straightforward error rather than a disagreement.

Chapter 4's pro forma held occupancy at \$95,200 in all three years. Chapter 6 then negotiated the lease and wrote a base-rent escalation of \$1.00 per square foot effective in year three. On 2,800 square feet that is **\$2,800**. Base rent goes from \$28.00 to \$29.00 per square foot; occupancy in Year 3 is therefore:

   base rent   2,800 sq ft x $29.00   =   $81,200
   NNN         2,800 sq ft x  $6.00   =   $16,800
   ────────────────────────────────────────────────
   Year 3 occupancy                       $98,000        (not $95,200)

Year-3 operating profit falls by exactly \$2,800, from the \$362,490 Chapter 4 published to **\$359,690**, which is **19.4%** of \$1,850,000 rather than 19.6%. The corrected three-year statement is in the Business Plan checkpoint at the end of this chapter, and that is now the version of record.

Two points of practice, because the error is more instructive than the correction.

Nobody was careless. Chapter 4 was written before the lease existed. Chapter 6 negotiated the lease and recorded the term correctly in the lease abstract. The failure is structural: there was no step in the process where a lease term walked back into the pro forma. In a real business, that step is a standing item — every time a contract is signed, someone updates the forecast. If nobody owns that job, you will discover your escalation the month it hits.

And NNN is not fixed either. Triple-net charges are an estimate that reconciles annually against the landlord's actual costs (Chapter 6). The \$16,800 is a budget, not a bill. Build the habit of treating it as a line that can move, and read the reconciliation statement when it arrives instead of paying it.

31.2 The chart of accounts: why a foodservice-specific one exists

A chart of accounts is the numbered list of buckets every dollar goes into. It sounds like bookkeeping trivia. It is actually the single decision that determines whether your P&L can answer questions, and most independent restaurants get it wrong by accepting whatever their accounting software installed by default — a generic small-business chart with "Cost of Goods Sold" as one line and "Office Expense" as another.

Here is why that fails. On a generic chart, you cannot answer what is my pour cost? — because beer, wine, and spirits are in one bucket with the food. You cannot answer is my labor problem in the kitchen or the dining room? — because there is one "Wages" account. You cannot answer how much am I paying to accept a credit card? — because processing fees landed in "Bank Charges" along with the monthly account fee. And you cannot benchmark against anything, because nobody else's buckets are shaped like yours.

A foodservice chart of accounts is built so that the P&L in §31.1 falls out of it automatically.

FIGURE 31.4 — Bellwether's chart of accounts (abbreviated)          [the Bellwether plan]

  4000  REVENUE
        4010  Food sales — dinner              4030  Private events and buyouts
        4011  Food sales — brunch              4035  Off-premise / delivery
        4020  Beverage — beer                  4040  Retail and merchandise
        4021  Beverage — wine                  4050  Gift card redemptions
        4022  Beverage — spirits/cocktails     4090  Comps          (contra-revenue)
        4023  Beverage — non-alcoholic         4091  Discounts      (contra-revenue)

  5000  COST OF SALES
        5010  Food purchases                   5040  Beverage — spirits
        5015  Food inventory adjustment        5045  Beverage inventory adjustment
        5020  Beverage — beer                  5080  Transfers: kitchen → bar
        5030  Beverage — wine                  5081  Transfers: bar → kitchen
                                               5085  Comped product at cost (contra)

  6000  LABOR
        6010  Salaries — management            6050  Employer payroll taxes
        6020  Wages — BOH hourly               6060  Workers' compensation
        6030  Wages — FOH hourly               6070  Benefits
        6040  Overtime premium                 6080  Employee meals (at cost)
                                               6090  Uniforms and training

  7000  OTHER OPERATING (direct operating expense)
        7010  Supplies and paper               7070  Technology — POS, KDS, reservations
        7020  Smallwares, china, glass, silver 7075  Technology — card processing
        7030  Linen and laundry                7080  Utilities
        7040  Cleaning and chemicals           7085  Waste, grease, pest, hood cleaning
        7050  Marketing and promotion          7090  Repairs and maintenance
        7060  Music licensing                  7095  Insurance — GL, liquor, EPLI

  8000  OCCUPANCY
        8010  Base rent                        8030  Property insurance
        8020  NNN / CAM                        8040  Property tax
        8025  Rent escalation                  8050  Percentage rent (if triggered)

  8500  GENERAL AND ADMINISTRATIVE
        8510  Accounting and bookkeeping       8540  Licenses and permits
        8520  Legal                            8550  Office and postage
        8530  Bank fees                        8560  Dues and subscriptions

  9000  BELOW THE OPERATING LINE
        9010  Depreciation                     9030  Interest expense
        9020  Amortization                     9040  Income tax

  BALANCE SHEET (see §31.9)
  1000  ASSETS      1010 operating cash · 1050 inventory — food · 1055 inventory — beverage
                    1200 leasehold improvements · 1250 equipment · 1290 accumulated depreciation
  2000  LIABILITIES 2100 accounts payable · 2200 SALES TAX PAYABLE · 2210 payroll liabilities
                    2220 gift card liability · 2230 tips payable · 2400 notes and lease obligations
  3000  EQUITY      3010 owner contributions · 3020 distributions · 3900 retained earnings

Three design choices in that chart are worth defending, because each one is a place operators disagree and each one changes a percentage you will later compare against somebody else's.

Beverage is split three ways. Beer, wine, and spirits have different cost structures — a keg yields differently from a bottle, and a cocktail's cost includes ice, garnish, and dilution (Chapter 15). One "beverage" account gives you a blended pour cost that hides which category is leaking. Bellwether's plan carries a 22% blended pour cost; the only way to defend that number is to be able to break it.

Card processing gets its own account, separate from other technology. On Bellwether's plan the whole technology line is \$73,273 — 4.7% of sales, the largest non-labor operating expense in the building. Card processing is typically the biggest single piece of it, generally running in the range of 2.5% to 3.0% of card sales for an independent. At Bellwether's volume, with roughly nine of every ten dollars arriving on a card, that is somewhere between \$34,900 and \$41,900 a year (Chapter 26 works the fee structure). You cannot negotiate a number you cannot see, and you cannot see it if it is inside "bank fees."

Comps and discounts are contra-revenue accounts, not expenses. This is the choice that trips up the most people, and §31.8 is devoted to why.

⚠️ Where the Money Leaks

The benchmark that isn't.

An operator tells you their occupancy runs 6.1% and yours runs 8.4%, and you conclude they negotiated a better lease. Maybe. Or maybe they put property insurance in general and administrative and you put it in occupancy.

Look at Bellwether's own chart. Insurance appears twice: general liability, liquor liability, and employment-practices coverage sit at 7095 inside other operating (\$29,300), while property insurance sits at 8030 inside occupancy. That is a defensible split — property insurance is a cost of the building, the others are costs of operating — but it is a choice. Move the \$29,300 into occupancy and Bellwether's occupancy percentage jumps from 6.1% to 8.0% while other operating falls from 14.0% to 12.1%. Same restaurant. Same money. Two points of difference in the number Chapter 1 told you to read prime cost against.

(Check: \$95,200 + \$29,300 = \$124,500 ÷ \$1,550,000 = 8.0%. \$217,000 − \$29,300 = \$187,700 ÷ \$1,550,000 = 12.1%.)

What the disciplined operator does: write the account definitions down, once, in a one-page document that says what goes in each account and what does not. Give it to your bookkeeper. Never compare your percentages to an industry figure without knowing how that figure was built. And when you change a classification — which you occasionally should — restate the prior periods so your trend line still means something. A trend across a classification change is not a trend.

Mapping the plan into the accounts

The two large aggregate figures the plan carries decompose exactly into the chart above. Both must foot, and both do.

Note what the second table tells you that "labor is \$500,000" does not: **only \$415,000 of it is wages.** The other \$85,000 — 20.5 cents on every wage dollar — is payroll tax, insurance, and benefits that arrive automatically the moment you put someone on the clock. When a manager says "I added four hours," the cost is not four hours times the rate. It is four hours times the rate times 1.205. Chapter 19 built the schedule; this is what the schedule costs.

31.3 COGS: computing it properly, and the inventory adjustment everyone skips

Cost of goods sold (COGS) is the cost of the food and beverage product you actually consumed during a period. Not what you bought. Not what you paid for. What you used.

That distinction is the entire chapter for most operators, and Chapter 1 already told you why: the restaurant in Figure 1.3 believed its food cost was 30% for eleven months because it was dividing invoices by sales, and it was actually 34.5%.

