Chapter 1: The Restaurant Business: Why Most Fail, Why Some Succeed, and What You Need to Know Before Day One

"Everybody wants to talk about the food. Nobody wants to talk about the walk-in." — constructed; the thing every veteran operator eventually says to somebody

Overview

The numbers in front of you are from a restaurant everybody loved. Two hundred and forty covers on a Saturday, a chef who got written up twice in the city paper, a bar three deep at eleven. It closed in March. I know, because I was standing there when the owner handed back the keys, and I know exactly which line on this profit-and-loss statement killed it. Food cost ran 34.5% for eleven months while he told himself it was 30%, because he had never once counted his walk-in on the last day of the month. Four and a half points on eight hundred sixty thousand dollars of food sales is thirty-eight thousand seven hundred dollars a year — almost exactly what he was short when the rent escalation hit in year three. He did not have a food problem. He had a counting problem, and by the time a counting problem reaches your bank balance it has been running for a year.

That is what this book is about. Everyone told him the food was the hard part. The food was the easy part.

This chapter does three things. First, it clears away the industry's most famous piece of folklore — the claim that ninety percent of restaurants fail in year one — because a book that intends to teach you to be rigorous about numbers has no business opening with a scary one it did not check. The real figure is smaller and considerably more useful. Second, it follows a restaurant dollar from the register to whatever is left at the end, so you can see the actual shape of the business: a low-margin, high-volume operation where a handful of cost lines decide everything. Third, it introduces prime cost, the sum of what you spend on food, beverage, and labor, which is the number this entire book orbits.

By the end you will not know how to open a restaurant. You will know what kind of problem opening a restaurant is — a financial and operational one, wearing a chef's coat — and you will have the vocabulary and the arithmetic to follow everything that comes after.

In this chapter, you will learn to:

• Separate the myth of restaurant failure from what the research actually documents, and explain why the correction makes the picture more alarming rather than less.
• Follow a dollar of revenue through every cost category on a restaurant P&L and say where it went.
• Compute prime cost from a set of operating numbers, and interpret it against the benchmark for a given service style.
• Name the four mechanisms that actually close restaurants and describe how each one announces itself before it is fatal.
• Compare quick service, fast casual, full service, fine dining, and bar-driven concepts on cost structure, and explain why a technique that works in one may bankrupt another.
• Estimate annual revenue from seats, turns, average check, and operating days — the arithmetic every chapter after this one assumes you can do.

Learning Paths

🏗️ Opening — this chapter is your foundation; read all of it, and do the arithmetic in §1.2 and §1.3 by hand. The failure mechanisms in §1.4 are the ones your business plan must answer. 📋 Managing — weight §1.3 and §1.4. If you manage someone else's restaurant, prime cost is the number your performance is actually judged by, whether or not anyone has told you so. 🍸 Beverage — note in §1.2 how beverage cost and food cost blend into a single COGS line, and why a strong bar program quietly subsidizes a kitchen. §1.6's bar-driven column is yours. 🚚 Small Format — §1.6 explains why a truck's economics differ, and §1.4 matters more to you, not less: small formats have thinner cushions and shorter warning periods.

1.1 The myth and the math: what the failure research actually says

Ask anyone — a banker, a culinary student, a person at a party who has just learned what you do — and you will hear the same number. Ninety percent of restaurants fail in the first year.

It is not true. It has never been demonstrated. It is usually attributed to a study at a large university that nobody has ever been able to produce, and it has been repeated so often, in so many otherwise careful places, that it has acquired the texture of fact. If you go looking for the source, you find citations pointing at other citations pointing at nothing.

What we do have is real research. The best-known work on the question comes from H.G. Parsa and colleagues, published through Cornell, which followed restaurant openings in a defined market over several years and actually counted what happened to them. The finding, in round terms:

• Roughly one in four restaurants does not reach its first anniversary — on the order of 26–27%, not 90%.
• Close to six in ten are gone within three years.
• "Failure," in this literature, generally means the business closed or changed ownership. That is not the same as bankruptcy, and it is not always a tragedy — some of those closures are a founder selling a going concern, retiring, or taking a better offer on the lease.

It is worth being precise about the uncertainty here. Studies of this question differ in market, period, and definition, and the figures move accordingly. Treat "about a quarter in year one, approaching six in ten by year three" as the honest shape of the thing rather than as a decimal you can quote to a lender.

🧮 Run the Numbers

Why the correction makes things worse, not better.

Suppose the myth were true: 90% failure in year one. What would that describe? A business whose outcome is essentially random — struck by lightning, unpredictable, not worth studying. If nine out of ten competent people fail immediately, skill is not the variable.

Now take the real numbers. Start with 100 restaurants:

• End of year 1: about 74 are still open.
• End of year 3: about 40 are still open.

So the largest share of the casualties happen in years two and three — after the honeymoon, after the press, after the opening loans are spent, after the first rent escalation. Thirty-four of those hundred restaurants survived a full year and then died.

A business that dies in month eleven was probably wrong from the beginning: bad location, bad concept, catastrophic undercapitalization. A business that dies in month twenty-nine was, at some point, working. Something drained it slowly enough that nobody caught it.

That is a bleeding pattern, not a lightning pattern. And bleeding is countable, visible, and — this is the entire premise of this book — preventable by people who count.

FIGURE 1.1 — Survival, out of 100 openings                    [after the published failure research]

  100 ┤████████████████████████████████████████████████████  opening
      │
   74 ┤█████████████████████████████████████                  end of year 1   (-26)
      │
   57 ┤████████████████████████████                           end of year 2   (-17)
      │
   40 ┤████████████████████                                   end of year 3   (-17)
      └──────────────────────────────────────────────────────
        Figures rounded from the published research; treat as shape, not precision.
        Note where the losses actually cluster: years two and three, not year one.

The shape of that curve is the argument of this book compressed into eight lines of ASCII. The restaurants dying in year two and year three are not being killed by bad food. Bad food kills you in month four. What kills you in month twenty-nine is a cost structure that was two or three points wrong the entire time, running quietly underneath a dining room that looked busy.

⚠️ Where the Money Leaks

The myth is not harmless. It causes two specific, expensive errors.

Error one: fatalism. If you believe failure is a 90% coin flip, there is no point building systems. Why count inventory weekly if the outcome is random? The myth licenses exactly the negligence that produces the real failures.

Error two: false relief. An operator who makes it past twelve months and believes the 90% figure concludes they are through the dangerous part. They are not. Statistically they have just walked into it. Year two is when the opening capital is gone, the novelty traffic has normalized, the first lease escalation lands, and the staff who opened the place start leaving.

Why a wrong number survives

It is worth spending a moment on why the myth is so durable, because the mechanisms that keep it in circulation are the same ones that will hand you a bad foot-traffic estimate from a broker, a bad yield figure from a vendor, and a confident "everybody in this neighborhood does forty percent bar" from the person trying to sell you their restaurant. Learning to see through this particular number is practice for the rest of them.

Four things keep it alive.

Closures are events; survivals are not. A restaurant that closes generates a story — a sign on the door, a post, a paragraph in the local paper, three weeks of people saying they loved that place. A restaurant that has been quietly open for eleven years generates nothing at all. What you hear about is therefore heavily enriched with closures, in a way that has nothing to do with how common closures actually are. That is not a flaw in the research; it is a flaw in the sample your memory collected. Try it right now: name the restaurants near you that have closed in the last two years. Now name the ones that didn't. The second list is longer, much harder to produce, and almost nobody ever makes it.

"Closed" and "changed ownership" get counted together — and then the word "failed" is applied to both. The research is explicit that failure generally means the business closed or changed ownership, and that is a defensible definition for its purpose: from the standpoint of the original venture, both are an ending. It travels badly, though. By the third retelling, "closed or changed ownership" has become "failed." By the fifth, "failed" has become "went under and lost everything." Nobody along that chain is lying. Each step is a small compression, and compression always runs toward the more dramatic word.

A restaurant closing is not the same as a restaurant failing. This one deserves to be said flatly, because it is the distinction the statistic most obscures. Restaurants close solvent all the time:

• A ten-year lease ends and the landlord has a national tenant willing to pay forty percent more.
• An owner is sixty-four and tired, and nobody in the family wants it.
• A partnership dissolves — a divorce, a death, a falling-out — and the cleanest resolution is a sale.
• A concept is deliberately wound down because the operator wants the room for a different one, or wants the capital for a second site.
• The chef takes a job that pays more than the restaurant ever will.
• The place was always a pop-up, a residency, or a five-year plan, and it reached the end of it.

Every one of those is a closure. Several of them are a success. The statistic cannot tell them apart, and the person quoting it to you at a party has never once asked the question that separates them: did the owner walk away with money, or without it?

