Chapter 35: Growth Decisions: A Second Location, Franchising, Catering, Retail Products, and When to Expand

"The second restaurant is not a bigger version of the first one. It is a different company that happens to own the first one." — constructed; the sentence I wish somebody had said to me before I signed

Overview

Here is how it happens. You have been open two years. The room is full on Friday, the reviews are good, the staff have stopped quitting every eight weeks, and for the first time since you signed the lease you can look at a bank balance without your stomach dropping. Somebody — a broker, a landlord, a friend, a guest who works in real estate — mentions a space. It is a good space. It is cheap, relatively, and it will not be available in ninety days.

And the arithmetic seems obvious. This restaurant makes money. A second one would make money too. Two restaurants, twice the money.

That is the sentence that closes more successful independent restaurants than any bad review ever written, and it is wrong in a specific, computable way. It is wrong because the first restaurant is not a machine that produces profit. It is a machine that produces profit while you are standing in it, and the thing you are proposing to duplicate is the building, the menu, and the equipment — not the part that actually works.

So this chapter opens with a question, and I want you to sit with it before you read another paragraph.

Is your business profitable without you in it?

Not "could it survive a long weekend." Not "the sous is very good." Profitable — measured, with a market-rate salary sitting in the labor line where your uncompensated sixty-hour week currently sits, over a period long enough to include a slow February and a health inspection and somebody's resignation. If the honest answer is no, then what you own is not a repeatable model. It is a job with excellent equipment, and duplicating it will produce two jobs and one person.

For our running project the honest answer is no, not yet — and this chapter proves it with numbers rather than asserting it with adjectives. Then it does the more useful thing, which is to show you every other way a restaurant grows, ranked by what each one costs in capital, in personal guaranty, in owner-hours, and in reversibility. Some of them are very good. One of them is a second restaurant, and it is nearly last.

In this chapter, you will learn to:

• Run the seven-gate second-location test on a real set of operating numbers and state whether the business passes, with the evidence for each gate.
• Recompute a unit's profit with market-rate management in it instead of the owner, and explain why that number — not the reported one — is what actually repeats.
• Build a full second-location pro forma including the overhead layer, the new debt service, and the profit the second unit takes out of the first one.
• Measure trade-area overlap from your own reservation data and convert it into covers per night against your cash break-even.
• Compare catering, a retail product, licensing, a delivery brand, a small format, and a second restaurant on the only four axes that matter: capital, guaranty, hours, and reversibility.
• Make the affirmative financial case for one great restaurant held for thirty years — and name honestly what that choice costs you.

Learning Paths

🏗️ Opening — you are not here yet, and that is exactly why you should read §35.1 and §35.5 now. The bench you would need before a second building is the same bench that makes the first one survivable, and it takes two years to build. Start it before you open, not after. 📋 Managing — §35.5 is your chapter. The path from manager to multi-unit leader runs through being the person an owner can leave a building with, and §35.5 tells you exactly what that looks like from the owner's side of the desk. 🍸 Beverage — §35.7 is yours twice over: a bottled cocktail or a house amaro is the most commonly attempted retail extension in this industry, and the jar arithmetic in Figure 35.9 applies to it exactly. Note also that a strong bar is the cheapest way to raise the check on the soft nights in §35.8. 🚚 Small Format — §35.3 and §35.7 are where your model wins. Chapter 30's comparison is the spine of this chapter's argument, and the reason is structural: the small format's advantage is not lower cost, it is a bounded downside.

35.1 The honest readiness test: is the business profitable without you in it?

Every growth conversation in this industry starts in the wrong place. It starts with the site, or the money, or the concept for the second one. It should start with a measurement of the first one, and the measurement is not "is it profitable."

It is: is it profitable without you in it?

Why that is the only question that matters first

A restaurant's reported operating profit is a blended figure. It contains the value of everything the owner does — the purchasing relationships, the schedule they write on Sunday night, the two hours a week they spend on the walk-in that nobody else spends, the table they touch that turns a complaint into a regular, the cook they can call at 3:40 on a Friday because that cook owes them personally.

None of that is in the operations manual, because there isn't one. All of it is in the profit.

When you open a second location, you do not duplicate that. You divide it. The owner who was producing that value in one building now produces it in two buildings at roughly half the intensity, in the best case, and in practice worse than half, because the second building is new and consumes attention at a rate the first one stopped consuming years ago. Meanwhile the first building — the one that is paying for everything — has just lost its most productive employee.

This is why the failure pattern is so consistent, and it is worth naming plainly rather than dressing up with statistics I would have to invent. The documented pattern in this industry, at every scale from two units to two hundred, is expansion past the management capacity that supports it. It is why chains that grew faster than they could train general managers ended up closing units they had opened three years earlier. I am not going to attach a percentage to that, because I do not have one I can defend. I will tell you what the mechanism is, which is more useful: growth converts an owner from an operator into an administrator, and the buildings that were being operated stop being operated.

👨‍🍳 On the Line

The two-building Friday.

You open the second one. It is a Friday in November. Unit two is eleven weeks old and still finding its feet; unit one is doing 118 covers with a 40-top at 6:30.