The usage formula belongs to Chapters 11 and 13 and we are not redefining it here. We are using it, and then adding the four adjustments that turn usage into a COGS number you can put on a statement.

   USAGE     =  beginning inventory  +  purchases  −  ending inventory

   COST OF SALES  =  usage
                     −  transfers out      (product sent to another department)
                     +  transfers in       (product received from another department)
                     −  employee meals at cost
                     −  comped and promotional product at cost

Every one of those subtractions moves a real cost somewhere else — it does not make it disappear. Employee meals move to labor (account 6080). Comped product moves to a contra account so that §31.8 can show you what generosity cost. Transfers move between the kitchen and the bar so that food cost and pour cost each own what they actually consumed. If a subtraction from COGS does not have a matching addition somewhere, you are not adjusting. You are hiding.

👨‍🍳 On the Line

Sunday night, 10:40, counting the walk-in.

Here is the actual job, because it is the part nobody in a business-plan class ever does.

The count happens at the same moment every week — for Bellwether, after close on the last night of the operating week — and it happens after the last delivery of the week and before the first of the next. It is done by two people with a count sheet organized in the physical order of the shelves, not alphabetically, because a sheet that follows the shelves gets counted correctly and a sheet that follows the alphabet gets guessed. Walk-in, then reach-ins, then freezer, then dry storage, then the line, then the bar.

Partial cases get counted as partials. Open containers get weighed or eyeballed to a tenth and the convention gets written down — "we call a half-full 5 lb bag of flour 2.5 lb" — because a convention applied consistently is worth more than a precision applied occasionally.

It takes two trained people about seventy minutes at Bellwether's size, and the first several counts take twice that. You do it yourself for the first six months. Not because your team is dishonest, but because you will learn more standing in your own walk-in on a Sunday night than from any report anyone ever sends you: that the pork is over-ordered, that there are four half-used containers of the same herb, that somebody is buying the expensive lemons.

The failure modes, in order of frequency: counting on a different day each week (which makes every comparison meaningless); counting before the Sunday delivery one week and after it the next; forgetting the bar's back stock; and — the worst one — writing "same as last week" for a category nobody wants to count. The last one is not laziness. It is the single most common way an inventory number becomes fiction, and it always starts with the dry goods.

What it looks like on a real week

Here is Bellwether in an ordinary week, food only, and then the same restaurant the following week.

🧮 Run the Numbers

Two weeks, identical food cost, and a seven-point swing in what the shortcut reports.

	Week A	Week B
Beginning food inventory	\$12,500 | \$13,100
+ Food purchases	\$7,400 | \$5,900
− Ending food inventory	\$13,100 | \$12,200
= Gross usage	\$6,800** | **\$6,800
Food sales	\$21,900 | \$21,900
Actual food cost (usage ÷ sales)	31.1%	31.1%
Shortcut (purchases ÷ sales)	33.8%	26.9%

The restaurant used exactly \$6,800 of food in both weeks and sold exactly \$21,900 in both weeks. Nothing changed. The shortcut reports 33.8% one week and 26.9% the next — a 6.9-point swing — entirely because the walk-in grew \$600 in Week A and shrank \$900 in Week B.

An operator watching the shortcut has a crisis in Week A and a celebration in Week B, and both are hallucinations. Worse, the operator learns the wrong lesson: buy light at the end of the week and the number improves. That is not cost control. That is running the walk-in down and calling it management, and it produces exactly the 86s and the emergency Sunday grocery-store run that Chapter 14 warned you about.

Over time the shortcut converges — across both weeks combined, purchases of \$13,300 on sales of \$43,800 is 30.4%, against actual usage of \$13,600 giving 31.1% — which is precisely why it survives. It is roughly right annually and useless weekly, and weekly is when you can still do something.

From usage to cost of sales

Gross usage is not yet the number that goes on the P&L. Take Week A and finish it.

And the bar, same week:

Total cost of sales for the week: \$6,370 + \$1,995 = \$8,365** on total sales of \$30,400 = 27.5%**.

Now look at what just happened to the food number. The same week reads:

• 33.8% — purchases ÷ sales, the shortcut
• 31.1% — usage ÷ sales, the inventory-adjusted number
• 29.1% — cost of sales ÷ sales, after transfers, employee meals, and comps at cost

Three answers to "what was my food cost this week?" and only the third one is comparable to a recipe cost card, because a cost card prices the food a guest was served. If you are going to compare your 29.1% against the Hearth Chicken's 29.4%, you must be measuring the same thing.

The other two are not wrong; they answer different questions. 31.1% answers what did the kitchen consume? — the right number for a waste and portioning investigation. 33.8% answers what did I spend? — the right number for a cash conversation. Name which one you are quoting. Most arguments about food cost are two people quoting different formulas at each other.

⚠️ Where the Money Leaks

The transfer nobody logs, and the bar that eats the kitchen's number.

Bellwether's bar takes \$185 of citrus, herbs, and garnish from the kitchen in an ordinary week. That is \$9,620 a year. If nobody logs the transfer, the kitchen carries the cost and the bar gets the credit: food cost is overstated by roughly 0.9 points and pour cost is understated by roughly 2.2 points.

(\$9,620 ÷ \$1,116,000 food sales = 0.86 points. \$9,620 ÷ \$434,000 beverage sales = 2.22 points.)

Two points of pour cost is the difference between a bar program that looks excellent and one that looks ordinary — and it is entirely an accounting artifact. The chef spends a month chasing a food cost problem that is sitting in a cocktail.

What the disciplined operator does: a transfer sheet on a clipboard by the bar door, filled in at the moment of transfer, entered weekly. It takes ninety seconds a day. It is the cheapest accuracy you will ever buy.

31.4 Controllable vs. non-controllable: what you can change this week

A controllable cost is one a manager's decisions move inside the current period. A non-controllable cost is one that was set by a contract, a policy, or a signature, and will arrive at the same size regardless of how well anyone runs the building this week.

The distinction is not cosmetic. It determines what you hold a manager accountable for, what you review weekly versus annually, and — most importantly — where you look first when a number is wrong.

Restructure Bellwether's Year 1 along that cut:

(Other controllable: \$27,900 supplies + \$23,250 marketing + \$15,500 R&M + \$40,000 utilities + \$7,777 waste/grease/pest/hood/music = \$114,427. Non-controllable operating: \$29,300 insurance + \$73,273 technology = \$102,573. Together \$217,000.)

Controllable income is \$505,293, or 32.6% of sales. That is the number a general manager should be measured on and bonused against, because everything in it is a decision somebody made this month. Below that line sits \$244,273 of cost that arrives whether the restaurant is brilliant or terrible.

Two cautions, both of which matter more than they sound.

"Controllable" does not mean "variable." These are different cuts of the same costs and people confuse them constantly. Card processing is almost perfectly variable — it rises and falls with sales, dollar for dollar — and almost entirely non-controllable, because the rate is contractual and a manager cannot change it on a Tuesday. Rent is the reverse: fixed and non-controllable. Salaried management wages are fixed and controllable — you decide how many managers you carry. Chapter 32 needs the fixed/variable cut for break-even; this chapter needs the controllable cut for accountability. Do not let one substitute for the other.

Utilities are semi-controllable and you should treat them that way. There is a base load — the walk-in runs on a Monday when you are closed — and a volume component. Putting the whole \$40,000 in "controllable" is a simplification that says: somebody is responsible for turning the hood off.

EBITDA, and what it does and does not tell you

EBITDA — earnings before interest, taxes, depreciation, and amortization — is the operating profit line before the four items that depend on how the business was financed and built rather than how it is run. On Bellwether's Year 1 it is \$261,020.

It is genuinely useful for exactly one thing: comparing the operating performance of two restaurants without the comparison being swamped by the fact that one owner paid cash and the other borrowed \$335,000. It is also what a buyer will eventually value the business on, and what a lender computes coverage from.

And it is dangerous for exactly one reason: EBITDA is not cash. It excludes depreciation, which is correct — depreciation is not a cash cost. But it also excludes the principal portion of debt service, which very much is. Bellwether pays \$69,500 a year in debt service and only \$39,700 of that is interest; the other \$29,800 leaves the bank account and appears nowhere on the income statement at all.