And a frightening number is more repeatable than an accurate one. "Ninety percent of restaurants fail in the first year" is simply a better sentence than "roughly a quarter don't reach the first anniversary, the definition includes ownership changes, and the figure moves with market and period." It is shorter, it is memorable, and it does useful work for a surprising number of people. It flatters the operator who survived. It excuses the one who didn't. It helps a consultant sell a service and a landlord justify a personal guarantee. It discourages the competition. A statistic with that many satisfied customers does not need to be true in order to survive.

The myth also circulates in a softer form — sixty percent fail in the first year — which is equally unsourced and equally false, and which is probably a corruption rather than an invention. Notice how close it sits to the genuine three-year figure. A cumulative number, quoted over three years and repeated by people who did not write down the period, collapses very easily into a first-year number. That is very likely how it happened, and it is a useful reminder that the most dangerous errors are not the wild ones. They are the ones that began as something true and lost a qualifier.

🧮 Run the Numbers

How many of the closures are actually failures?

Take the cohort from Figure 1.1 again: 100 openings, 40 still open at the end of year three. Sixty are gone.

Nobody has counted how many of those sixty left solvent, so we are going to make an assumption and label it as one. Assume — illustratively, and deliberately generously — that one in five of the exits after year one is voluntary and solvent: a sale of a going concern, a retirement, a lease not renewed, a planned wind-down. Apply it only after year one, because almost nobody sells a nine-month-old restaurant at a profit. A going concern takes time to become one.

text Exits during year 1 26 essentially all distressed Exits during years 2-3 34 of which voluntary/solvent (20%) 7 (6.8, rounded) of which distressed 27 ───────────────────────────────────────────── Total exits by end of year 3 60 voluntary / solvent 7 DISTRESSED 53

So even under an assumption written to be kind to the industry, 53 of 100 openings are gone within three years for reasons nobody chose — and 27 of those had been open for more than a year before they died.

The correction to the myth rescues nobody. It relocates the problem. It says the danger is not a lightning strike in month six; it is a slow one in month twenty-nine, in a business that was working.

What to do with a number somebody hands you. For the rest of this book — and for the rest of your working life — when anyone quotes you an industry statistic, ask three questions before you spend a dollar against it: over what period, in what market, and what exactly counted? If the person cannot answer all three, the number is folklore and you should treat it as folklore, however confidently it was delivered. You are going to be quoted a great many numbers by people who want something from you. This is the filter.

What "failure" actually looks like from inside

Restaurants very rarely close on a single catastrophic day. The pattern, in my experience and in almost every closure I have watched up close, runs something like this:

1. Months 1–4. Revenue is strong. Everyone is thrilled. Costs are high, but everyone expects that — it's opening. Nobody is measuring against a benchmark because there isn't one yet.
2. Months 5–10. Revenue normalizes downward from the opening spike. This is normal and expected. Costs, however, do not normalize downward, because the systems that would have brought them down were never built. Food cost is "about thirty" — a number nobody has computed. The schedule is written by feel.
3. Months 11–18. The owner notices there is less money than there should be. There is a period — this is the critical window — where the problem is diagnosable and fixable, costing perhaps three to six points of prime cost.
4. Months 19–30. Something exogenous arrives: a rent step, a bad winter, a piece of equipment, a key employee leaving, a road closure, a new competitor. It is not large. It is simply larger than the cushion, and the cushion was spent in step two.
5. The end. Which is usually a landlord conversation, not a bankruptcy filing.

Notice that step three is where this book lives. Almost every technique you will learn — costing a recipe, running a menu-engineering analysis, building a staffing guide, reading a weekly flash report — exists to move the discovery in step three from month sixteen to month two.

1.2 Where the money goes: the anatomy of a restaurant dollar

Here is the thing about restaurants that surprises people who have not run one: the margins are terrible, and the volume is enormous. A restaurant doing $1.2 million a year in sales — which sounds like a great deal of money, and is a perfectly respectable independent — might keep three to five cents of each dollar. A very good year might be six or seven.

Compare that to almost any other small business and it looks insane. Software keeps seventy cents on the dollar. A law practice keeps thirty or forty. A restaurant keeps four, and it has to handle perishable inventory, employ twenty-five people, satisfy a health department, and do it all again tomorrow because none of today's product survives to be sold again.

This is not a complaint. It is the defining structural fact of the business, and everything else follows from it. When you keep four cents on the dollar, a four-point error is not a problem. It is the whole business.

The standard cost categories

Restaurant financials use a fairly consistent set of buckets. You will meet all of them properly in Chapter 31; here is the map.

Those ranges are industry rules of thumb, and they are ranges for a reason — a coffee shop and a steakhouse do not share a cost structure. Do not treat them as targets for your specific restaurant. Treat them as the shape of the thing, and build your own numbers from Chapter 31 onward.

Notice one thing immediately: the first two lines are more than half the business. That observation is important enough to have its own name, and it gets §1.3 to itself.

FIGURE 1.2 — Where a dollar of full-service revenue goes    [constructed teaching example]

  Cost of goods sold      ██████████████       30¢  ┐
  Labor (all-in)          ████████████████     34¢  ┘  PRIME COST = 64¢
  Occupancy               ████                  8¢
  Other operating         ████████             17¢
  General & admin         ██                    4¢
  ────────────────────────────────────────────────
  = Operating profit      ███                   7¢     before debt service and taxes

  Of that 7¢: roughly 3–4¢ typically goes to debt service on the money that built
  the restaurant. What the owner actually keeps is the remainder — and in many
  independents, the "remainder" is the owner's own salary for a 65-hour week.

That last line is not a joke and it is not cynicism. A very large fraction of American independent restaurants are, financially speaking, a job that owns a building lease. There is nothing wrong with that — it can be an excellent job — but you should know which business you are entering, and you should know it before you sign a personal guarantee.

👨‍🍳 On the Line

What "thin margin" feels like on a Tuesday.

Abstract percentages do not convey this, so here it is concretely. You are the manager. It is a Tuesday in February. You have 41 covers on the books and two servers scheduled.

At 6:15 you decide whether to cut one of them. If you cut and a nine-top walks in at 7:00, the remaining server drowns, four tables have a bad night, and two of them do not come back — a cost you will never see on any report. If you don't cut, you have paid a server roughly four hours to work a section that produced maybe $180 in sales.

That single decision is worth about $60 either way. It seems too small to matter. You make some version of it about six hundred times a year, and the aggregate of those six hundred decisions is larger than your entire annual profit.

This is why systems beat instinct. Not because instinct is bad, but because instinct does not aggregate. A staffing guide built from sales-per-labor-hour data (Chapter 19) makes that decision the same way every time, and the same way is worth more than the right way six hundred times a year.

Reading a real one

Let's go back to the restaurant from the Overview — the one that closed in March. Here is what its final full year actually looked like.