At 3:40 the grill cook at unit two calls out. You are at unit one. The new chef de cuisine at unit two has been in the job nine weeks and has never run that station short. You have exactly three options and all of them cost money:

1. Drive over. Twenty-two minutes each way. You cook the station at unit two, which you are good at, and unit one runs a 118-cover Friday with a 40-top and no owner in the building. Historically that is the night your FOH partner and you are both on the floor. Tonight nobody is.
2. Stay. Unit two runs short. The tickets go long, the new chef de cuisine gets buried, and eleven weeks into a new restaurant's life you serve a bad Friday to a room that is still deciding about you. Those guests do not come back, and you will never see the line item.
3. Call somebody in at overtime. Which is the right answer, and requires that there be somebody to call — a person on unit two's roster who can run that station cold. In week eleven of a new restaurant, there usually is not.

Here is what I want you to notice. The problem is not that the cook called out. Cooks call out. Chapter 21's Friday night is the same problem at one building, and at one building you solve it, because you are there and you have depth built over two years.

The second building did not create a new kind of problem. It created the same problem in a place where you are not standing, in a room that has no depth, on a night when the profitable restaurant also needed you. Every operational problem you already know how to solve becomes a problem you have to solve remotely, with a team that has not earned the reflexes yet.

That is what "the owner does not scale" means. It is not a metaphor.

The seven gates

I want to give you an instrument rather than an attitude. Here is the second-location test: seven questions, each of which has an answer that is a fact rather than a feeling, and every one of which must be a yes before you sign anything.

FIGURE 35.1 — The second-location test                      [constructed teaching instrument]

  #  THE GATE                                                    BELLWETHER, TODAY
  ─────────────────────────────────────────────────────────────────────────────────
  1  THE ABSENCE TEST                                                    ✗
     Can both owners be out of the building for fourteen
     consecutive days — the same two weeks, no phone calls —
     with prime cost, covers, and guest scores holding?
     Evidence: Ch. 21 found no management bench at all.
     The chef-owner works six days; the FOH partner runs
     the room. Neither can be absent a week without the
     operation degrading.

  2  THE REPLACEMENT TEST                                                ?
     Is the unit profitable with market-rate management
     in it instead of the owners?
     Evidence: never computed. Depending on what the
     partners actually draw, the answer is somewhere
     between 15.0% and 5.5% operating profit. See §35.2.

  3  THE REPEATABILITY TEST                                              ✗
     Do you know WHY it works, precisely enough to write
     it down and hand it to a stranger?
     Evidence: no operations manual exists. Ch. 37 sets
     the standard; Bellwether has not started.

  4  THE BENCH TEST                                                      ✗
     Are the people who will run both buildings already
     employed and already performing — not people you
     would have to go hire?
     Evidence: four salaried, two of whom are the owners.
     Two buildings need five leaders. See §35.5.

  5  THE SLACK TEST                                                      ✗
     Is there room in the salaried week to absorb a
     second building?
     Evidence: Ch. 19 found the salaried week fully
     committed. Ch. 29's catering coordination already
     takes 3.5 hrs × 14 events = 49 hours a year out
     of it.

  6  THE CASH TEST                                                    marginal
     Can the existing unit fund the second unit's first-
     year hole out of its own cash, without touching the
     working-capital reserve or the partners' income?
     Evidence: the modeled hole is $163,996 (§35.3)
     against cash after debt service of $191,520 — which
     is also what the partners live on.

  7  THE AUDIT TEST                                                      ✗
     Who audits the owners?
     Evidence: Ch. 34 closed on this question and left
     it open. With two buildings it stops being
     theoretical: cash, comps, voids, and inventory at
     unit two will be handled by people neither owner
     is standing next to.
  ─────────────────────────────────────────────────────────────────────────────────
  SCORE: this is a GATE, not a scorecard. Five of seven is not a pass.
         Bellwether clears none of them cleanly.

It is a gate and not a scorecard, and I want to defend that. Scorecards are for decisions where the downside is proportional to the error. This decision is not. A restaurant with an 18-cover cushion — Chapter 32's finding, and we will use it hard in §35.4 — has no room to be partly wrong. Each of those seven failures is independently capable of consuming the entire cushion. Failing four of them simultaneously is not four small problems; it is one large one wearing four hats.

Gate 6 deserves a note, because "marginal" is doing a lot of work there. On paper, \$191,520 of annual cash after debt service exceeds a \$163,996 first-year hole. In practice that \$191,520 is not sitting in a drawer. It is the partners' household income, the source of every unbudgeted equipment failure, and the reason the business survived its first slow February. Calling it "available" is the same error as calling the sales-tax collection in your account a healthy balance. Chapter 33 was blunt about that and it applies here.

Owner dependence, and how to measure it

The concept underneath all seven gates has a name.

Owner dependence is the share of a business's results produced by the owner's personal presence, judgment, and relationships rather than by systems that would operate without them. It is not a character flaw. In a first restaurant it is correct — you are supposed to be the system for the first two years, because building the system takes longer than opening the doors and the doors have to open. The error is not being owner-dependent. The error is not knowing how owner-dependent you are, and then trying to copy the business.

Here is a measurement you can run this month. It costs nothing and it will change how you think.