🧮 Run the Numbers

Profit is \$162,820. Cash is \$191,520. Both are correct.

text ACCOUNTING VIEW CASH VIEW Operating profit (EBITDA) $261,020 Operating profit (EBITDA) $261,020 less depreciation & amort. $58,500 less debt service — interest $39,700 less interest $39,700 less debt service — principal $29,800 ───────────────────────────────────── ───────────────────────────────────── = pre-tax income $162,820 = cash before tax and capex $191,520

The gap is **\$28,700**, and it is exactly the depreciation you did not pay (\$58,500) minus the principal you did (\$29,800). Two entirely defensible numbers, \$28,700 apart, describing the same twelve months.

Both matter and they matter to different people. Your tax return cares about \$162,820. Your bank account cares about \$191,520. You have to be able to produce both, and you have to know which one somebody is asking for.

Debt service coverage — net operating income ÷ total debt service, the ratio Chapter 5 defined — is \$261,020 ÷ \$69,500 = 3.76× on plan. Hold that number; §31.5 does something uncomfortable to it.

Accrual vs. cash accounting

Cash accounting records revenue when money arrives and expense when money leaves. Accrual accounting records revenue when it is earned and expense when it is incurred, regardless of when the money moves.

Restaurants have an unusual relationship with this distinction because revenue is nearly always cash and accrual simultaneously — a guest pays as they leave, and cards settle in a day or two. It is the expense side that diverges, and it diverges in ways that will mislead you if you run on cash.

• Payables. Your produce invoice is 21 days. On cash accounting, a month in which you stretch vendors looks profitable and the following month looks catastrophic — and you have learned nothing about either month. On accrual, the food you cooked in March is a March expense.
• Prepaid and annual items. Bellwether's insurance is \$29,300. Paid annually, cash accounting puts the entire premium in one month — a \$29,300 hole in a single period against an average monthly revenue of \$129,167 — and shows eleven months that look 1.9 points better than they are. Accrual spreads it: \$2,442 a month.
• Inventory. This is the big one, and §31.3 already made the argument. Cash accounting has no concept of an inventory adjustment. It reports purchases. You have seen what purchases report.

Run accrual. Have your bookkeeper accrue at minimum: inventory, payroll through the period end, and any prepaid expense over about a thousand dollars. Then have them produce a cash statement alongside it, because Chapter 33 is going to make the case that in the first eighteen months, the cash statement is the one that keeps you open.

🔍 Check Your Understanding

1. A manager reports that "labor was \$9,400 this week." What three follow-up questions do you ask before that number means anything?
2. Bellwether's EBITDA is \$261,020 and its pre-tax income is \$162,820. Which number should the general manager's bonus be computed on, and why is the answer actually neither?
3. Card processing is variable but non-controllable; salaried management wages are fixed but controllable. Give one example of a cost that is both variable and controllable, and one that is both fixed and non-controllable.

(1: Does it include payroll taxes, workers' comp, and benefits, or only wages? Does it include the salaried managers? Does it include accrued hours worked after the period cutoff? A wages-only figure understates Bellwether's true labor by about 20.5%. 2: Neither — controllable income (\$505,293) is the right basis, because it stops before rent, insurance, the loan, and the depreciation schedule, none of which the GM negotiated. 3: Variable and controllable — food cost, or hourly labor hours. Fixed and non-controllable — base rent, or the annual license fee.)

31.5 Prime cost revisited: the weekly calculation and the benchmark by service style

Chapter 1 promised you this section. Here it is.

Prime cost is cost of sales plus total labor, as a percentage of total sales. You have known that since page one. What Chapter 1 could not do — because you did not yet have a chart of accounts, a usage formula, or a payroll structure — was show you how to compute it in eight days instead of seven weeks.

The seven inputs

You need exactly seven numbers, and every one of them exists inside your building on Monday morning.

FIGURE 31.5 — The weekly prime cost calculation                    [the Bellwether plan]

   INPUT                                              WHERE IT COMES FROM        WEEK 22
   ────────────────────────────────────────────────────────────────────────────────────
   1. Net sales, food                                 POS, week close            $21,900
   2. Net sales, beverage                             POS, week close             $8,500
   3. Food cost of sales                              count sheet + invoices      $6,370
   4. Beverage cost of sales                          count sheet + invoices      $1,995
   5. Hourly wages, incl. overtime                    payroll export              $6,483
   6. Salaried wages, 1/52 of annual                  fixed                       $2,365
   7. Payroll taxes + workers' comp + benefits        computed from 5 + 6         $1,740
   ────────────────────────────────────────────────────────────────────────────────────
   TOTAL SALES              (1 + 2)                                             $30,400
   TOTAL COST OF SALES      (3 + 4)                                              $8,365    27.5%
   TOTAL LABOR              (5 + 6 + 7)                                         $10,588    34.8%
   ════════════════════════════════════════════════════════════════════════════════════
   PRIME COST                                                                   $18,953    62.3%
   ════════════════════════════════════════════════════════════════════════════════════

   Input 7 detail:  payroll taxes  9.25% x $8,848  =   $818
                    workers' comp  2.90% x $8,848  =   $257
                    benefits, meals, uniforms      =   $665   ($34,580 / 52)
                                                       ─────
                                                       $1,740

Inputs 3 and 4 require the count from §31.3 — the seventy minutes on Sunday night. Inputs 5, 6, and 7 come out of your payroll system in about four minutes. Inputs 1 and 2 come off the POS. Total time for a trained manager: under two hours, most of it counting. That is the whole discipline. There is no sophisticated part.

Two mechanical notes that will save you arguments.

Use the same seven-day window for every input. If your operating week runs Tuesday to Monday, your inventory count, your sales, and your payroll must all cover Tuesday to Monday. Payroll systems love to run Sunday to Saturday. Make them match or your labor percentage will oscillate for reasons that have nothing to do with the schedule.

Round at the end, never in the middle. Prime cost is \$18,953 ÷ \$30,400 = 62.345%, which prints as 62.3%. If you had instead rounded cost of sales to 27.5% and labor to 34.8% and added the percentages, you would get 62.3% here — but the same shortcut on the annual plan gives 27.8 + 32.3 = 60.1% against a true 60.02%, and computing debt service coverage from a rounded 10.6% instead of the dollars turns 2.35× into 2.36×. Carry the dollars. Round once, at the point of printing.

The benchmark, and the caveat Chapter 1 attached to it

Chapter 1 gave you the bands, and they have not changed:

And Chapter 1 attached a caveat that most readers skip: read prime cost alongside occupancy. A restaurant at 63% prime with 5% rent is in a very different position from one at 63% prime with 11% rent. We are about to need that caveat badly.

Reconciliation one: the roster does not reconcile, and nobody wrote down the mapping

Before we can defend the labor line, we have to know how many people are in the building, and the plan currently says three different things.

• Chapter 17 (the staffing plan) says 31 people — 4 salaried and 27 hourly — and about 22 FTE.
• Chapter 19 (the labor model) says 24 scheduled positions — 3 salaried and 21 hourly — totaling 453.5 hourly hours a week at a blended \$14.70.
• Chapter 18 (training) works from a 29-head hourly roster, which is illustrative for building a training calendar and is not a payroll figure.

These are not three answers to one question. They are answers to three different questions, and the reason nobody caught the discrepancy is that each chapter was internally right. Here is the mapping, which should have been published thirteen chapters ago.

Three things follow, and they are the teaching.

Heads and scheduled positions are different quantities. Twenty-seven hourly people fill twenty-one scheduled slots because a "line cook" slot on a seven-day schedule is covered by two or three different people across the week, and because part-timers, on-call staff, and the person you hired last Tuesday are all heads who fill fractions of slots. Neither number is wrong. Thirty-one is the number you must hire and keep certified; twenty-four is the number the schedule fills; you cannot substitute one for the other.

The two "FTE" figures are not comparable, and the difference is a definition rather than seven missing people. Chapter 17's ~22 counts full-timers as 1.0 and part-timers as 0.5 — a common staffing-plan shorthand. Chapter 19's 14.3 is paid hours ÷ 40, which is the only FTE that multiplies by a wage rate to produce dollars. Publish the hours-based figure in anything financial. Keep the headcount-weighted one in the hiring plan where it belongs, and label it so nobody adds them.

The fourth salaried head is real and is not a scheduled position. Chapter 17's org chart carries four salaried roles; Chapter 19 schedules three, because a schedule is a list of slots that must be covered and the chef-owner covers no fixed slot — they float, and they are in the building far more than forty hours. The dollars are in the labor line either way. A schedule is not a payroll register, and an operator who builds one from the other will be wrong in one direction or the other every single week.