🧾 Read the Numbers

```text FIGURE 1.3 — "The restaurant everybody loved" [constructed teaching example] THE ARTIFACT Annual profit-and-loss summary, final full year of operation. A 92-seat neighborhood American restaurant, dinner six nights, full bar. THE CONTEXT Year three. Strong reviews, strong Friday and Saturday, soft Tuesday through Thursday. Owner-operator, no general manager. Bookkeeping done monthly by an outside firm; inventory counted "when there was time."

                 Revenue                                  $1,200,000    100.0%
                   Food sales           $860,000
                   Beverage sales       $340,000
                 Food cost        (34.5% of food sales)     $296,700
                 Beverage cost    (22.0% of bev sales)       $74,800
                 TOTAL COGS                                 $371,500     31.0%
                 Labor, all-in                               $408,000     34.0%
                 ─────────────────────────────────────────────────────────────
                 PRIME COST                                  $779,500     65.0%
                 Occupancy                                    $96,000      8.0%
                 Other operating                             $204,000     17.0%
                 General & administrative                     $48,000      4.0%
                 ─────────────────────────────────────────────────────────────
                 OPERATING PROFIT                             $72,500      6.0%
                 Debt service                                 $42,000
                 NET                                          $30,500      2.5%

WHAT IT SHOWS A restaurant that was profitable — barely. It cleared $30,500 on $1.2 million of sales. The single largest deviation from plan is food cost: 34.5% against a 30% target, on $860,000 of food sales, is $38,700 of margin that should have been there and wasn't. Had food cost held at 30%, operating profit would have been $111,200 (9.3%) and net $69,200 (5.8%) — more than double. WHAT IT DOESN'T It does not say WHY food cost ran 34.5%. That number could be theft, over-portioning, waste, spoilage, poor purchasing, uncosted specials, menu drift, or simply prices that were set two years ago against costs that have moved since. An annual P&L identifies the wound; it never identifies the weapon. Chapters 11 and 13 do that. It also does not show cash — which month the account got thin, or how close this business came to missing payroll in February. THE DECISION Count inventory every single week, starting this week, and compute food cost weekly against theoretical usage. Nothing else on this statement is worth touching until you know which of the seven possible causes it is. THE LESSON A restaurant does not need to be unprofitable to close. It needs to be insufficiently profitable when something ordinary goes wrong — and $30,500 of annual cushion does not survive one rent escalation and one bad winter. ```

Read that P&L again and notice what is not wrong with it. Labor at 34% is unremarkable. Occupancy at 8% is fine. Other operating at 17% is a little high but within normal. General and administrative at 4% is fine. Revenue of $1.2 million on 92 seats is healthy.

One line was four and a half points off, and it consumed more than half of the profit. That is the business. That is what a four-cent margin means in practice.

1.3 Prime cost: the one number that predicts survival

If you take a single thing from this book, take this one.

Prime cost is the sum of two things: your total cost of goods sold — food and beverage — and your total labor cost, including wages, payroll taxes, and benefits, for everyone in the building from the dishwasher to the owner's salary. Expressed as a percentage of total sales:

$$\text{Prime cost \%} = \frac{\text{COGS} + \text{Total labor}}{\text{Total sales}}$$

In the P&L above: $(\$371{,}500 + \$408{,}000) \div \$1{,}200{,}000 = 65.0\%$.

Why this number and not another

Three reasons, and they compound.

First: it is most of your controllable cost. Rent is fixed the day you sign. Your insurance premium is what it is. Your utility bill moves a little with volume and season but not with your decisions. Food, beverage, and labor are the categories where a manager's choices — what to buy, what to charge, who to schedule, how to portion — change the outcome this week. Prime cost is, almost exactly, the set of costs you can actually do something about.

Second: the two halves trade off against each other, which means watching either one alone will mislead you. This is the part most operators miss, and it is why "what's your food cost?" is the wrong question.

🧮 Run the Numbers

Why food cost alone lies.

Two restaurants, both doing $1,200,000 in sales.

	Restaurant A	Restaurant B
COGS	26.0% ($312,000) | 32.0% ($384,000)
Labor	39.0% ($468,000) | 30.0% ($360,000)
Prime cost	65.0% ($780,000)** | **62.0% ($744,000)

Restaurant A has excellent food cost. If you asked its owner at a party, they would tell you 26% and you would be impressed. Restaurant A is also losing $36,000 a year relative to Restaurant B, because it is buying that low food cost with labor — scratch production, house-made everything, extensive butchery, three extra prep hours a day.

That is not automatically wrong! Scratch production may be exactly the right choice for the concept, and it may be why guests come. But it is a choice with a price, and an operator who only watches food cost cannot see the price.

Conversely, Restaurant B buys portioned proteins and some prepared components. Its food cost is six points worse and its labor is nine points better. On prime cost, it wins.

The trade is real and it runs in both directions. Every "we should make this in house" decision and every "we should buy this pre-portioned" decision moves cost from one half of prime to the other. The only way to know whether it was a good trade is to watch the total.

Third: it has a benchmark that actually means something. For full-service restaurants, the widely used rule of thumb is prime cost at or below 60% of sales, with the mid-60s being survivable but tight and 70%+ generally indicating a business in trouble. For quick service, the target is lower — often around 55% — because the model runs less labor. These are rules of thumb, not laws; a fine dining room with an extraordinary check average and a big beverage attachment may run higher and survive. But when an operator tells me their prime cost is 68%, I already know a great deal about their year without seeing anything else.

FIGURE 1.4 — Prime cost, read as a diagnosis          [industry rules of thumb; ranges, not laws]

   ≤ 55%   ████                    strong; likely QSR/fast casual or an exceptionally
                                   well-run full-service operation
   55–60%  ████████                healthy full service; the target Chapter 31 builds toward
   60–65%  ████████████            workable but tight; little cushion for anything going wrong
   65–70%  ████████████████        distressed; profit is mostly gone, cash is probably strained
   > 70%   ████████████████████    the business is consuming itself; act now, not next quarter

   Read alongside occupancy: a restaurant at 63% prime with 5% rent is in a very
   different position from one at 63% prime with 11% rent.

A prime-cost walk, from raw lines to a diagnosis

Assertion is cheap. Here is the whole calculation, on a restaurant that is not the running example of this book, from the raw lines exactly as they arrive to a diagnosis you could act on Monday.

The restaurant. A 74-seat neighborhood Italian room, dinner six nights a week, a small bar, one chef and one general manager on salary. This is a constructed teaching example — the figures are realistic and internally consistent, and they are not any real business's records.

Step one: get the sales, and split them.

Nobody hands you a number called "prime cost." They hand you a point-of-sale report, a payroll register, and a folder of invoices. Start at the top.

Beverage is 21.9% of sales (\$285,000 ÷ \$1,300,000) — a modest bar, which is going to matter in about thirty seconds.

Step two: get cost of goods sold, both halves.

Counted properly — beginning inventory plus purchases minus ending inventory, the method the end of this section returns to — this restaurant used \$314,650 of food and \$62,700 of beverage over the year.

Look at what just happened, because it is one of the most common arithmetic confusions in this business. Food cost is 31.0% and total COGS is 29.0%, and both numbers are correct. They have different denominators: food cost is measured against food sales, COGS against all sales. The bar, at a 22.0% pour cost, drags the blended figure down. An operator who compares their own 31.0% food cost against somebody else's 29.0% COGS and concludes they are losing has compared two different things — which is the first of several reasons the industry's favorite question is a bad one.

Step three: get labor. All of it.

That final addition is where most operators' number quietly goes wrong. Ask this restaurant's general manager what labor runs and you will hear "about twenty-four percent" — and they are not making it up. They mean the hourly crew: \$316,400 of wages and overtime against \$1,300,000 of sales is 24.3%, which is a perfectly respectable figure. But two salaried people are another \$118,000, and the payroll taxes, workers' compensation, and benefits are \$82,500 — just under nineteen cents on every wage dollar, arriving automatically the moment anybody is on the clock. Total labor is \$516,900, which is 39.8% of sales. The number the manager quotes and the number on the statement are nearly sixteen points apart, and neither of them is a lie.

Step four: prime cost, and the diagnosis.

$$\text{Prime cost} = \frac{\$377{,}350 + \$516{,}900}{\$1{,}300{,}000} = \frac{\$894{,}250}{\$1{,}300{,}000} = 68.8\%$$

Read that against Figure 1.4 and you have a diagnosis before you look at anything else on the statement: distressed. Now read the two halves separately, which is the point of this whole section. COGS at 29.0% is fine. Food at 31.0% of food sales is a shade high for the style and worth a look, but it is not what is wrong here. Labor at 39.8% is what is wrong here — six to ten points above where a casual full-service room ought to sit.

And if you had asked this operator the industry's favorite question — what's your food cost? — you would have heard "thirty-one, a little high, we're working on it," and you would have learned absolutely nothing about the thing that is killing them.

Step five: finish the statement, and then move prime cost by two points.

The rest of the costs are ordinary, and none of them is the problem: occupancy \$104,000 (8.0% of sales), other operating \$208,000 (16.0%), general and administrative \$45,500 (3.5%). Those three total \$357,500.

Note the thing that trips people up: this restaurant is profitable and distressed at the same time. It cleared ten thousand dollars. It is also carrying a prime cost that Figure 1.4 puts in the band where the profit is mostly gone and the cash is probably strained. Both readings are correct, and the combination is exactly the bleeding pattern §1.1 described — a business that works, for a while, on a cushion too thin to absorb anything.

Two points of prime cost on \$1,300,000 is \$26,000. Move it both ways.

FIGURE 1.5 — What two points of prime cost is worth      [constructed teaching example]