The absence audit. For eight weeks, log every decision that reached an owner. Not the ones you initiated — the ones that came to you. Then sort them into three bins:

Owner dependence is the share that lands in A plus B. In my experience a typical owner-operated independent runs somewhere north of seventy percent — most days, most of what reaches the owner is something a manual and a trained manager should have handled. A unit that is genuinely ready to be duplicated has pushed that under about fifteen percent, which means the owner's week has become mostly bin C. That is a practitioner's judgment, not a research finding, and I offer it as a target to aim at rather than a benchmark to be graded against.

The audit's real value is not the percentage. It is the list. Every bin-B item is a page of the operations manual you have not written, and eight weeks of logging produces a table of contents that a year of good intentions never did. Chapter 37 turns that list into the document.

🤝 Hospitality

What the second restaurant costs the first one's guests.

There is a version of this argument that is purely financial, and I have been making it. Here is the version that is not.

A neighborhood restaurant's regulars are not buying dinner. They are buying the fact that the chef-owner comes out at 8:40 and asks how the lamb was, and remembers that they hated the beets in March. That is the entire product. Chapter 23 put a number on it: the second visit is where the business lives, because the first one barely covers what it cost to acquire.

When the owner is at the other building three nights a week, the regulars notice in about six weeks. Nobody says anything. Nobody writes a review that says "the owner wasn't there." What happens is that the visit interval stretches — from every three weeks to every five — and the cover count on Tuesday softens by four, and nothing on any report tells you why.

Four covers a Tuesday is 208 covers a year, and at a \$46 check with a 57% contribution margin that is \$5,455 of profit that disappeared without appearing anywhere as a line. It is not the biggest number in this chapter. It is the one you will never see coming, and it compounds, because the guests who stretch to five weeks are the same ones who were going to bring their in-laws in December.

The commercial point and the human point are the same point here, which does not happen often enough to waste: the reason the second restaurant hurts the first one is that hospitality is a personal transaction, and you cannot be in two rooms.

35.2 Unit economics: proving the model repeats before you repeat it

Unit economics is the revenue, cost, and capital of a single operating location, measured independently of the company that owns it, so you can tell whether the model itself makes money and how much capital it takes to buy that money.

The phrase gets used loosely. Used properly it answers four questions, and the fourth one is the one nobody asks:

1. What does one unit cost to build? — the project cost, all in, including the working-capital reserve and the money you will spend before you sell anything.
2. What does one unit earn at maturity — with a fully-paid manager in it instead of you? — the owner-adjusted unit profit.
3. How long does it take to get there? — the ramp, in months, and what it costs to fund the months before maturity.
4. What is the return on the capital and on the guaranty? — because those are two different denominators and the second one is the one that can take your house.

Bellwether has clean answers to (1) and (3): \$620,000 and, on the plan, year one. Question (2) is where the trouble is.

Owner-adjusted unit profit

Bellwether's plan year produces \$1,550,000** of revenue and **\$261,020 of operating profit — 16.8% — before \$69,500** of debt service, leaving **\$191,520 of cash. Those are good numbers. They are better than most independents ever see.

They also contain two owner-partners who are, in every sense that matters to a second location, working for whatever is left over.

The question unit economics asks is: what would it cost to buy that labor at market? Two buildings need paid leadership in both. So price it.

Now the honest part. I do not know what the partners currently draw, and neither, in my experience, do most operators in their second year — they take what is left after the vendors, which is not a salary, it is a residual. So run it as a fork.

🧮 Run the Numbers

Owner-adjusted unit profit: three versions of the same restaurant.

Reported operating profit: \$261,020** on \$1,550,000 (16.8%). Market-rate replacement management costs \$175,680 loaded. The adjustment is the difference between what the partners currently draw and what replacing them costs.

If the partners' combined draw is…	Loaded cost of that draw	Replacement delta	Owner-adjusted profit	As % of sales
\$144,000 (already at market) | \$175,680	\$0 | **\$261,020**	16.8%
\$120,000 | \$146,400	\$29,280 | **\$231,740**	15.0%
\$90,000 | \$109,800	\$65,880 | **\$195,140**	12.6%
nothing above distributions	\$0 | \$175,680	\$85,340	5.5%

(Check the middle row: \$120,000 × 1.22 = \$146,400. \$175,680 − \$146,400 = \$29,280. \$261,020 − \$29,280 = \$231,740, which is 15.0% of \$1,550,000.)

Read the last row, and read it slowly. A restaurant that reports a 16.8% operating margin and whose owners are paying themselves out of what is left over is a restaurant with a 5.5% repeatable margin. That is not a bad business — 5.5% is a perfectly ordinary independent restaurant, and Chapter 1 told you the range. But it is not a business that supports being copied, because copying it means buying the management you are currently getting for the residual, in both buildings.

The number you are proposing to duplicate is the number in the last column, not the number on your P&L.

This single adjustment is the most common reason a second location surprises its owner. The first unit's margin looked like 17%. The model's actual margin was 12%, or 8%, or 5%. The second unit — which has to pay market for everything, because the owners are already spent — runs the model's margin, not the first unit's.