Reconciliation two: the labor line does not fit, and the gap widens twice

Now the uncomfortable part, and the reason this chapter exists.

The plan carries labor at \$500,000 — 32.3%. Chapter 19 built the schedule position by position, hour by hour, and priced it. Here is what a bottom-up build actually produces.

Read the gap column carefully, because it says something precise. The plan and the bottom-up build use identical loading rates — 9.25% payroll tax, 2.90% workers' comp — and carry essentially identical benefits (\$34,577 versus \$34,580). The entire disagreement is \$62,825 of wages; the remaining \$7,636 is tax and insurance riding along behind it.

So this is not an argument about benefits policy or about whether the plan forgot workers' comp. It is an argument about how many hours the restaurant needs and what it must pay for them — which is the hardest question in the business and the one Chapter 19 was built to answer.

Then Chapter 20 made it worse.

⚖️ Code and Compliance

The sous chef's classification, and \$27,000.

Chapter 20 walked the exempt/non-exempt analysis under the Fair Labor Standards Act (FLSA) and found that Bellwether's sous chef position fails the duties test. Paying a salary does not make a position exempt; the exemption depends on what the person actually does, and a sous chef who spends most of a shift on a station cooking is performing non-exempt work regardless of the title on the schedule or the fact that the paycheck is the same every week.

The correction is not optional and it is not expensive to make in advance. Classified correctly, the position is owed overtime. At Bellwether's numbers that is roughly 642 overtime hours a year — just under twelve and a half a week, which is what a sous chef in a hearth kitchen actually works — at a time-and-a-half rate of \$37.50 derived from a \$52,000 salary:

text 642 OT hours x $37.50 = $24,075 employer payroll taxes 9.25% x $24,075 = $2,227 workers' compensation 2.90% x $24,075 = $698 ────────────────────────────────────────────────────── annual cost of correct classification $27,000

\$27,000 a year — 1.7% of revenue — to be right. Getting it wrong costs the back pay anyway, plus liquidated damages in many circumstances, plus the fact that misclassification claims tend to arrive as a group rather than one at a time.

Jurisdiction matters enormously here. Several states set salary thresholds and duties tests more demanding than the federal ones, and the federal salary threshold itself has moved more than once in recent years. Verify locally, with an employment attorney, before you write a single salaried offer letter. Chapter 20 owns the analysis; this chapter owns the arithmetic.

Put the three columns side by side. This is the most important table in the chapter.

🧮 Run the Numbers

The plan's most contested number, in three columns.

	A: Plan	B: Bottom-up	C: Lawful
	Ch. 4 target	Ch. 19 schedule	Ch. 19 + Ch. 20
Revenue	\$1,550,000 | \$1,550,000	\$1,550,000
Cost of sales	\$430,280 | \$430,280	\$430,280
Labor	\$500,000 | \$570,461	\$597,461
Labor %	32.3%	36.8%	38.5%
Prime cost	\$930,280** | **\$1,000,741	\$1,027,741
Prime %	60.0%	64.6%	66.3%
Occupancy	\$95,200 | \$95,200	\$95,200
Other operating	\$217,000 | \$217,000	\$217,000
General and administrative	\$46,500 | \$46,500	\$46,500
Operating profit	\$261,020** | **\$190,559	\$163,559
Operating profit %	16.8%	12.3%	10.6%
Debt service	\$69,500 | \$69,500	\$69,500
Cash after debt service	\$191,520 | \$121,059	\$94,059
DSCR	3.76×	2.74×	2.35×

(Column B prime is 64.56%, which Chapter 19 rounds to 64.5%; carried here as 64.6% because this chapter rounds once, at printing. Column C's DSCR is \$163,559 ÷ \$69,500 = 2.35×; computing it from the rounded 10.6% instead of the dollars gives 2.36×. Round at the end.)

We are not going to resolve this. All three columns are published, and the plan of record carries column A because that is what thirty chapters computed against — but column A is now explicitly labeled as the plan's most contested number, and Chapter 40 will have to defend it or revise it.

Here is what makes the honest presentation bearable: column C still works. At 66.3% prime — which Chapter 1's bands call "distressed" — Bellwether still clears 10.6% operating profit and 2.35× debt service coverage. That is not a fudge. It is the caveat Chapter 1 attached to those bands, arriving exactly on schedule.

Why does 66.3% prime survive here when Chapter 1 said 65–70% means profit is mostly gone? Because occupancy is 6.1%. Chapter 1's bands assume a typical full-service occupancy in the 6–10% range, and Bellwether negotiated the low end of it (Chapter 6) in a second-generation space in a neighborhood that had not finished gentrifying.

Watch what happens if it hadn't.

🧮 Run the Numbers

Same prime cost, four more points of rent.

Take column C exactly as it stands — 66.3% prime, the lawful labor line — and change one thing: occupancy at 10.0% of sales (\$155,000) instead of 6.1% (\$95,200). Everything else identical.

text occupancy 6.1% occupancy 10.0% Operating profit $163,559 $103,759 Operating profit % 10.6% 6.7% Debt service $69,500 $69,500 Cash after debt service $94,059 $34,259 DSCR 2.35x 1.49x

Four points of rent — \$59,800 — takes the business from comfortable to tight. Not fatal, but there is now almost no room between the plan and a bad quarter.

The lesson: the prime-cost band is not a verdict. It is a verdict conditional on your occupancy. Read them as a pair. Bellwether's plan is 60.0 + 6.1 = 66.1; the lawful case is 66.3 + 6.1 = 72.4. Seventy-two would worry me — and it would worry me less than a 72.4 built as 62 prime plus 10.4 occupancy, because prime is the half you can still move. Rent you signed.

Reconciliation five: two fixed/variable splits, and only one of them can go to Chapter 32

One more labor disagreement, and it is subtle enough that neither author could have caught it.

Chapter 4, building the sensitivity analysis, modeled labor as \$252,000 fixed plus 16.0% of sales variable. At \$1,550,000 that is \$252,000 + \$248,000 = **\$500,000** — the plan's labor line, exactly.

Chapter 19, building the schedule position by position, put the fixed floor — the salaried managers plus the minimum crew that must be in the building whether you do 40 covers or 140 — at \$191,895, with the genuinely volume-sensitive part running 24.4% of sales. At \$1,550,000 that is \$191,895 + \$378,566 = \$570,461 — Chapter 19's bottom-up labor line, exactly.

Both splits foot. Each one reproduces its own author's total to the dollar, which is why nobody saw a problem: a fixed/variable split always foots at the volume you calibrated it to. The disagreement is invisible on a P&L and enormous in a break-even, because break-even is determined entirely by the shape — fixed dollars divided by contribution margin ratio — and not at all by the total.

FIGURE 31.6 — Two fixed/variable splits of the same cost line       [the Bellwether plan]

                             FIXED       VARIABLE RATE    AT $1,550,000
  Chapter 4 estimate        $252,000        16.0%           $500,000
  Chapter 19 build          $191,895        24.4%           $570,461
  ──────────────────────────────────────────────────────────────────────
  Difference                -$60,105       +8.4 pts         +$70,461

     variable dollars:  $378,566 - $248,000  =  +$130,566
     fixed dollars:     $191,895 - $252,000  =   -$60,105
                                                ─────────
     net                                        +$70,461   <-- the labor gap, again

  The fixed/variable disagreement and the labor gap are not two problems. They are one
  problem with two faces. At plan volume they are the same $70,461.

Chapter 32 inherits Chapter 19's split: \$191,895 fixed, 24.4% variable. Not because Chapter 19 outranks Chapter 4, but because Chapter 19 built it from the schedule — from actual positions with actual start times — and Chapter 4 estimated it from a target. When a bottom-up build and a top-down estimate disagree about shape, take the build.

For the rest of the cost structure, the honest position is:

That last row is not a dodge. Other operating contains card processing (nearly pure variable), utilities (a base load plus a volume component), insurance (fixed for the policy year), and marketing (entirely discretionary, which is neither). Chapter 32 needs a defensible split of \$217,000 and it should build one line by line rather than inherit a percentage from here.

31.6 The weekly flash report: designing the one page you read every Monday

Everything above this section is machinery. This section is the product.

A weekly flash report is a single page, produced within one business day of the operating week's close, that shows an operator what last week cost and whether it was what they expected. It is called a flash because it is fast and approximate: it does not wait for the bookkeeper, it does not accrue perfectly, and it is not the statement your accountant will produce three weeks later. It is 95% accurate on Monday morning, which is worth infinitely more than 100% accurate on the twenty-second.