                          two points        as counted        two points
                            better                              worse
   Prime cost              $868,250          $894,250          $920,250
                             66.8%             68.8%             70.8%
   All other cost          $357,500          $357,500          $357,500
   ──────────────────────────────────────────────────────────────────────
   Operating profit         $74,250           $48,250           $22,250
   less debt service        $38,000           $38,000           $38,000
   ──────────────────────────────────────────────────────────────────────
   = NET                    $36,250           $10,250         ($15,750)

   Two points is $26,000. The full swing across four points is $52,000, on a
   business whose best case here nets $36,250. The cost line moves a little.
   The bottom line moves by more than its own size. That is the leverage, and
   it runs in both directions.

Two points. Not a catastrophe, not a scandal, not anything anybody would put in a memo — two points, in a business whose own rules of thumb (Figure 1.4) are written in five-point bands. It is the difference between netting \$36,250 and losing \$15,750. On a four-to-six-point margin, the profit line moves by more than its own size, and that single fact is the reason the operators who count every week are not being neurotic.

The diagnosis is also now specific enough to be useful. It is not "we need to raise prices." It is: the labor half of prime is six to ten points heavy, and I do not yet know whether that is the schedule, the wage rates, the salaried load at this volume, or a menu that takes too many hands to execute. Those are four different investigations with four different fixes, and Chapters 19 and 12 are where you run them. But you cannot begin any of them until somebody has done Step three honestly, including the \$82,500 that nobody ever remembers.

👨‍🍳 On the Line

Where two points actually live.

Two points of prime cost on that restaurant is \$26,000 a year. Divide it down: \$500 a week, and across six services, about \$83 a night.

Eighty-three dollars. Here is one entirely unremarkable Thursday:

	Cost
A server held ninety minutes past the point anyone needed them (\$16.60/hr with taxes and benefits × 1.5) | \$25
Thirty entrées plated an ounce heavy, on a \$9.00/lb protein (1.9 lb) | \$17
One comped entrée and one comped round, at cost	\$15
A prep cook forty minutes past close (\$17.80/hr with taxes and benefits × 0.67) | \$12
Ten pounds of usable trim into the bin instead of the stock pot (\$1.40/lb) | \$14
Total	\$83

(Each line rounded to the dollar.)

Not one of those is a fireable offense. Not one of them appears on any report anybody reads. The comp was gracious. The server was kept because at 7:40 it looked like it might get busy. The extra ounce made the plate look better and the cook was proud of it. The trim went in the bin because the stock pot was already full and there was nowhere to put it. Every one of them is a defensible decision made by a competent person having a normal night.

That is the difficulty of this business in one table. The money does not leave through a hole. It leaves through six hundred small doors, every one of which somebody had a good reason to open, and the only thing that closes them is a number somebody looks at every week.

Computing it weekly, which is the whole point

Here is the operational discipline that separates the survivors, and it is almost embarrassingly simple: compute prime cost every week.

Not monthly. Not when the accountant sends the statement six weeks after the period closed. Weekly.

The reason is arithmetic. If you compute prime cost monthly and the statement arrives three weeks after month-end, you learn about a problem seven weeks after it started. Seven weeks of a four-point overrun on a $23,000-a-week restaurant is about $6,400 — real money, and more importantly, seven weeks of the cause running unchecked and becoming habitual. Weekly, you catch it in eight days.

To do it you need three inventories (food, beverage, and a purchases figure) and your payroll for the week. It takes a trained manager under two hours. Chapter 31 builds the one-page weekly flash report that carries it.

⚠️ Where the Money Leaks

"I know my food cost."

Almost nobody who says this can produce the calculation. Ask a follow-up: how do you know?

There is exactly one honest answer, and it involves counting. Food cost is:

$$\text{Food cost} = \frac{\text{beginning inventory} + \text{purchases} - \text{ending inventory}}{\text{food sales}}$$

If you have not counted your ending inventory, you do not have the numerator. What most operators actually mean by "my food cost is 30%" is "I divided my food invoices by my food sales," which ignores the fact that inventory levels move. Buy heavy the last week of the month and your "food cost" spikes; run the walk-in down and it magically improves. Neither has anything to do with what you actually used.

The restaurant in Figure 1.3 believed its food cost was 30% for eleven months. The invoices-over-sales shortcut told it so, roughly, and nobody counted. The full method is in Chapter 13.

1.4 The four real killers: undercapitalization, cost drift, labor, and cash timing

Restaurants close for four financial reasons. Concept problems, bad locations, and bad food are real, but they act through these four mechanisms — a bad concept kills you by producing insufficient revenue against a fixed cost base, which is a cash problem.

Killer one: undercapitalization

Undercapitalization means opening with less money than the business will need before it reaches sustainable operation. It is the single most common cause of first-year failure, and it is almost always a planning error rather than a bad-luck event.

The mechanism is straightforward. Build-outs run over — not occasionally, routinely. Openings slip, and every week of slippage is rent paid on a building generating nothing. Revenue ramps more slowly than the plan assumed, because the plan assumed the opening spike was the baseline. Meanwhile you are carrying full payroll from two weeks before opening.

The result is that an operator who budgeted precisely enough to open has, by definition, budgeted nothing to operate. They open with an empty reserve, and then the first slow month is existential rather than annoying.

⚠️ Where the Money Leaks

The reserve is not optional, and it is not the contingency.

Two different pots of money get confused constantly:

• The construction contingency covers build-out overruns. Budget 10–15% of the construction number. You will use it. If you don't, you were lucky.
• The working-capital reserve covers operations after you open: payroll, inventory, and rent during the months when revenue does not yet cover them. This is separate money.

An operator who raises the contingency and calls it the reserve has done the arithmetic wrong twice. Chapter 33 builds the reserve properly. Chapter 5 explains why lenders ask about it — and why an application without one gets declined.

The running project in this book budgets \$45,000 of working-capital reserve inside a \$620,000 project. Whether that is enough is a live question the plan will have to defend.

Killer two: cost drift

This is the one that killed the restaurant in Figure 1.3, and it is the most insidious because nothing announces it.

Cost drift is the slow upward movement of prime cost that occurs when nothing is measured. Its components are individually trivial:

• A protein price rises 8% and menu prices don't move.
• A cook starts plating six ounces instead of five because it looks better.
• A new prep cook wastes a little more trim than the last one.
• Three specials a week are never costed at all.
• Comps run 2% instead of 0.8% because nobody is reviewing the comp report.
• A new server gets an extra half-hour on the clock every shift for a month.

None of these is a scandal. Together they are three to five points of prime cost, which on a $1.2 million restaurant is $36,000 to $60,000 a year — the whole profit.

The countermeasure is the entire middle of this book: standardized recipes (Chapter 10), cost cards (Chapter 11), specs and par levels (Chapter 13), a staffing guide (Chapter 19), and the weekly flash report (Chapter 31) that makes drift visible within eight days rather than eleven months.

Killer three: labor

Labor deserves separate mention from cost drift because it moves faster and for different reasons.

Food cost changes over a purchasing cycle. Labor changes this shift. A manager who does not cut the floor at nine on a dead Tuesday has spent money that no subsequent decision recovers. Multiply by three hundred and twelve services a year.

Labor also has a structural component that is hard to see. Every restaurant has a fixed labor floor — the salaried managers, the chef, the opening prep cook, the closing dishwasher — that must be paid whether you do 40 covers or 140. Below a certain volume, that floor is a catastrophic percentage of sales. This is why slow dayparts are so dangerous: a lunch service doing $900 with $400 of labor on the floor is running 44% labor on that service, and it is dragging the whole week's number with it. Chapter 32 shows you how to find out whether lunch is actually carrying its weight or quietly subsidized by dinner.

And labor has a third dimension: turnover. An industry running roughly 75% annual turnover is paying, over and over, for recruiting, onboarding, training, and the reduced output of people who are still learning the menu. Chapter 17 makes you compute that cost for a specific position, because until it is a number nobody funds the prevention.

Killer four: cash timing

This is the one that surprises people, and it is the reason Chapter 33 exists.

A profitable restaurant can run out of money. Profit is an accounting result over a period. Cash is what is in the account on a Thursday when payroll clears. They are not the same, and in a restaurant they diverge for specific structural reasons:

• You get paid instantly — cards settle in a day or two — which sounds like an advantage and produces a dangerous illusion of liquidity. Money arriving daily feels like money you have.
• Payroll lands every two weeks and does not negotiate.
• Produce and protein invoices land weekly, dairy twice a week.
• Rent lands on the first, and does not care that the first is a Tuesday in February.
• Sales tax is collected daily and remitted monthly or quarterly — and it was never your money, though it has been sitting in your account making the balance look healthier than it is.
• Insurance, licenses, and equipment maintenance arrive in irregular lumps, frequently in the slow season.

Put those together and there are specific weeks each year — usually late January and February in most American markets — when obligations pile up against the lowest revenue of the year. An operator who has never built a thirteen-week cash forecast does not know which week that is until it arrives.

All four, on one page

You will rarely meet these mechanisms one at a time. They compound. Cost drift eats the cushion, which turns an ordinary February into a cash-timing crisis, which gets solved by not replacing a departing cook, which puts the rest of the crew on overtime and moves the labor line. Here is a statement with three of the four running in it simultaneously, from a restaurant nobody in its market would have described as being in trouble.

🧾 Read the Numbers