The four tests a model must pass to be repeatable

Owner-adjusted profit is necessary and not sufficient. A model that repeats has four properties, and Bellwether has a mixed record on them:

Site-independence deserves its own paragraph, because it is where most concept-driven restaurants quietly fail the test. Chapter 2 established that Bellwether is share-taking in a supplied market — the Rivermill District was not short of restaurants; Bellwether won covers from operators who already had them. That is a perfectly good way to run a restaurant. It is a terrible basis for assuming the model repeats, because "we win share in a gentrifying warehouse district eight years into its arc, with a hearth, at a \$46 check" is a claim about one neighborhood at one moment. It is not a claim about a format.

The test for site-independence is uncomfortable and simple: can you state, in one falsifiable sentence, the customer and the occasion the concept wins, without naming the neighborhood? If the sentence needs the neighborhood, you have a restaurant, not a model. That is fine. It is just a different plan.

Chapter 30 gave us the cheapest possible version of this test and we should not walk past it. A residency — ten Monday nights cooking your menu in someone else's kitchen in a different part of town — tests demand for the concept at a price point, in a trade area you do not operate in, for \$430 of downside, which is 0.07% of the project cost. There is no other test in this book with that ratio of information to money. If the question is "does this work outside Rivermill," ten Mondays answers it better than a pro forma ever will, and the failure costs less than one bad Saturday.

35.3 The growth options, ranked by capital and risk

Now the useful part. "Should we grow?" is almost always the wrong question, because growth is not one thing. It is at least a dozen things with wildly different price tags, and operators tend to consider exactly two of them: stay as you are, or build another restaurant. That is a false binary and it is expensive.

Rank them properly and the ranking is not close.

FIGURE 35.2 — The growth ladder, ordered by new personal guaranty     [constructed]

  NEW GUARANTY      OPTION                                        HOW FAST YOU CAN UNDO IT
  ─────────────────────────────────────────────────────────────────────────────────────
  none          │   price and menu work on the existing unit      the same day
  none          │   fill Tuesday and Wednesday                    the same week
  none          │   a residency in someone else's kitchen         the same night
  none          │   a delivery-only second brand                  30 days' notice
  none          │   catering out of the kitchen you already have  one season
  none          │   a retail SKU through a co-packer              one production run
  none          │   licensing your name to another operator       the contract term
  modest,       │   a food truck                                  sell the truck
   asset-backed │
  ══════════════╪══════════════════════════════════════════════════════════════════════
  ~$545,000     │   a small-format second unit                    hard: a lease + a note
  ~$1,471,000   │   a full second restaurant                      very hard: 10 years
  ~$150,000+    │   buying a franchise (Ch. 36)                   very hard: a franchise
                │                                                  agreement is a term
  varies, large │   franchising your own concept (Ch. 36)         a different company
  ─────────────────────────────────────────────────────────────────────────────────────
  The double line is the only line on this page that matters. Everything above it can
  be undone with a phone call, a notice period, or a sale. Everything below it is a
  signature that outlives your opinion of it.

The four axes

Every option gets scored on four things, and only four:

1. Capital — what it costs to start, including the money you will spend before revenue.
2. New personal guaranty — how much of your personal balance sheet you have pledged. This is the axis operators skip, and it is the one that determines whether a failure is a bad year or a bankruptcy.
3. Owner-hours per week — against a salaried week that Chapter 19 already found full.
4. Reversibility — how fast, and at what cost, you can stop.

Here is the same ladder with the numbers attached. Every dollar figure below is either inherited from an earlier chapter or built in this one; the ones built here are worked in §35.7 and §35.8.

Read the last row against the row above it and then against the four rows above that. The option with the largest capital requirement, the largest guaranty, the largest time cost, and the worst reversibility also has the worst first-year return. That is not a coincidence and it is not bad luck. It is what happens when you add a fixed cost base and a management layer to a business that had neither.

Building the second-location pro forma properly

Most second-location pro formas are wrong in the same three ways, and all three flatter the decision:

• They model the new unit and stop. They do not model what the new unit does to the old one.
• They forget the overhead layer. Two units need bookkeeping, insurance, technology, and coordination that one unit got for free because the owner did it at the kitchen table.
• They compare the new unit to zero. The correct comparison is not "unit two versus nothing." It is "the group with two units versus the group with one" — because the alternative to building unit two is not going out of business, it is continuing to run a restaurant that clears \$191,520 a year.

Here is the pro forma done properly.