The design principles

One page. If it is two pages, it will be read once. Bellwether's fits on one.

Every number gets a comparison. A number alone is trivia. \$30,400 in sales means nothing; \$30,400 against a \$29,300 forecast and a \$28,950 prior week means something. Give every figure at least one comparison and preferably two.

The target ramps. This is where most flash reports fail, and Chapter 9 already told us why.

🧮 Run the Numbers

Why a flat 60% target would make the flash report lie for six months.

Chapter 9 established that Bellwether's first quarter of operation runs at roughly 66.6% prime cost — opening waste, over-scheduling because nobody knows the volume yet, training hours, a menu the kitchen has not built muscle memory for, and a dining room that is still learning where things are. That is not failure. That is what opening costs, and a plan that pretends otherwise is lying about the hardest thirteen weeks in the business.

But the year still has to land at 60.0%. So the rest of the year has to be better than the average. How much better?

```text Let R1 = revenue in weeks 1-13 at 66.6% prime R2 = revenue in weeks 14-52 at 58.0% prime

R1 + R2 = $1,550,000 0.666(R1) + 0.580(R2) = $930,280 (the plan's prime cost dollars)

Solving: R1 = $363,720 (23.5% of the year's revenue, in 25% of its weeks) R2 = $1,186,280

Check: 0.666 x $363,720 = $242,238 0.580 x $1,186,280 = $688,042 ────────────────────────────── $930,280 = 60.0% of $1,550,000 ✓ ```

Weeks 14 through 52 must average 58.0% prime cost — two full points better than the plan's headline — for the plan's headline to be true. That is a completely different management instruction from "hit 60%," and an operator reading a flat 60% target would spend the first quarter thinking they were failing and the third quarter thinking they were winning. Both wrong.

One more thing falls out of that arithmetic: \$363,720 over thirteen weeks is **\$27,978 a week, which is within a few hundred dollars of the \$27,130 base grid. That makes sense — the revenue bridge in §31.1 is private events, patio, off-premise, and gift cards, and essentially none of it exists in your first quarter. Your opening quarter runs on the dining room alone.**

FIGURE 31.7 — The prime cost ramp: what the flash report prints as "target"   [the Bellwether plan]

  weeks  1– 4    71.0%  ████████████████████████████████████   opening; everything is wrong
  weeks  5– 8    66.0%  █████████████████████████████████      systems starting to hold
  weeks  9–13    63.5%  ████████████████████████████████       menu settling, schedule tightening
  weeks 14–26    60.0%  ██████████████████████████████         ← the plan's annual number
  weeks 27–39    57.5%  █████████████████████████████          the year has to be bought back here
  weeks 40–52    56.5%  ████████████████████████████           and here

  Q1 (weeks 1-13) sales-weighted: 66.6%   |   weeks 14-52 sales-weighted: 58.0%
  Year, sales-weighted: 60.0% on $1,550,000

  WARNING: these are WEEKLY targets. The annual 60.0% is SALES-weighted, and the simple
  average of the weekly percentages is 60.1%, not 60.0%, because the light weeks are at the
  front. You cannot average percentages across periods of different size. Re-add the dollars.

That warning is not pedantry; it is the single most common arithmetic error in restaurant reporting. A percentage is a ratio of two sums. Averaging ratios is not the same as taking the ratio of the sums, and the gap grows with the variance between periods. Every time you want an average percentage across periods, add the numerators, add the denominators, and divide.

The report itself

🧾 Read the Numbers

```text FIGURE 31.8 — "The Monday page" [the Bellwether plan] THE ARTIFACT Bellwether's one-page weekly flash report, week 22 of operation (Tuesday through Monday, early September). Produced Monday at 9:40 a.m. from the Sunday-night count, the POS export, and the payroll file. THE CONTEXT Late summer. Patio open. A strong week — 779 covers against a base-grid forecast of 695. Management has been actively cutting the floor since week 18; hourly hours came in 12.5 under the standard schedule.

══════════════════════════════════════════════════════════════════════════════════ BELLWETHER — WEEKLY FLASH — WEEK 22 week ending Monday, Sept 8 ══════════════════════════════════════════════════════════════════════════════════ SALES ACTUAL FORECAST VAR PRIOR WK VAR Food $21,900 $21,100 +$800 $20,850 +$1,050 Beverage $8,500 $8,200 +$300 $8,100 +$400 NET SALES $30,400 $29,300 +$1,100 $28,950 +$1,450

COVERS AND CHECK COVERS CHECK vs BASE GRID Dinner (5 services) 532 $46.00 +57 covers (base 475) Brunch (2 services) 247 $24.00 +27 covers (base 220) TOTAL 779 +84 covers (base 695) [ do NOT divide $30,400 by 779 — see §31.1 ]

COST OF SALES ACTUAL % OF CAT TARGET VAR Food cost of sales $6,370 29.1% 30.0% -0.9 pt Beverage cost of sales $1,995 23.5% 22.0% +1.5 pt TOTAL COST OF SALES $8,365 27.5% 27.8% -0.3 pt

LABOR ACTUAL % SALES TARGET VAR Hourly hours 441.0 453.5 -12.5 hrs Overtime hours 6.0 0.0 +6.0 hrs Hourly wages $6,483 Salaried $2,365 Taxes, WC, benefits $1,740 TOTAL LABOR $10,588 34.8% 32.3% +2.5 pt Sales per labor hour $68.94 (net sales / hourly hours) Covers per labor hour 1.77

══════════════════════════════════════════════════════════════════════════════════ PRIME COST $18,953 62.3% RAMP TARGET WK 22: 60.0% VARIANCE +2.3 pt = +$713 this week annualized at this rate: $37,076 ══════════════════════════════════════════════════════════════════════════════════

COMPS, DISCOUNTS, VOIDS Gross sales $31,196 less comps (manager-authorized) $486 1.56% of gross (target ≤1.0%) less promotional discounts $310 0.99% of gross = NET SALES $30,400 Voids (41 items) $128 does NOT reduce sales — audit item

CHANNEL COST ACTUAL % NET SALES Card processing (2.81%, Ch. 26) $854 2.81% Takeout packaging + first-party fees $38 0.13% Third-party delivery commission $0 0.00% (first-party only) Event coordination (0 events wk 22) $0 0.00% TOTAL CHANNEL COST $892 2.93%

CASH Sales tax collected (7%, illustrative) $2,128 LIABILITY, not revenue Expected deposits + card settlement $32,528 Actual deposits + card settlement $32,890 OVER / (SHORT) ($18)

THE ONE ACTION THIS WEEK: Pour cost is 1.5 points over on a strong beverage week. Pull the comp detail by employee and re-count the well spirits Thursday. ══════════════════════════════════════════════════════════════════════════════════

WHAT IT SHOWS A good sales week and a mediocre cost week. Food cost is running 0.9 points BETTER than target, which on a week this busy is real and worth understanding rather than celebrating. Labor is 2.5 points over despite the manager cutting 12.5 hours below the standard schedule -- which tells you the standard schedule, not the manager, is the problem. Six overtime hours in a week with hours cut means someone stayed late on a specific night; the schedule did not cause it, a shift did. Prime lands at 62.3% against a 60.0% ramp target: $713 this week. WHAT IT DOESN'T It does not say WHY pour cost ran 23.5%. Over-pour, a comped round, a mis-keyed transfer, a bad count, or a genuinely heavier cocktail mix are all consistent with this page. It does not show which night lost money -- a weekly total hides a catastrophic Tuesday inside a great Saturday (Chapter 32 splits it by daypart). It does not show cash position, only cash reconciliation; the $2,508 of sales tax sitting in the account is not spendable and this page cannot tell you that. THE DECISION Two actions, both this week. (1) Pull the comp-by-employee report and re-count well spirits Thursday -- 1.5 points of pour cost is $128 a week and $6,600 a year. (2) Rebuild the standard hourly schedule: a manager cutting 12.5 hours and still landing 2.5 points over is not a management problem, it is a staffing-guide problem, and it is the Chapter 19 build showing up in real life exactly as predicted. THE LESSON A flash report's job is not to be accurate. It is to be EARLY, and to end with one action. A page that produces a feeling instead of a task has failed, no matter how many numbers are on it. ```

👨‍🍳 On the Line

Monday, 9:40 a.m.

The report above does not get emailed. It gets read, out loud, in the office, by whoever is accountable, with the chef and the FOH manager in the room, for fifteen minutes. Then everybody goes back to work.