```text FIGURE 1.6 — "The biggest restaurant in town" [constructed teaching example] THE ARTIFACT One-page annual profit-and-loss summary with the prior year beside it, plus the guest count. A 120-seat suburban American grill, lunch and dinner seven days, full bar, freestanding building with its own parking. THE CONTEXT Year seven, and the best-known restaurant in its market -- the place people take visiting family. Three-page menu. Two rounds of menu price increases in the last eighteen months. Books done monthly by an outside firm; no weekly numbers of any kind. The owner's summary of the year, delivered at a chamber lunch: "sales are up."

                                                THIS YEAR       LAST YEAR
                 Revenue                       $2,340,000      $2,199,000    +6.4%
                   Food sales      $1,825,000
                   Beverage sales    $515,000
                 Food cost      (32.8% of food sales)  $598,600
                 Beverage cost  (26.0% of bev sales)   $133,900
                 TOTAL COGS                             $732,500    31.3%
                 Labor, all-in                          $846,000    36.2%
                 ─────────────────────────────────────────────────────────
                 PRIME COST                           $1,578,500    67.5%     (64.9% last year)
                 Occupancy                               $210,600     9.0%
                 Other operating                         $397,800    17.0%
                 General & administrative                 $70,200     3.0%
                 ─────────────────────────────────────────────────────────
                 OPERATING PROFIT                         $82,900     3.5%
                 Debt service                             $96,000
                 NET                                    ($13,100)    -0.6%

                 Covers                            78,400          81,200    -3.4%
                 Sales per cover                   $29.85          $27.08   +10.2%

WHAT IT SHOWS A restaurant whose revenue line is the healthiest thing about it. Sales rose 6.4% and the business lost $13,100. Prime cost is 67.5% against a full-service rule of thumb at or below 60%, and it moved 2.6 points in a single year -- $60,840 on this year's sales, more than four times the size of the loss. Occupancy at 9.0% is the aggravating condition Figure 1.4 told you to read alongside prime: 67.5% prime with 9.0% rent leaves nothing at all to work with. And the composition of the revenue growth is the real finding. Covers fell 3.4% while sales per cover rose 10.2%. The price increases produced the growth; the dining room did not. Fewer people came, each paid more, and it still did not cover what costs did. WHAT IT DOESN'T It does not say which half of prime moved, because last year's split is not on this page -- and a 2.6-point move in food is an entirely different investigation from a 2.6-point move in labor. It does not say why covers fell: price, a new competitor down the road, a service problem, or one daypart collapsing while the rest holds. It does not break covers out by lunch and dinner, which is exactly where a decline like this usually hides. And it says nothing about cash -- a business losing $13,100 on paper may have been out of money in February or may be perfectly liquid. THE DECISION Monday: start counting weekly, with food and labor computed separately, so that within eight days you know which half moved. Pull covers by daypart for the last twenty-four months. The one thing you do not do is raise prices a third time -- that is the lever that produced this statement, and pulling it again buys another year of the same. THE LESSON Revenue is the least informative number on a restaurant P&L, and it is the only one everybody quotes. "Sales are up" is not a result. It is a question: up on more guests, or up on a higher price charged to fewer of them? ```

Count the mechanisms in that statement. Cost drift is unmistakable — 2.6 points in twelve months with nobody measuring. Labor at 36.2% sits just outside the 30–36% band in §1.2, which does not prove it moved, only that it is now the first of the two halves you would rule out. Cash timing is invisible on this page and is very likely the thing that decides whether there is a year eight. Only undercapitalization is absent, and it is absent for an unhelpful reason: this restaurant survived long enough to prove it opened with enough money, seven years ago, against a cost structure that no longer exists.

What is genuinely missing from that page is not a line. It is a date. Every figure on it is annual, which means every problem on it was twelve months old before anyone saw it — and the two-point sensitivity in Figure 1.5 tells you what twelve months of an unwatched number is worth.

🔍 Check Your Understanding

1. A restaurant has COGS of $28,000 and labor of $31,000 on weekly sales of $92,000. What is its prime cost percentage, and how would you characterize it?
2. Two operators both report 29% food cost. One runs 38% labor, the other 30%. Which has the better business, and what additional information would you want before answering confidently?
3. Explain, in one sentence each, how undercapitalization and cash timing differ as failure mechanisms — given that both end with an empty bank account.

(1: $(28{,}000+31{,}000)\div 92{,}000 = 64.1\%$ — workable but tight, with essentially no cushion. 2: The 30%-labor operator is at 59% prime versus 67%, a large gap; but you would want to know occupancy cost, the service style, and whether the low-labor operation is buying convenience products that guests are noticing. 3: Undercapitalization is a stock problem — you never had enough money to begin with; cash timing is a flow problem — you have enough money over a year but not in a particular week.)

1.5 What the survivors have in common (and what they don't)

I want to be careful here, because this is where restaurant books usually start lying. There is no formula. There are restaurants doing everything right that closed because a two-year streetcar construction project took their parking, and restaurants doing everything wrong that are somehow still packed on a Wednesday.

But having watched a lot of both, some things are genuinely common to the survivors, and some famous things are not.

What they have in common

They measure weekly. Every single one. It is the most reliable predictor I know. The survivors can tell you their prime cost for last week. The casualties can tell you their food cost "runs about thirty."

Somebody in the building understands the numbers. Not necessarily the chef, not necessarily the owner. But somebody, and they have authority. A great many restaurants have a talented chef-owner and an outsourced bookkeeper and nobody in between who can look at a Tuesday and say "that shouldn't have cost that."

They have a cushion, and they respect it. Cash reserve, an undrawn line of credit, an owner who did not take a distribution in a good quarter. The survivors treat the reserve as untouchable.

They control the schedule. Not "keep labor low" — control. They forecast, they staff to the forecast, they adjust in real time, and they know before service what the labor number will be.

They keep people. Lower turnover shows up in food cost (trained cooks waste less and portion correctly), in labor (fewer training hours, faster stations), in service, and in the review score. Chapter 21 makes the case; it is one of the strongest in the book.

They are honest about the slow shifts. The survivors know which dayparts make money and which are vanity. They either fix the losing service, price it differently, or close it. The casualties keep a Monday lunch open for years because closing feels like defeat.

What they don't have in common

Cuisine. Every cuisine works. Every cuisine fails.

Price point. There are $12-check operations printing money and $180-tasting-menu rooms that never made a dollar, and vice versa.

Critical acclaim. This is worth dwelling on. Press is a demand-side event; it fills the room for a while. It does nothing at all to your cost structure. A restaurant with a bad prime cost and great press dies with a full dining room, which is a genuinely strange thing to watch — and it happens constantly. The restaurant in Figure 1.3 was written up twice.

The owner's culinary talent. Necessary, not sufficient, and occasionally an active liability when it produces a menu that is expensive to execute and impossible to staff.

🤝 Hospitality

The thing the numbers don't capture, and why it is still a number.

Everything above is financial, and it might read as though this book thinks a restaurant is a spreadsheet. It isn't, and the reason is commercial rather than sentimental.

A first visit is expensive. You spent marketing money, or press attention, or the guest's curiosity, to get one person through the door once. On the economics of Figure 1.2, a single visit contributes maybe thirty-six cents on the dollar toward covering your fixed costs. The second visit is where the business actually lives — it costs nothing to acquire and contributes the same thirty-six cents.

So the question "did they have a good time?" is not soft. It is the single largest driver of customer lifetime value in a business with no contracts and no switching costs. A guest chooses you again tonight or they don't.

This is why the third theme of this book is you sell hospitality, not plates. Not because hospitality is nice — because a restaurant that produces second visits has a fundamentally different revenue model from one that produces first visits, at identical food cost. Chapter 23 works the arithmetic properly.

1.6 Service styles and why the economics differ

"Restaurant" covers an enormous range of businesses that share almost nothing but a health permit. The techniques in this book apply broadly, but the targets do not, and applying a full-service benchmark to a quick-service operation will produce nonsense.

Here are the major service styles and how their economics differ.

(Ranges are industry rules of thumb — treat them as orientation, not as targets for your specific business. Build your own from Chapter 31.)

One thing to notice before moving on, because it is the first live example of a habit this book will ask of you constantly: the COGS range and the labor range are independent, and the prime target is not their sum. Take the top of both fine-dining ranges — 38% food and beverage cost, 42% labor — and you get 80% against a prime target of ≤65%. That is not an error in the table. It is what a range means: the two lines move against each other, and a fine-dining room running 38% COGS is a room that has bought product quality and must pay for it with labor discipline somewhere below 27%. A restaurant sitting at the top of both ranges at once is not a restaurant with high standards. It is a restaurant that is closing. Read every benchmark pair in this book that way — as a constraint on the pair, never as two independent permissions.