🧾 Read the Numbers

```text FIGURE 35.3 — "The second Bellwether" [the Bellwether plan, extended] THE ARTIFACT A first-year pro forma for a second location, presented three ways: the new unit alone, the group with two units, and the group with one — which is the comparison the decision actually turns on. THE CONTEXT A 64-seat second unit in a comparable neighborhood, 2,600 sq ft, same format, same check averages, opened by the same two partners. Unit one is held at its year-one plan figures so the comparison isolates the effect of the second building.

PROJECT COST — UNIT TWO Construction $355,000 Equipment, including a second hearth 195,000 Smallwares and FF&E 48,000 Pre-opening: labor, training, licensing, initial inventory 37,000 Working-capital reserve 45,000 ────────────────────────────────────────────────────────────────── TOTAL PROJECT COST $680,000

CAPITAL STACK — UNIT TWO Owner injection, from accumulated cash $120,000 Landlord tenant-improvement allowance 70,000 Equipment lease 65,000 Term debt 425,000 ────────────────────────────────────────────────────────────────── $680,000

UNIT TWO — YEAR ONE Covers: 64 seats x 1.3 turns = 83 dinner covers, five nights 100 brunch covers, two services Revenue: (83 x $46 x 5) + (100 x $24 x 2) = $23,890/week x 52 $1,242,280 rounded for the pro forma to $1,240,000 100.0% Cost of goods sold 365,800 29.5% Labor, all-in (a full leadership team plus hourly) 440,200 35.5% ──────────────────────────────────────────────────────────────── PRIME COST 806,000 65.0% Occupancy (2,600 sq ft at $30 base + $7 NNN) 96,200 7.8% Other operating 173,600 14.0% General & administrative 37,200 3.0% ──────────────────────────────────────────────────────────────── OPERATING PROFIT, UNIT TWO $127,000 10.2%

WHAT UNIT TWO DOES TO UNIT ONE AND TO THE GROUP Unit one operating profit, on plan $261,020 less contribution lost to cannibalization (§35.4) (38,176) less a general manager for unit one, loaded (80,520) ──────────────────────────────────────────────────────────────── UNIT ONE, ADJUSTED $142,324 Unit two operating profit 127,000 less above-unit overhead, year one (Fig. 35.5) (87,000) ──────────────────────────────────────────────────────────────── GROUP OPERATING PROFIT $182,324 on group revenue of $1,483,024 + $1,240,000 = $2,723,024 6.7% Debt service: unit one $69,500 + unit two $85,300 (154,800) ──────────────────────────────────────────────────────────────── GROUP CASH AFTER DEBT SERVICE $27,524

THE COMPARISON THE DECISION TURNS ON One unit, today: cash after debt service $191,520 Two units, year 1: cash after debt service 27,524 ──────────────────────────────────────────────────────────────── THE SECOND RESTAURANT COSTS THE PARTNERS $163,996 in its first year — on 76% more revenue and roughly double the personal exposure.

WHAT IT SHOWS A second unit that performs well — 10.2% operating profit in year one is a good first year for a new restaurant — still reduces the owners' cash by $163,996, because it adds $80,520 of management to unit one, $87,000 of overhead to the group, $85,300 of debt service, and takes $38,176 of contribution out of the building that was paying for everything. Those four numbers total $290,996 against unit two's $127,000 of profit. WHAT IT DOESN'T It does not model a bad first year, a construction delay, or a ramp slower than plan — each of which is more likely than not. It does not show the month-by-month cash trough, which is deeper than the annual figure implies because the hole is front-loaded. And it does not price the owners' time, which is the scarcest input in the whole document. THE DECISION Do not sign. Compute the same table with a small format and with the line extensions in §35.7 before considering a full second unit again, and do not reconsider until the seven gates in Figure 35.1 are green. THE LESSON The correct comparison is never "the new unit versus nothing." It is "the group with it versus the group without it" — and the second unit has to pay for the damage it does to the first one before it earns its first dollar. ```

Two things about that pro forma deserve emphasis, because they are where the argument actually lives.

First: unit two is not modeled as a failure. It does \$1,240,000 in its first year, 80% of unit one's plan, which is a good ramp. It runs a 65% prime cost, which is high but entirely normal for a first year with a new team. It clears 10.2%. If you showed that P&L to an operator with no context they would call it a solid opening year. The problem is not the second restaurant. The problem is what a second restaurant costs a two-unit company.

Second: the deltas do not depend on when you open it. I held unit one at its year-one plan for clarity, but the four drag items — the unit-one GM, the above-unit overhead, unit two's debt service, and the lost contribution — are all independent of what unit one grew to. Run the same table against unit one's year-three plan of \$1,850,000 and the arithmetic changes by nothing: unit two still has to clear \$290,996 before the partners see a dollar. That is 23.5% of unit two's sales. Bellwether's best-ever modeled margin is 16.8%.

The stabilized case, because year one is not the whole argument

It would be dishonest to rest the case on a first year. New restaurants lose money in year one; that is what year one is. So run it out to unit two's third year, when it should be mature.

At stabilization, in year three, with both restaurants running well, the partners have less cash than they have today — on 89% more revenue and 108% more personal exposure.

That is the finding. It is not a rhetorical flourish; it is a subtraction. And it lets us state the hurdle exactly:

For the group to merely match the \$191,520 the single unit already produces, unit two must clear **\$262,620 of operating profit on \$1,420,000 of sales. That is an 18.5% operating margin** — better than Bellwether's own plan has ever called for, in a building with none of the first one's history, run by people who do not work there yet.

(Check: \$157,700 + U − \$74,000 − \$154,800 = \$191,520, so U = \$262,620; and \$262,620 ÷ \$1,420,000 = 18.5%.)

Why two units is the worst number of units

Here is the structural reason, and it is the most useful idea in this chapter.

A one-unit business has no above-unit overhead, because the owner is the above-unit overhead and does it unpaid on Sunday. A four-unit business has real above-unit overhead and four units to spread it across. A two-unit business buys the entire infrastructure of a company and gets two units to amortize it over.

(Illustrative. The step at three units is real: two units can be supervised by two owners, three cannot, and the person you hire to supervise three costs the same whether you have three or five.)

Notice the shape. Overhead per dollar of sales rises from one unit to three and then starts falling. You have to cross a valley to get to the far side, and the valley is two units and three units. Operators who reach five or ten units are on the far side and their economics are genuinely better. Operators who stop at two are standing in the deepest part of the valley, paying company overhead on restaurant volume.

This is why "we'll just do one more and see how it goes" is such an expensive sentence. There is no see how it goes at two units. Two units is not a smaller version of a group; it is the most structurally disadvantaged size a restaurant company can be. You either commit to crossing the valley — which means the bench, the manual, the capital, and probably five or six units — or you stay on this side of it.

⚠️ Where the Money Leaks

The overhead layer nobody budgets.

When you open a second location, a set of costs appears that were previously free, because the owner absorbed them. Here is the year-one list for a two-unit group, and it is the single most commonly omitted page of a second-location pro forma.