The sequence that works: sales first (are we where we thought?), then covers (is it traffic or check?), then prime cost against the ramp, then the one action. Not a discussion of every line — one action. If two things are wrong, do the bigger one. If nothing is wrong, say "nothing is wrong" and go, because a meeting that always finds a problem trains people to manufacture one.

The failure modes are predictable. The report becomes a performance review — the labor number becomes an accusation and the FOH manager starts protecting the number instead of managing the floor. The report becomes optional in a busy week, which is exactly the week you needed it. Or the report grows: somebody adds a column, then a section, then a second page, and within four months nobody reads it. Fight for the one page.

And when it says something you do not want to hear — as week 22's does, about the schedule — the whole value of the exercise is in believing it the first time rather than the fourth.

A definitional note before anyone reports it as an error

While we are standardizing definitions: Chapter 22 plans the floor on a 96-minute table cycle and Chapter 24 measures a 95-minute dine time. These are not in conflict and neither is wrong. A dine time runs seated-to-paid and belongs to the guest. A table cycle runs seated-to-seated and belongs to the table, so it necessarily includes the bus-and-reset. The cycle is therefore always the larger of the two, and it is here.

The one-minute spread is a planning round-up in the reservation grid, not a claim that Bellwether resets a four-top in sixty seconds — a real reset runs three to five minutes, absorbed inside the buffer the host stand already builds into a quote. Do not net the two numbers against each other and call the difference "reset time." Report the cycle when you are building a grid; report the dine time when you are measuring a service.

31.7 Accounting periods: monthly vs. 4-4-5 and 13-period calendars

Bellwether opens on the first Tuesday in April. That is a fine day to open a restaurant and a terrible day to start a fiscal year on a monthly calendar, and the reason is one that catches almost every independent.

Months are not comparable to each other. February has 28 days. March has 31. And more to the point, months contain different numbers of Fridays and Saturdays, which for a dinner restaurant is the only day-count that matters.

🧮 Run the Numbers

What an extra weekend does to a month you did not run any differently.

Bellwether's base grid, by day:

text an extra Friday (dinner) $5,520 an extra Saturday (dinner $5,658 + brunch $2,640) $8,298 an extra Sunday (brunch) $2,640 ───────────────────────────────────────────────────────────── total $16,458

Average monthly revenue on plan is \$1,550,000 ÷ 12 = **\$129,167. A month with five Fridays, five Saturdays, and five Sundays carries \$16,458 — 12.7% — more revenue** than a month with four of each. Nobody did anything differently. The calendar did it.

Now watch it distort your percentages. Rent is \$7,933 a month either way:

text 4-weekend month 5-weekend month Revenue $129,167 $145,625 Occupancy (fixed) $7,933 $7,933 Occupancy % 6.14% 5.45%

Occupancy "improved" 0.7 points and you did not renegotiate anything. Every fixed cost in the building does the same thing, in the same direction, at the same time — which means the month you compare against is doing as much work as the month you are reporting.

The fix is a fiscal calendar that makes periods comparable.

The 13-period calendar divides the year into thirteen periods of exactly four weeks — 28 days each, 364 days total. Every period has exactly four Tuesdays, four Fridays, four Saturdays. Period 7 this year is directly comparable to Period 7 last year and to Period 6 this year. The four inventory-count weeks inside each period line up perfectly with the weekly flash report, which is the real prize: your weekly and periodic reporting share the same building blocks.

The 4-4-5 calendar keeps twelve months but defines each quarter as thirteen weeks split into periods of four, four, and five weeks. It preserves quarterly comparability and keeps something that looks like a month, which is why retailers and larger restaurant groups use it. It costs you the clean four-week period.

Both leave one day a year unassigned (364 versus 365), which accumulates. Every five or six years you run a 53rd week, and that period will look extraordinary in every comparison. Note it in the report so nobody celebrates.

FIGURE 31.9 — Three calendars, one year                            [structure]

  MONTHLY          Jan     Feb     Mar     Apr     ...      periods of 28-31 days
                   31d     28d     31d     30d              4 or 5 weekends
                   |___ not comparable to each other ___|

  4-4-5            [ 4wk | 4wk | 5wk ]  [ 4wk | 4wk | 5wk ]  ...
                   |___ Q comparable ___|                    "months" still differ

  13-PERIOD        [4wk][4wk][4wk][4wk][4wk][4wk][4wk][4wk][4wk][4wk][4wk][4wk][4wk]
                   |___ every period identical: 28 days, 4 weekends ___|
                   |___ each period = exactly 4 flash reports ___|

  Bellwether opens the first Tuesday in April. On a 13-period calendar, Period 1 runs
  four weeks from that Tuesday, and the operating week (Tue-Mon) nests inside it cleanly.
  On a monthly calendar, "April" is a stub period of about four weeks that will be
  compared against a full May and look terrible.

⚠️ Where the Money Leaks

The stub period, and the panic it causes.

Bellwether opens the first Tuesday in April — call it the 7th. On a monthly calendar, "April" is about 24 operating days, carries the tail of pre-opening payroll, and includes a soft-open week at nearly zero revenue. May is a full month. The May-over-April comparison will show a revenue increase of something like 40% and a prime cost that appears to have collapsed.

Neither is real, and the danger is that both get believed. I have watched an owner conclude from a stub-period comparison that a menu change worked, keep it, and spend a year wondering why the lift never repeated.

What the disciplined operator does: pick the fiscal calendar before you open, tell your bookkeeper and your POS about it on day one, and label the opening period as a stub on every report it appears on. Changing calendars later means restating history, and restating history is the kind of project that never gets finished.

31.8 Comps, voids, and discounts, and how they distort every number above them

Three words that get used interchangeably and mean entirely different things.

A comp — short for complimentary — is an item that was ordered, made, delivered to a guest, and then removed from the check by a manager. The product was consumed. The revenue was not collected.

A discount is a reduction in the price of an item that was sold: an industry-night 20%, a loyalty offer, a staff meal at half price. Revenue is reduced; the item was sold at a lower price.

A void is the cancellation of an item before it was produced and delivered — a mis-keyed entry, a guest who changed their mind, a duplicate ring. Nothing was consumed and nothing was sold.

The accounting consequence differs sharply, and this is the part that trips people:

Voids never touch your P&L. They are an audit item, not an accounting item, and they matter for exactly the reason Chapter 34 will spend a section on: a void is the one transaction type that can make a sale vanish without a trace on the income statement. A server who rings an item, collects cash, and voids the item has stolen money that no financial statement will ever show. You watch voids by employee, by count, and by time of day — not because your staff are thieves, but because the pattern is invisible anywhere else.

What a comp does to every number above it

Comps do not merely reduce revenue by their menu value. They corrupt every ratio on the statement, and in the direction that makes you look worse than you are — which is why operators who do not understand the mechanism chase phantom cost problems.

🧮 Run the Numbers

\$486 of generosity, and a full point of prime cost.

Week 22 from the flash report. Comps of \$486 at menu value; the food and beverage in them was consumed and is sitting in cost of sales.

text AS REPORTED IF NOTHING HAD BEEN COMPED Net sales $30,400 $30,886 Cost of sales $8,365 $8,365 Labor $10,588 $10,588 ────────────────────────────────────────────────────────────────────── Cost of sales % 27.5% 27.1% Labor % 34.8% 34.3% PRIME COST % 62.3% 61.4%

\$486 of comps moved prime cost almost a full point — 0.98 points, to be exact. The numerator did not change at all: the product was made and the staff were paid either way. Only the denominator moved.

Annualize that. At \$486 a week, Bellwether comps **\$25,272 of menu value a year — 1.63% of revenue. The product cost inside it is only about \$7,000, so the comp policy is not primarily a food cost problem. It is a revenue** problem wearing a food cost costume, and an operator reading only food cost percentage will investigate the walk-in for a year and never find it.

Track comps as a percentage of gross sales, with a target, by authorizing manager. Bellwether's target is ≤1.0%; week 22 ran 1.56%. That 0.56-point overage is \$175 in a week and \$9,100 a year.

🤝 Hospitality

The comp that is an investment and the comp that is a leak.

None of the above is an argument against comping. It is an argument for knowing which comp you just made.

A recovery comp — the entrée that came out wrong, the twenty-minute wait on a table that was quoted ten, the bottle that was off — is an investment in a second visit, and Chapter 23 already made the case that the second visit is where the business actually lives. A dessert that turns a ruined anniversary into a story the guest tells is the cheapest marketing you will ever buy. Comp it, comp it fast, comp it before they ask, and do not make anyone perform gratitude for it.