Quick service buys low labor with systems, equipment, and a narrow menu. The entire model depends on throughput: the same square footage serves many more people per hour, so fixed costs spread across far more transactions. It fails when volume drops, because the fixed labor floor is proportionally brutal.

Fast casual sits between: counter service and no table staff, but higher-quality product and a higher check. It emerged as a category because it captured full-service perceived quality on a quick-service labor model — which is, in prime-cost terms, exactly the arbitrage you would predict someone would find.

Full service is what most of this book assumes by default, because it is the most complex case and it contains every problem the others have. You pay a table staff, you own the guest for ninety minutes, and your revenue ceiling is set by seats times turns.

Fine dining is a genuinely difficult business and deserves to be said plainly. Higher food cost (better product, more trim, more waste on expensive ingredients), much higher labor (more staff per guest, more skilled staff, longer prep), and low turns. It works when the check average and the beverage attachment are high enough to carry it, and a lot of fine dining is quietly subsidized by the owner's uncompensated labor or by another business. That is not a reason not to do it. It is a reason to go in with the arithmetic done.

Bar-driven operations invert the usual relationship: beverage is the profit center and food supports it. Pour cost around 18–24% is dramatically better than food cost around 30%, so a beverage-weighted sales mix pulls blended COGS down hard. The tradeoffs are regulatory exposure (see below), a different labor pattern, and a revenue model far more sensitive to late-night trends and local competition. Chapters 15 and 16 are yours.

⚖️ Code and Compliance

Service style changes your regulatory load, not just your margins.

Choosing a model chooses a compliance burden:

• Alcohol service brings licensing (which in some jurisdictions is quota-limited and extraordinarily expensive to acquire), dram shop liability exposure, server training requirements, and hours-of-operation restrictions. Chapter 8 covers this.
• Raw or undercooked service — sushi, tartare, rare burgers, oysters — puts you in a higher health-department risk category, usually with more frequent inspections and sometimes with consumer-advisory and variance requirements. Chapter 25 covers this.
• Table service brings the tip credit and tip-pooling rules into your wage model, and those rules vary enormously by state — several states have no tip credit at all. Chapter 20 covers this.
• Mobile and off-site operations — trucks, catering — add commissary requirements, mobile permits, and transport-temperature obligations. Chapters 29 and 30 cover these.

All of it varies by state, county, and city, and all of it changes. Verify locally, with a professional, before you commit money to a model.

Independent versus chain

One more distinction worth naming. An independent is a single owner-operated restaurant or a very small group; a chain is a multi-unit brand operating under common systems, whether company-operated or franchised.

The difference that matters here is not size, it is where the systems live. A chain unit arrives with a costed menu, a staffing guide, a purchasing contract, an operations manual, and a district manager who will notice a prime-cost variance in week two. An independent has to build all of that, usually while also cooking.

This book is written primarily for independents, because independents have to do it themselves. If you are managing a chain unit, everything here still applies — but you should read it as a map of why the systems handed to you exist, which will make you far better at operating within them and considerably more promotable.

1.7 Who this book is for, and the promise it makes

The arithmetic you'll need immediately

Before the chapter closes, one calculation, because every chapter after this assumes you can do it.

A cover is one guest served. Not one table, not one check — one person. If a party of four dines, that is four covers. Covers are the fundamental unit of restaurant volume, and nearly every operating metric in this book is denominated in them.

The average check, also called per-person average or PPA, is sales divided by covers.

$$\text{Average check} = \frac{\text{Sales}}{\text{Covers}}$$

And revenue, at its simplest:

$$\text{Annual revenue} = \text{seats} \times \text{turns per day} \times \text{average check} \times \text{operating days}$$

🧮 Run the Numbers

The four-variable revenue estimate.

Take a 68-seat restaurant serving dinner five nights a week plus two brunches.

Dinner. 68 seats × 1.4 turns = 95 covers a night. At a $46 average check that is $4,370 per service. Five services: $21,850 a week.

Brunch. Say 110 covers each weekend day at a $24 average check = $2,640 per service. Two services: $5,280 a week.

Weekly total: $27,130.** Annualized over 52 weeks: **$1,410,760.

Now watch how sensitive that is. Move turns from 1.4 to 1.6 — one extra table turning over on a busy night — and dinner covers go from 95 to 109. That is $644 more per service, $3,220 a week, **$167,440 a year**, on a business whose entire annual profit at 6% would be around $85,000.

Or move the average check by two dollars, from $46 to $48, through better beverage attachment and one more appetizer per table. That is $190 a service, $950 a week, $49,400 a year.

This is the most important lesson in the arithmetic of this business. In a four-to-six-point margin operation, small movements in the operating variables produce enormous movements in profit — in both directions. It is why cost drift of three points is fatal, and it is why the operators who measure obsessively are not being neurotic. They are responding correctly to the leverage.

The six arguments this book makes

This book has an argument, and it runs in six parts. You will meet them in every chapter, and they are not decoration. Each one is a claim that could turn out to be false, and each one gets tested with actual numbers somewhere in the next thirty-nine chapters. It is worth stating them plainly now, along with where each gets settled, so you can hold the book to them.

The food is the easy part. This is not a claim that food does not matter; it is a claim about variance. Within any given market and price band, the difference between restaurants in food quality is fairly small — almost everyone who opens one can cook, and the ones who can't mostly don't. The difference between restaurants in cost control, scheduling discipline, and cash management is enormous. There is a crueler version of the same point: a chef-owner has spent ten years building systems for the food — mise en place, prep lists, station setups, a trained palate — and roughly zero years building systems for anything else. The part they are best at is the part with the least leverage. Settled in Chapters 11 and 13, which turn cooking into arithmetic, and in the money chapters, 31 through 33.

Prime cost is the number that keeps you open. §1.3 made the case; here is what makes it an argument rather than a preference. Prime cost is the only operating figure that is simultaneously (a) most of the cost you can actually change, (b) computable inside your own building without waiting for anybody, and (c) comparable against a published benchmark. Nothing else in a restaurant is all three at once. Sales are comparable but not controllable in the short run. Occupancy is comparable but fixed the day you signed. Food cost alone is controllable and computable and — as §1.3 showed — actively misleading on its own. Settled in Chapter 31, which computes it weekly, and Chapter 32, which shows what a single point of it does to the number of people you need through the door.

You sell hospitality, not plates. The 🤝 callout in §1.5 gave the commercial argument; here is the structural one. A restaurant has no contract, no subscription, no switching cost, and no lock-in of any kind. Every night is a re-election in which every guest votes again from scratch, and the incumbent's only advantage is the memory of last time. That makes the guest's memory the sole durable asset the business owns — it is the entire moat — and memories are produced by how people were treated far more reliably than by what they ate. There is a direct financial edge, too: hospitality is the only lever that raises revenue without raising cost. The arithmetic just above showed that a two-dollar move in average check is worth \$49,400 a year, and that move is a server who knows the wine list, not a purchase order. Settled in Chapter 23, with Chapters 21 and 22 supplying the people and the room.

Every seat-hour is inventory you can't store. Restaurants describe their inventory as the walk-in. It isn't. The walk-in keeps; a seat at 7:15 on a Saturday does not. What a restaurant actually manufactures is capacity — a fixed number of seats multiplied by a fixed number of hours — produced continuously whether or not anyone buys it, and expiring on the hour. That puts the business much closer to an airline or a hotel than to a shop, and it makes the questions that matter yield questions: how many seat-hours did you have, how many did you sell, and at what check. Settled in Chapters 22 and 24, which turn the idea into a metric and then into a reservation policy.

Cash is not profit. A profit-and-loss statement is a story about a period. A bank balance is a fact about a Thursday. Restaurants pull those two apart harder than most businesses do, for the structural reasons §1.4 listed — money that arrives daily and feels like liquidity, sales tax sitting in the account that was never yours, payroll that does not negotiate, and an irregular pile of annual bills that tends to land in the slowest weeks of the year. It produces one genuinely counterintuitive consequence that operators usually learn the hard way: the most dangerous month is often the one directly after your best month, because you bought inventory and staffed up for a volume that has just gone away. Settled in Chapter 33.

Your people are the product. At the industry's turnover rate you are not so much running a restaurant as running a permanent training program with a dining room attached. The cost of that is not the job posting. It is a green cook portioning by eye for six weeks, a green server who cannot sell the list or read a table, and the guest who came once and will not come again. Every one of those lands inside prime cost — which means turnover is not a human-resources topic that happens to be expensive. It is a prime-cost line item wearing different clothes. Settled in Chapters 17, 18, and 21.

Those six get repeated. They also get argued with: several later chapters take a theme and show you where it breaks down, because a principle you are never allowed to push against is not a principle. It is a slogan, and slogans do not survive contact with a Tuesday in February.