```text FIGURE 35.5 — Above-unit overhead, year one, two units [constructed teaching example]

Bookkeeper, moving from part-time to full-time (incremental) $26,400 Group accounting, tax, legal, and payroll service (incremental) 14,800 Second-entity insurance, umbrella, and employment-practices 9,600 Technology: a second POS, reservations, scheduling, inventory, and accounting seats 11,700 Vehicle, mileage, and inter-unit product transfers 6,500 Recruiting and training: four salaried leaders and 26 hourly hires at unit two 18,000 ────────────────────────────────────────────────────────────────────────── TOTAL, YEAR ONE $87,000 Ongoing, after the one-time recruiting and training $69,000 ```

Not one dollar of that appears in either restaurant's P&L, and it is 3.2 cents of every dollar the group takes in. On a business whose whole operating margin is 16.8 cents, you have just handed away nearly a fifth of it to run the company that owns the restaurants.

And note what is not on the list: a director of operations. At two units, the partners are the above-unit management, working unpaid on top of two jobs. Add \$92,000 for that role — which you must, at three units — and the overhead line becomes \$161,000. That is the valley.

FIGURE 35.4 — Where the group's dollar goes, two units, year one     [constructed]

  Combined unit-level operating profit   ████████████████        9.9¢
    unit one $142,324 + unit two $127,000 = $269,324
    on group revenue of $2,723,024

  − above-unit overhead                  █████                   3.2¢
  ────────────────────────────────────────────────────────────────────
  = GROUP OPERATING PROFIT               ███████████             6.7¢

  − debt service, both units             █████████               5.7¢
  ────────────────────────────────────────────────────────────────────
  = CASH TO THE PARTNERS                 ██                      1.0¢

  For comparison, one unit today:
    operating profit                     ███████████████████████████ 16.8¢
    debt service                         ███████                      4.5¢
    cash to the partners                 ████████████████████        12.4¢

  Columns may not foot to the tenth of a cent because of rounding.

The picture is the argument. A single unit sends 12.4 cents of every dollar to the people who own it. Two units send one cent. The revenue nearly doubled and the take fell by 92%, and every number in that waterfall assumes both restaurants perform.

35.4 Cannibalization and trade-area overlap

Cannibalization is revenue at a new unit that was transferred from an existing unit rather than newly created — plus the second-order costs of the transfer, which are usually larger than the transfer itself.

Operators consistently underestimate it for a reason that is almost endearing: they think of the second restaurant as reaching new people. And it does, some. But the people most likely to try your second restaurant in its first six months are, by an enormous margin, the people who already like your first one. Your opening is announced to your own list. Your regulars are the ones who show up in week two. Your press coverage reaches your existing audience first.

The easiest cover to take is one you already have, and that is exactly the problem.

Chapter 2's finding, and why it is decisive here

Chapter 2 established that Bellwether is share-taking in a supplied market. Rivermill was not underserved. Bellwether did not create demand; it won covers from operators who already had them, on the strength of a hearth, a short menu, and a room people wanted to be in.

That fact, which was a strength in Chapter 2, is a liability in this chapter, and the logic is worth walking slowly.

If your restaurant succeeds by taking share, then your competitive advantage is relative — you are better than the alternatives within a given drive time. Now open a second unit and ask: against whom does it take share? In a trade area that overlaps the first one, the strongest competitor in the set is your own restaurant, because it shares your menu, your service standard, your price point, and your list. You have built a competitor with perfect knowledge of your positioning and no ability to be differentiated from you.

A restaurant that creates demand — that fills a genuine gap — has more room to open a second unit nearby, because its second unit competes with the gap, not with itself. A restaurant that takes share does not have that room.

Measuring the overlap, with data you already own

You do not have to guess at this. Every restaurant with a reservation system and any kind of guest list is sitting on the answer.