A relationship comp — the round for the regulars, the industry friend, the neighbor who sends you business — is real marketing and should be budgeted like marketing, with a number attached and a person accountable for it.

A leak comp is the third category and it is the one that kills you: the comp with no reason written on it, the comp entered at 11:40 p.m. by the same employee eleven weeks running, the "manager comp" authorized by someone who is not a manager. It is not usually theft. Usually it is a well-meaning bartender being generous with your money because nobody ever told them what generosity costs.

The fix is not a lower comp budget. It is a required reason code on every comp, three or four choices deep, reviewed weekly by name. Chapter 34 builds the authorization structure. What this chapter contributes is the arithmetic that makes the case: 0.56 points of unbudgeted comps is \$9,100 a year, which is a line cook's worth of hours you could have scheduled instead.

One accounting instruction to close: record comps and discounts as contra-revenue, not as marketing expense. Both treatments produce identical operating profit. But if you route comps to marketing, your reported revenue is overstated, every cost percentage improves artificially, and your marketing line contains something that is not marketing. The percentages are the whole point of a restaurant P&L. Do not corrupt the denominator.

31.9 The balance sheet in brief; sales tax is not your money

The P&L covers a period. The balance sheet is a photograph of one instant: what you own, what you owe, and the difference, which is equity. Restaurants under-use it badly, partly because it changes slowly and partly because most operators were never shown one that meant anything.

Here is Bellwether's, on opening day, built entirely from the project cost and the capital stack.

It balances, because a balance sheet always does — that is what the name means. Four notes on it.

The \$75,000 tenant-improvement allowance is not equity and not a gift. Under lease accounting it is generally recorded as a lease incentive and amortized against rent expense across the lease term — which means Bellwether's reported rent expense is lower than its cash rent by roughly \$7,500 a year for ten years, and the P&L and the checkbook will disagree about occupancy by that amount every single year. Know which one you are reading. (Ask your accountant how they are handling it; treatment varies and Chapter 6 owns the lease terms.)

Pre-opening costs generally get expensed as incurred, not capitalized, under US accounting rules. They are shown as an asset above only because this is an opening-day snapshot of where the \$620,000 went; by the end of the first period most of that \$35,000 has moved to the income statement. Your accountant will tell you exactly how to treat yours.

Accumulated depreciation is where the build-out goes to die. The \$310,000 of leasehold improvements is amortized over the lease term, the equipment over its useful life, the smallwares faster. Illustratively — and your accountant sets the real schedule — \$310,000 over ten years plus \$185,000 over ten years plus \$45,000 over five is \$31,000 + \$18,500 + \$9,000 = **\$58,500 a year** of non-cash expense. That is the D&A line in Figure 31.2, and it is 3.8% of revenue: a substantial number that never leaves your bank account and never shows up in prime cost.

The debt line shrinks by the principal, not the payment. Bellwether pays \$69,500 of debt service in Year 1. Only \$39,700 of that is interest and hits the P&L; the other \$29,800 reduces the loan balance on the balance sheet and appears nowhere on the income statement.

FIGURE 31.10 — Where the $69,500 of debt service actually goes, Year 1    [the Bellwether plan]

                                    PAYMENT     INTEREST     PRINCIPAL
   SBA 7(a) note                     $54,300     $34,200       $20,100
   Equipment lease                   $15,200      $5,500        $9,700
   ──────────────────────────────────────────────────────────────────
   TOTAL DEBT SERVICE                $69,500     $39,700       $29,800
                                                  |             |
                                          P&L expense      balance sheet only
                                          (account 9030)   (reduces account 2400)

   In later years the split shifts toward principal as the balances amortize down --
   which means your reported profit RISES slightly each year even if nothing else
   changes, while your cash outflow stays flat at $69,500. Do not mistake that for
   operational improvement.

Sales tax is not your money

Of everything in this chapter, this is the one that has bankrupted the most restaurants, and it is almost entirely a psychological failure rather than an accounting one.

When a guest pays \$46 for dinner plus sales tax, the tax portion is not revenue. You are collecting it as an agent for a taxing authority. It goes into account 2200 — Sales Tax Payable, which is a liability, and it sits in your bank account until you remit it.

The problem is that it sits in your bank account.

⚠️ Where the Money Leaks

The float that isn't a float.

At an illustrative 8.25% rate — verify your own; rates vary by state, county, and city, and what is taxable varies too — Bellwether collects:

text annual taxable sales $1,550,000 x 8.25% = $127,875 collected per year per week = $2,459 per month = $10,656

If Bellwether remits monthly, then on the 20th of a month the operating account contains up to \$10,656 that is not Bellwether's — plus whatever has accrued since. On a business whose entire working-capital reserve is \$45,000, that is nearly a quarter of the cushion, and it looks exactly like cash.

Here is the mechanism, and I have watched it happen twice. February is slow. The account is thin. The rent clears, payroll clears, and the sales-tax remittance is due on the 20th — but there is a produce invoice and a repair, and the tax gets paid a little late. March is better; you catch up. Then the next slow month you do it again, and now you are a month behind, and being a month behind feels survivable because the balance still looks fine.

It is not survivable. Sales tax is generally treated as a trust obligation — money you held for someone else — and the consequences of not remitting it are categorically more serious than being late to a vendor. In many jurisdictions, penalties and interest accrue quickly, collection authority is broad, and personal liability can attach to the individual responsible for remitting, regardless of the entity structure you set up in Chapter 8. Specifics vary enormously by state and you must verify locally with an accountant.

What the disciplined operator does: move the sales tax out of the operating account. Weekly. A separate account, a standing transfer of the collected amount every Monday when the flash report is produced — the flash report in §31.6 prints the number precisely so that this transfer can happen the same morning. The money never sits somewhere it can be mistaken for yours. It costs nothing and it removes the temptation permanently.

Two other liabilities work the same way and deserve the same discipline.

Gift cards (account 2220). A gift card sold is cash received and revenue not yet earned. It is a liability until it is redeemed. A December of strong gift-card sales is a wonderful cash month and a mediocre revenue month, and an operator who books gift cards as revenue will have a spectacular December followed by a January in which people eat for free.

Tips payable (account 2230). Tips collected on cards and not yet paid out are held for employees; they are never your revenue and never your expense. Chapter 20 owns the wage-and-hour side of this, and the accounting side is simple: they pass through a liability account and touch neither revenue nor labor cost.

🔍 Check Your Understanding

1. A void, a comp, and a discount all "remove money from a check." Which one does not appear on the P&L at all, and why does that make it the most dangerous of the three?
2. Bellwether's debt service is \$69,500 and its interest expense is \$39,700. If someone tells you the restaurant's expenses were understated by \$29,800, what is the correct response?
3. Why does an operator who books gift-card sales as revenue end up with a January that looks like a collapse?

(1: The void — nothing was consumed and nothing was sold, so it touches neither revenue nor cost. It is dangerous precisely because a fraudulent void is invisible on every financial statement and can only be caught in the POS audit trail. 2: They are wrong; the \$29,800 is principal repayment, which reduces a liability rather than being an expense. It is a real cash outflow and belongs on the cash statement, not the P&L. 3: Because the revenue was recognized in December when the card was sold, so January's redemptions produce food cost and labor cost against no revenue at all.)

🍽️ The Business Plan

Checkpoint 31 of 40 — the money section opens, and the plan is made to foot.

This chapter contributes three artifacts to the plan: the three-year P&L, the chart of accounts, and the weekly flash-report design. It also contributes something less comfortable — a published list of the places the plan disagrees with itself.

1. The three-year P&L (version of record)

Year 3 occupancy is corrected to \$98,000 per the Chapter 6 escalation. Every column foots.

(Year 3 was previously published at \$362,490 operating profit with occupancy held flat at \$95,200. The \$1.00/sq ft base-rent escalation on 2,800 sq ft adds \$2,800, giving \$359,690 — 19.4%, not 19.6%.)

And the shadow column the plan owes. Years 2 and 3 inherit the Year-1 labor assumption. If §31.5's lawful column is right, the improvement curve is real but it starts six points higher. Same assumptions, same 0.8- and 1.0-point annual improvement, starting from 38.5%:

The plan carries the first table. It now also carries the second, labeled, because a plan that shows only its good case is not a plan.