The promise

Here is what this book undertakes to do.

You will finish it able to cost a recipe to the cent, price a plate to a target and know when to ignore the target, engineer a menu into its four quadrants and act on each one, write a schedule against a forecast, read a profit-and-loss statement and find the leak in it, build a thirteen-week cash forecast, and state your break-even in covers per night. You will know what a lease obligates you to, what a health inspector looks at, what the tip credit is and whether it exists where you live, what a third-party delivery order actually contributes after commission, and what a lender means when they ask about your debt service coverage ratio.

You will also have built something. Across these forty chapters you will develop a complete business plan for a specific restaurant — concept, market, brand, lease, floor plan, costed menu, staffing model, three-year financials, marketing plan, operations manual — and at the end you will see what a lender did with it. If you are planning a real restaurant, Appendix C is the blank version. Work it in parallel; the plan is worth more than the reading.

What this book does not promise is that your restaurant will succeed. Roughly forty of a hundred survive three years, and no amount of technique overrides a bad location or a concept nobody wanted. What it promises is that you will not be one of the ones who never saw it coming. You will know your numbers, weekly, and you will know them early enough to do something.

🔍 Check Your Understanding

1. A 90-seat restaurant serves dinner six nights a week at 1.3 turns and a $38 average check. What is its estimated annual dinner revenue?
2. Why does a book about running restaurants introduce prime cost before it introduces the profit-and-loss statement?
3. Name the four failure mechanisms from §1.4 and identify which one is a stock problem and which are flow problems.

(1: 90 × 1.3 = 117 covers; × $38 = $4,446 per service; × 6 = $26,676 per week; × 52 = **$1,387,152. 2: Because prime cost is the operator's weekly decision variable, while the P&L is a periodic report; the book teaches the thing you act on before the thing you receive. 3: Undercapitalization, cost drift, labor, cash timing. Undercapitalization is a stock problem — insufficient money at a point in time. The other three are flow problems — money moving out faster than it should, or at the wrong moment.)

1.8 What actually closes restaurants: the named cause and the operative one

One more distinction before the plan, and it is the one that will make the next thirty-nine chapters legible to you.

Ask an operator why their restaurant closed and you will get a sentence. The sentence is usually true, usually sincere, and almost never the mechanism. That matters more than it sounds, because you cannot build a countermeasure against a sentence. You can only build one against a mechanism.

Here is the translation table. On the left, what gets named. On the right, what was actually running underneath it.

Every entry in the right column resolves into one of the four mechanisms in §1.4. The left column is vocabulary. The right column is physics. Work through the first four.

"We were undercapitalized" — or was it a cash-timing failure?

Both end with an empty account, and §1.4's Check Your Understanding already gave you the distinction: one is a stock problem, the other a flow problem. Here is why the distinction is worth money.

The test is a single question. Would more money at opening have changed the outcome? A genuinely undercapitalized restaurant is short in aggregate — the reserve was never built, the ramp took nine months instead of four, and no amount of scheduling discipline closes a gap of that shape. A cash-timing failure is short in a window: the twelve months produced enough, and the second week of February did not.

The countermeasures are opposite, which is precisely why the misdiagnosis is expensive. Undercapitalization is fixed before you open, by raising more or building less. Cash timing is fixed after you open, by forecasting — and it is frequently fixed for nothing, by moving an insurance renewal to a different month, negotiating terms with one vendor, or drawing on a line of credit you arranged in October when you did not need it.

An operator who diagnoses a timing failure as undercapitalization goes looking for more money. They often get it. And then they fail again, in the same week the following February, because nothing about the timing changed — the new money simply funded one more year of the same pattern. Chapter 5 sizes the raise; Chapter 33 builds the forecast that tells you which of the two you actually have.

"The location was wrong" — or was the lease priced for a room that couldn't produce?

Sites are rarely wrong in the abstract. Leases are frequently wrong in the arithmetic, and the two get confused because the location is right there in front of you and the lease is invisible.

Do the calculation the way it should have been done before anybody signed. Take a restaurant paying \$140,000 a year in all-in occupancy, aiming — per §1.2 — to hold occupancy under 8% of sales. That requires \$1,750,000 of annual sales (\$140,000 ÷ 0.08). Now measure what the room can physically produce: 60 seats, five services a week, a realistic 1.8 turns on a good night, a \$38 average check. That is 108 covers a night, \$4,104 a service, \$20,520 a week, \$1,067,040 a year — and that is the optimistic version, because it assumes every service is a good night. (Constructed figures.)

The site needed \$1.75 million. The room produces \$1.07 million. Occupancy lands at 13.1% (\$140,000 ÷ \$1,067,040), which is five points above target — and five points of sales is most or all of the operating profit §1.2 says a full-service restaurant gets to keep.

Nothing about that is a location problem. Marketing does not fix it, because it is not a demand problem. It is geometry: sixty seats will not produce \$1.75 million at a \$38 check no matter how many people want to come. The failure happened at the signing table, roughly three years before anybody noticed, and Chapters 6 and 7 exist to make you do this multiplication before you initial the page.

"Food cost got away from us" — or was it never held?

"Got away" implies it was once held. Usually it wasn't. It was estimated — invoices divided into sales — and an estimate is not a measurement.

The distinction is diagnostic, and it changes what you do on Monday. A restaurant that counts weekly and finds 33% has a food cost problem: it knows the number, it knows which week the number moved, and it can go find out whether the cause is purchasing, portioning, waste, or theft. That is a fortnight's work for a competent manager. A restaurant that does not count has a measurement problem, which becomes a food cost problem, which becomes a cash problem, in that order, over roughly fourteen months. The first is solvable. The second is what closed the restaurant in Figure 1.3.

So when you hear "food cost got away from us," the useful follow-up is not how high did it get? It is when did you find out? The gap between those two answers is the actual failure, and Chapters 11 and 13 exist to close it.

"The concept didn't work" — or did it work at a price nobody here would pay?

Concepts almost never fail because nobody wanted the food. They fail because the concept's cost structure required a check average the market would not support.

A menu that needs three cooks on the line, twenty-two minutes of ticket time, and four components finished to order is a \$52-check concept. It is a \$52-check concept whether or not you printed \$34 prices on it. If the neighborhood pays \$34, you have not built a restaurant — you have built a subsidy, and it is funded by whoever is not being paid, which in an independent is almost always the owner. That is how a place can be beloved, busy, well reviewed, and quietly insolvent for four years.

This is why "will people like it?" is the wrong test, and why it produces so many pleasant failures. The right test has three parts: will enough people pay the number this costs to produce, often enough, in this specific room? Chapter 2 turns that into a claim precise enough to be wrong, and Chapter 12 prices it.

⚠️ Where the Money Leaks

The post-mortem that teaches nothing.

When a restaurant near you closes, you will hear the reason within a week, usually from somebody who heard it from somebody. Collect a dozen of these and you will notice they are almost all demand-side: nobody came, the neighborhood changed, the landlord was impossible, the apps took our lunch business, people don't go out on Tuesdays anymore.

Demand-side explanations are comfortable because they are external, and because they are usually partly true. But nearly every closure has a demand-side story and a cost-side mechanism, and only one of the two was ever within reach of the person telling you the story.

Build this habit instead. For every closure you hear about, ask three questions: what was prime cost, what was occupancy as a percentage of sales, and how many weeks of payroll were in the account? You will almost never get an answer, and the silence is the answer. An operator who could have answered all three on any given Monday is an operator who saw it coming — and people who see it coming generally have twelve to eighteen months in which to act.

Apply exactly the same discipline to your good news, by the way. A strong quarter gets a sentence too — "the summer was great" — and it is precisely as uninformative. Up on covers or up on check? Up in the bar or up in the dining room? A win you cannot explain is a win you cannot repeat.

None of this is about catching anybody in a lie. Operators who give you the left column are being honest; they are telling you the last thing that happened rather than the first. But you are about to spend thirty-nine chapters building instruments, and an instrument is only useful if you know what it is measuring. The left column is what a restaurant looks like as it closes. The right column is what belongs on the dashboard.

1.9 How to read this book

The chain

Almost everything in the next thirty-nine chapters is a single chain, and each link is a number that becomes the input to the next one.