🧾 Read the Numbers

```text FIGURE 35.6 — "Where the covers live" [the Bellwether plan, extended] THE ARTIFACT A postal-code distribution of dinner covers, pulled from the reservation system and the email list, covering the trailing twelve months, set against drive times to the current restaurant and to a site under consideration. THE CONTEXT Bellwether, end of year one. 475 dinner covers a week across five services. The site under consideration is 14 minutes from Rivermill.

POSTAL AREA        SHARE OF     DRIVE TO      DRIVE TO      OVERLAP
                   COVERS       UNIT ONE      THE SITE      RISK
──────────────────────────────────────────────────────────────────────
Rivermill             31%       walk          14 min        low
Eastbank              22%        9 min         6 min        HIGH
The Heights           14%       12 min         5 min        HIGH
Southline              9%       15 min        21 min        low
University             8%       11 min        16 min        low
Outer ring / away     16%       20+ min       20+ min       low
──────────────────────────────────────────────────────────────────────
TOTAL                100%

Covers living closer to the new site than to the current one:  36%
36% of 475 dinner covers a week = 171 covers in the overlap zone.

WHAT IT SHOWS More than a third of the existing dinner business lives in two postal areas that are materially closer to the proposed site than to the current restaurant. These are not marginal guests — Eastbank and the Heights are the two largest non-local sources, which means they are the guests who already drive past three alternatives to get here. They are precisely the guests most likely to switch, because switching gets them the same experience with less driving. WHAT IT DOESN'T It does not say how often each guest visits, which is the variable that actually determines transfer. A postal code with 22% of covers and a six-visit-a-year average behaves very differently from one with 22% of covers and a one-visit average. Pull frequency before you model. It also tells you nothing about the site's own resident demand, which is the only thing that would make the second unit additive rather than substitutive. THE DECISION Model transfer explicitly, by night, before running any pro forma. Then re-run the site search with a minimum separation set by this table rather than by what is available: a site where overlap is under a fifth of covers, not over a third. THE LESSON Your reservation system already knows whether your second restaurant will compete with your first one. Ask it before you ask a broker. ```

A working threshold, offered as practitioner judgment rather than research: if more than about a fifth of your existing covers live closer to the new site than to your current one, you are not opening a second restaurant. You are opening a competitor with your own name on it. Bellwether's overlap is 36%.

Converting overlap into covers, which is where it gets real

Percentages of a guest list are abstract. Covers per night are not, and covers per night is how this damage actually presents.

Take the 171 covers a week in the overlap zone and apply a visit-shift rate — the share of those visits that move to the new unit. At 27.5%, which is conservative for a same-brand unit five to six minutes closer, that is 47 covers a week of gross transfer.

But not all 47 are lost. On Friday and Saturday, Bellwether has a waitlist; a transferred reservation is backfilled within the hour. On Tuesday and Wednesday there is no waitlist, because Chapter 32's whole point is that Tuesday and Wednesday are demand-constrained. So the transfers are backfilled exactly where they don't hurt and not backfilled exactly where they do.

FIGURE 35.7 — Unit one's dinner week, before and after unit two opens    [the Bellwether plan]

  covers          0     25    50    75   100   125   150
                  ├─────┼─────┼─────┼─────┼─────┼─────┤
                                  ▲ cash break-even, 77 covers (Ch. 32)
  BEFORE                          │
    Tuesday    66   ██████████████████
    Wednesday  72   ████████████████████
    Thursday   92   ██████████████████████████
    Friday    118   █████████████████████████████████
    Saturday  127   ████████████████████████████████████
                    average 95      ceiling 132 (the hearth)     cushion: 18 covers

  AFTER — gross transfer 47/wk, backfill 19 on Fri and Sat, net loss 28/wk
                                  │
    Tuesday    60   █████████████████         −6   (17 below break-even)
    Wednesday  65   ██████████████████        −7   (12 below break-even)
    Thursday   83   ███████████████████████   −9
    Friday    114   ████████████████████████████████  −12 gross, +8 backfill = −4
    Saturday  125   ███████████████████████████████████  −13 gross, +11 backfill = −2
                    average 89.4                                 cushion: 12.4 covers

Walk that figure slowly, because it contains the chapter's most important operational point.

Before: Tuesday runs 11 covers below cash break-even and Wednesday runs 5 below. That is already the business's structural weakness — two of five dinner services do not cover their own cash. They are carried by Friday and Saturday. Chapter 32 told us the cushion between the 95-cover plan and the 77-cover break-even is 18 covers, not the 37 covers of headroom up to the 132-cover ceiling. Those are different numbers measuring different things, and only the first one protects you.

After: Tuesday is 17 below and Wednesday is 12 below. The weekly average falls from 95 to 89.4, and the cushion falls from 18 covers to 12.4 — a 31% reduction in the only protection this business has against a bad quarter.

And notice where the damage landed. Twenty-two of the 28 net covers lost came off Tuesday, Wednesday, and Thursday, the three nights that were already the problem, because those are the nights with no waitlist to backfill from. Cannibalization does not spread evenly. It concentrates on your weakest nights, because your strongest nights defend themselves.

The annual dollar figure: 28 covers × 52 weeks = 1,456 covers × \$46 = **\$66,976 of sales out of unit one. At a 57% incremental contribution margin — blended COGS at 27.8%, variable labor at about 11%, and variable other operating (card fees, guest supplies, linen) at about 4.2%, so 43% variable cost — that is \$38,176 of contribution gone**.

A note on that 57%, because a rate used carelessly will mislead you. An incremental contribution margin is only valid for modest changes in volume around your current operating point. Lose 28 covers a week and none of your fixed costs move, so 57% is right. Lose 300 covers a week and you would close a shift, cut a salaried position, and change your purchasing — the rate would be wrong. Never use a marginal rate to evaluate a structural change.