2. The chart of accounts

Figure 31.4 in §31.2, adopted as written, with a one-page account-definitions document specifying what belongs in each account. Three definitions are called out explicitly because they change published percentages: property insurance sits in occupancy while general/liquor/EPLI sits in other operating; card processing gets its own account separate from other technology; comps and discounts are contra-revenue, never marketing.

3. The weekly flash-report design

Figure 31.8 in §31.6, produced Monday by 10:00 a.m., read aloud in fifteen minutes, ending in one action. The prime-cost target column ramps per Figure 31.7 rather than sitting flat at 60%. The sales tax figure it prints triggers a same-morning transfer out of the operating account.

What this checkpoint settles

• The plan foots. Every statement in this chapter adds up, in dollars, twice checked.
• Year 3 occupancy is corrected. Year 3 operating profit is \$359,690.
• The two revenue bases are named and bridged: \$1,410,760 base grid + \$139,240 bridge = \$1,550,000. Average check is quoted per daypart, per base, or not at all.
• Chapter 32 inherits Chapter 19's fixed/variable labor split: \$191,895 fixed, 24.4% variable.
• The roster mapping is published: 31 heads, 24 scheduled positions, 14.3 hours-based FTE, 453.5 hourly hours.
• The 96-minute cycle and the 95-minute dine time are reconciled definitionally.

What this checkpoint does not settle

The labor line. Three defensible numbers are on the table — 32.3%, 36.8%, and 38.5% — and this chapter deliberately does not choose. That choice belongs to Chapter 40, and it is the single largest open question in the plan.

The bridge's composition. \$139,240 is bridged in total but not attributed line by line to events, patio, off-premise, and gift cards. Chapter 32 needs that attribution because the contribution margins differ sharply by channel.

Other operating's fixed/variable split. \$217,000 is genuinely mixed and Chapter 32 must build it line by line.

Cash by week. This entire chapter is annual and weekly profitability. It says nothing about which Thursday in February the account gets thin. That is Chapter 33, and it is the chapter that actually decides whether a plan this good on paper survives its first winter.

Open questions carried forward:

1. Is the labor line 32.3%, 36.8%, or 38.5% — and what would have to change operationally to make the lower number true? (Chapters 19, 21, 32, 40)
2. What is the break-even in covers per night, using Chapter 19's split, in each of the three labor scenarios? (Chapter 32)
3. Does the \$45,000 working-capital reserve survive a first quarter running 66.6% prime cost with sales tax accruing in the same account? (Chapter 33)
4. What comp, void, and discount thresholds should the flash report flag, and who authorizes? (Chapter 34)

Conclusion

A restaurant P&L is not hard. It is revenue, then the two lines that make up prime cost, then the things you can still change, then the things you cannot, then the financing. Once you have seen the order, you have seen every restaurant statement you will ever read.

What is hard is making thirty chapters of decisions agree with each other. We found six places Bellwether's plan did not, and none of them was carelessness — they were the ordinary consequence of a plan built in pieces by people solving different problems. A roster counted three different ways. A labor line built top-down that a bottom-up schedule could not reproduce, and then a compliance finding that widened the gap by \$27,000. Two revenue bases \$139,240 apart. A lease escalation that never walked back into the pro forma. Two fixed/variable splits that each footed to their own total and nowhere else. A table cycle and a dine time measuring different clocks.

Five of the six are now closed. The sixth — the labor line — is deliberately open, published in three columns, and it is the honest state of the plan. A plan that has been reconciled and still has one large disagreement in it is worth more than a plan that has been smoothed until everything agrees, because the first one tells you where to look.

The instruments matter more than the statement. Weekly prime cost from seven inputs. A one-page flash report with a ramping target and one action at the bottom. A fiscal calendar chosen before you open. A chart of accounts written down. Sales tax moved out of the operating account every Monday. None of it is sophisticated. All of it is the difference between finding a problem in week eight and finding it in month sixteen, which is the difference this whole book has been about since Figure 1.1.

Chapter 32 takes the fixed/variable split published here and turns it into the number every operator should be able to state and almost none can: how many guests do you need tonight before you stop losing money? It is the same arithmetic from the other end — and it is where the labor disagreement in §31.5 stops being an accounting question and starts being a question about how many covers have to walk through the door.

Key Terms

Profit and loss (P&L) statement — a report of revenue and expense over a period, arriving at profit. A restaurant P&L is ordered so that controllable costs — cost of sales and labor — sit at the top, and contracted costs sit below. Also called an income statement. (Ch. 31)

Restaurant chart of accounts — the numbered list of accounts every dollar is recorded into, structured so a foodservice P&L falls out of it automatically: revenue split by category, cost of sales split food/beer/wine/spirits, labor split by department and burden, and comps and discounts as contra-revenue. (Ch. 31)

Cost of goods sold (COGS) — the cost of food and beverage product actually consumed in a period: beginning inventory plus purchases minus ending inventory, then adjusted for transfers, employee meals, and comped product at cost. Not the same as purchases. (Ch. 31)

Controllable cost — a cost a manager's decisions move within the current period: cost of sales, labor, supplies, marketing, repairs, and most utilities. The basis for holding an operator accountable. (Ch. 31)

Non-controllable cost — a cost fixed by contract or signature that arrives at the same size regardless of how the building is run this week: rent, insurance policies, contracted technology, licenses, depreciation, and interest. (Ch. 31)

EBITDA — earnings before interest, taxes, depreciation, and amortization; on a restaurant P&L, the restaurant operating income line. Useful for comparing operating performance across differently financed businesses, and dangerous because it excludes the principal portion of debt service, which is real cash. (Ch. 31)

Accrual accounting — recording revenue when earned and expense when incurred, regardless of when cash moves. The correct basis for a restaurant P&L, because it is the only one that recognizes an inventory adjustment. (Ch. 31)

Cash accounting — recording revenue and expense when money actually moves. Simpler, and systematically misleading in a business with inventory, payables, and annual prepaid expenses. (Ch. 31)

4-4-5 calendar — a fiscal calendar of twelve "months" in which each thirteen-week quarter is split into periods of four, four, and five weeks, preserving quarterly comparability. (Ch. 31)

13-period calendar — a fiscal calendar of thirteen four-week periods (364 days), in which every period contains exactly four of every weekday and exactly four weekly flash reports, making all periods directly comparable. (Ch. 31)

Weekly flash report — a one-page report produced within a business day of the operating week's close showing sales, covers, cost of sales, labor, prime cost against a ramping target, comps and voids, and a cash reconciliation. Approximate and early, which beats exact and late. (Ch. 31)

Sales tax as a liability — sales tax collected from guests is held as an agent for a taxing authority and recorded in a liability account, never as revenue. It sits in the operating account looking like cash and is generally treated as a trust obligation with serious consequences for non-remittance. (Ch. 31)

Comp — an item ordered, produced, delivered, and then removed from the check by a manager. The product cost stays in cost of sales; the revenue disappears, so every cost percentage on the statement worsens. (Ch. 31)

Void — the cancellation of an item before it was produced and delivered. It appears nowhere on the P&L and is visible only in the POS audit trail, which makes it the most dangerous of the three check reductions. (Ch. 31)

Discount — a reduction in the price of an item that was sold. Recorded as contra-revenue; the item was sold at a lower price and the product cost remains. (Ch. 31)

Spaced Review

1. Without looking back: name the seven inputs to a weekly prime cost calculation, and say which two of them require somebody to physically count something.
2. A manager reports food cost of 26.9% one week and 33.8% the next, on identical sales and identical usage. What did they compute, what should they have computed, and what would have happened if they had "fixed" the 33.8% week by buying less at the end of it?
3. Bellwether's plan shows prime cost at 60.0% and Chapter 19's bottom-up build shows 64.6%. Explain in one paragraph why the business still clears a 2.35× debt service coverage at 66.3% prime, and what single line on the P&L is doing that work.
4. From Chapter 11: the Hearth Chicken costs \$8.52 and sells for \$29.00 — a 29.4% food cost. Bellwether's adjusted food cost of sales in §31.3 came out at 29.1%. Are those two numbers comparable? What three adjustments had to be made before they were?
5. From Chapter 1: Figure 1.4 called 65–70% prime cost "distressed." Bellwether's lawful-labor case is 66.3%. Reconcile those two statements without changing either number.
6. The recurring question: an operator moves from a monthly calendar to a 13-period calendar mid-year and reports that period-over-period revenue "improved." What is the first thing you check, and why should you refuse to believe the comparison at all?