FIGURE 1.7 — The chain this book builds                                    [structure]

   MENU ITEM        an idea, until somebody costs it                    Ch. 10, 12
        │
        ▼
   COST CARD        every component priced, plus a waste allowance      Ch. 11
        │              → the only honest basis for a menu price
        ▼
   COGS             what the guests were actually served, counted:      Ch. 13, 31
        │              beginning + purchases − ending
        ▼
   PRIME COST       COGS + all labor, as a % of sales — weekly          Ch. 19, 31
        │              → read against the band for your service style
        ▼
   BREAK-EVEN       the sales level at which the fixed costs are        Ch. 32
        │              covered and the business stops losing money
        ▼
   COVERS/NIGHT     the same line, expressed as people through the      Ch. 32
                      door — a number you can stand in your own
                      dining room and count

   Read DOWN to build a plan. Read UP to diagnose one: when the covers number
   comes back impossible, the broken link is somewhere above it.

That chain is the book. It begins with something a cook thought of in a walk-in and ends with a number of human beings who have to come through your door on an average Tuesday, and every step between the two is arithmetic that either foots or doesn't.

Read down and you are building a plan. Read up and you are diagnosing one — which is what §1.8 was really about, and what §1.3's walk demonstrated in miniature: the number at the bottom is wrong, so you go back up a link, and then another, until you find the one that put it there. Everything in the book hangs off that spine. The lease and the build-out set the fixed costs. The hiring and the culture set what labor costs and what service is worth. The health code and the wage law set the floor under all of it. The marketing decides how many of those covers actually show up.

The four paths

Not everyone reading this is opening a restaurant, so four tracks run through the book. Every chapter opens with a short Learning Paths note telling you what your track should weight in that particular chapter.

• 🏗️ Opening. Front to back, and do the arithmetic by hand rather than reading it. Keep Appendix C open beside you and fill in your own figures as each chapter produces a section. The chapters you will be tempted to skim — leases, financing, compliance — are the ones that cost the most to get wrong, and they are the ones you cannot renegotiate later.
• 📋 Managing. You can move quickly through the opening and financing chapters on a first pass. The operations, labor, and money chapters are your job description, and §1.3 is your performance review whether or not anyone has ever told you so.
• 🍸 Beverage. Read the beverage chapters closely, then read every place in this book where a blended cost of goods appears — because a bar's contribution is usually invisible inside a blended number. §1.3's walk is the demonstration: a 22.0% pour cost was quietly pulling a 31.0% food cost down to a 29.0% COGS, and nobody looking at the blended figure would have known who did that.
• 🚚 Small Format. Read all of it, halve the numbers, and double the attention on cash. A truck or a stall runs the same chain with a much shorter warning period; there is far less distance between a bad month and a closed business.

The project, and one honest warning

Every chapter ends with a 🍽️ The Business Plan checkpoint that adds one section to a plan for a single constructed restaurant. Forty checkpoints, one finished document, built the way a real one is built — in pieces, out of order, by people who are learning as they go. Appendix C is the blank version for a restaurant of your own. Work the two in parallel, because a plan you build alongside the reading is worth several times a plan you merely read.

And here is the warning, which is a feature rather than an apology. The running example is going to be caught out, more than once, on purpose. Its plan is assembled the way real plans are assembled — by different hands, at different times, for different purposes — and later chapters are going to find figures in it that do not agree with each other, along with a few that are simply optimistic. When that happens it is not an error that slipped past somebody. It is the lesson. Chapter 31 in particular takes the accumulated plan apart in public and makes it foot, and that reconciliation is one of the skills this book fully intends you to leave with.

A book that presented a fictional restaurant as flawless would be teaching you the one thing that has never once been true of a real one. Your plan will not foot the first time either. The difference between operators is not whether the plan is wrong. It is whether somebody checks.

🍽️ The Business Plan

Checkpoint 1 of 40 — the file opens.

Throughout this book you will build a complete business plan for one restaurant. It is a constructed teaching example — it does not exist — but every number attached to it is realistic and internally consistent, and by Chapter 40 the finished document is one you could put in front of a lender.

Here is what we know so far, which is almost nothing:

Bellwether (constructed teaching example)

A 68-seat chef-driven neighborhood American restaurant — 56 in the dining room, 12 at the bar, plus a 16-seat patio in season — in the Rivermill District, a former warehouse neighborhood of a mid-size Midwestern city about eight years into gentrification. Wood-fired hearth, short seasonal menu, dinner Tuesday through Saturday plus weekend brunch, full bar with a tight cocktail list and a roughly 40-bottle wine list.

Two partners: a chef with fourteen years of experience and no ownership experience, and a front-of-house partner who has run dining rooms but never a P&L.

That is the concept in one paragraph. It is not yet a business.

What this chapter adds to the plan: the target it has to beat.

Before a single further decision, write down the arithmetic the plan is obligated to satisfy. Using the estimate from §1.7 and the benchmarks from §1.3 and §1.6, a 68-seat full-service restaurant of this type must, to be viable:

What this checkpoint does not settle. Everything. We have no market analysis, no site, no lease, no menu, no cost cards, no staffing model, and no idea what it costs to build. The revenue figure above is an order-of-magnitude estimate from four assumed variables, three of which — turns, check, and operating days — are assumptions the next thirty-nine chapters exist to test.

Open questions carried forward:

1. Is there a market in the Rivermill District for a $46 dinner check? (Chapter 2)
2. Can this concept produce 95 covers a night on 68 seats? (Chapters 7, 22, 24)
3. What does it cost to build? (Chapters 6, 7)
4. Where does that money come from? (Chapter 5)
5. Can a chef with no ownership experience and a manager who has never read a P&L run a 60% prime cost? (Chapters 11, 19, 31 — and honestly, Chapter 21)

Conclusion

The restaurant business is not mysterious. It is a low-margin, high-volume, perishable-inventory business with a large hourly workforce and a fixed cost base that does not care how Tuesday went. Everything that makes it hard follows from that sentence.

Ninety percent of restaurants do not fail in the first year. About a quarter do, and about six in ten are gone within three — which means most of the casualties are businesses that worked, for a while, and then bled. They bled through the two cost lines that make up prime cost, which is why prime cost is the number this book returns to in every chapter, and why the operators who compute it weekly are so consistently the ones still open.

Four mechanisms do the actual killing: opening without enough money, letting costs drift because nothing is measured, losing control of the schedule, and running out of cash in a week when the obligations stack up. Each one is visible in advance. Each one has a countermeasure. Almost every technique in the next thirty-nine chapters is a countermeasure to one of them.

Chapter 2 starts building. Before you can cost a menu or write a schedule, you need to know what restaurant you are running, for whom, and why they would choose you over the eleven other places within a ten-minute walk. A concept is not a cuisine and it is not a feeling — it is a testable claim about a market, and the next chapter is about making that claim precise enough to be wrong.

Key Terms

Prime cost — the sum of cost of goods sold (food and beverage) and total labor cost, expressed as a percentage of total sales. The most important single operating number in a restaurant; the target for full service is at or below 60%. (Ch. 1)

Cover — one guest served. A party of four is four covers. The fundamental unit of restaurant volume. (Ch. 1)

Average check (also per-person average, PPA) — total sales divided by covers; the average amount spent per guest. (Ch. 1)

Full service — a restaurant where orders are taken at the table by service staff. Higher labor cost, higher check average, lower turns than counter models. (Ch. 1)

Fast casual — counter-service ordering with higher-quality product and a higher check than quick service; a hybrid that captures full-service perceived quality on a quick-service labor model. (Ch. 1)

Quick service — counter or drive-through ordering, narrow menu, high throughput, low labor percentage; the model depends on volume to spread fixed costs. (Ch. 1)

Independent — a single owner-operated restaurant or a very small group, which must build its own operating systems. (Ch. 1)

Chain — a multi-unit brand operating under common systems, whether company-operated or franchised, where systems are supplied to the unit rather than built by it. (Ch. 1)

Undercapitalization — opening with less money than the business needs to reach sustainable operation; the most common cause of first-year failure, and a planning error rather than bad luck. (Ch. 1)

Restaurant failure rate — the proportion of restaurants that close or change ownership within a given period. Roughly one in four in year one and close to six in ten by year three, according to the published research — not the 90% of industry folklore. (Ch. 1)

Spaced Review

1. Without looking back: what two cost categories make up prime cost, and what is the rule-of-thumb benchmark for a full-service restaurant?
2. A restaurant's owner tells you their food cost is 27% and they are very pleased. What is the first follow-up question you should ask, and what is the second?
3. Explain why the losses in Figure 1.1 cluster in years two and three rather than year one, and what that pattern implies about the cause of most restaurant failures.
4. A 60-seat restaurant serves dinner five nights a week at 1.5 turns with a $52 average check. Estimate its annual dinner revenue. If turns fell to 1.3, how much annual revenue would be lost?
5. The recurring question: an operator decides to switch from butchering whole chickens in house to buying pre-portioned chicken at a higher per-pound cost. Does this decision move prime cost, and in which direction? What would you need to measure to find out?