The three second-order costs nobody models

The \$38,176 is the visible damage. Three more costs are real and harder to quantify, and I want to name them rather than fake numbers for them.

1. You lose the waitlist. A restaurant that turns away 15 people on a Saturday has pricing power, a reason to hold its check average, and a genuine cushion against a bad month. A restaurant that fills exactly to capacity has none of those things, and it does not find out until the month it doesn't fill.
2. Scarcity is part of the product. A neighborhood restaurant that is hard to get into on Friday is a different product from one with two locations and tables available at both. This is not sentimentality; it is a real effect on reservation behavior and on how far people will drive.
3. The transfer is permanent-ish. Guests who switch to the closer unit mostly do not switch back. In the stabilized model I assumed \$40,000 of the \$66,976 sticks. That is a guess, and it is the guess I would most want to test before signing.

🔍 Check Your Understanding

1. A restaurant runs 95 covers a night with a cash break-even of 77 and a kitchen ceiling of 132. An operator says "we have 37 covers of room." What is wrong with that statement?
2. Why does cannibalization concentrate on the weak nights rather than distributing across the week?
3. Your reservation data shows 12% of covers living closer to a proposed site. Your partner's data, using the email list, shows 34%. Which do you trust, and what would you check?

(1: The 37 covers is headroom to the ceiling — upside you may never sell. The cushion is the 18 covers between the plan and break-even, which is the only thing protecting you from a bad quarter. Confusing them makes a fragile business look robust. 2: Because strong nights have a waitlist that backfills a transferred reservation and weak nights do not — the damage lands wherever there is no queue. 3: Neither, yet. Reservation data covers only booked parties and misses walk-ins and the bar; an email list is skewed toward whoever signed up. Check both against actual visit frequency, and weight by covers rather than by guest count.)

35.5 Management bench: the general manager you must have before you sign

A management bench is the group of people already employed and already performing who can take over a unit's leadership, in place, without a search. It is measured not by titles but by a single question: how long does the business run correctly with each of them in charge and the owner unreachable?

By that measure Chapter 21 was unambiguous. Bellwether has no bench at all. The chef-owner works six days. The FOH partner runs the room. Neither can be absent for a week without the operation degrading. Four salaried people, two of whom are the owners.

The arithmetic of a bench

The rule is simple and operators hate it: you need one leader per leadership seat, plus one in development.

Two buildings have four leadership seats — a general manager and a kitchen leader in each — plus you need a fifth person deep enough to absorb a resignation without triggering a search during service. That is five leaders. Bellwether has two, and both of them are owners, which means they are also the above-unit management, the capital allocators, and the people who sign the guaranty.

FIGURE 35.8 — The bench you have and the bench two buildings need     [the Bellwether plan]

  TODAY — one building, 31 people, four salaried, two of them owners
                        ┌──────────────────────┐
                        │  chef-owner          │  BOH lead, six days a week
                        │  FOH partner         │  runs the room, every service
                        └──────────┬───────────┘
                   ┌───────────────┴───────────────┐
             ┌─────┴──────┐                  ┌─────┴──────┐
             │   sous     │                  │    AGM     │
             └────────────┘                  └────────────┘
                            27 hourly staff
  Leadership seats filled: 2.  Depth behind them: 0.

  REQUIRED — two buildings
                        ┌──────────────────────┐
                        │  the two partners    │  above-unit; no station of
                        │                      │  their own in either building
                        └──────────┬───────────┘
              ┌────────────────────┴────────────────────┐
        ┌─────┴───────┐                           ┌─────┴───────┐
        │  UNIT ONE   │                           │  UNIT TWO   │
        │  GM     NEW │                           │  GM     NEW │
        │  exec chef  │                           │  chef   NEW │
        │  sous       │                           │  sous   NEW │
        └─────────────┘                           └─────────────┘
                     + one leader in development  NEW
  Leadership seats needed: 4, plus 1 in development.  Must be bought: 4.

Price the four purchases at market, loaded:

(Check: \$68,000 + \$72,000 + \$58,000 + \$66,000 = \$264,000; × 1.22 = \$322,080. Unit two's three seats total \$241,560 and sit inside its \$440,200 labor line, leaving \$198,640 of hourly.)

The fourth row is the one operators argue with, so let me defend it. Why does unit one need a general manager when it already has two owners in it?

Because after the second building opens, it doesn't. The partners are now the above-unit management of a two-unit company: they are hiring for two buildings, doing two sets of books, negotiating two leases, sitting in two health inspections, and driving between them. The seats they currently fill at unit one become part-time seats, and a 68-seat restaurant doing \$1.55M with a part-time floor leader is a restaurant whose numbers move. You do not get to keep the partners in unit one and also have them run the company.

Operators try to dodge this in exactly one way: put the FOH partner in unit two's GM chair, and defer unit two's \$82,960 hire for a year. It works arithmetically — you save \$82,960 — until you notice that unit one's \$80,520 general manager was purchased specifically to replace the FOH partner. **You cannot save the same salary twice.** The two structures are within about \$2,400 of each other and differ in only one respect: which building has an owner standing in it. Choosing to put the owner in the new building means choosing to remove the owner from the building that is paying for everything.

The bench has to exist first, and for a year

Here is the part that costs time rather than money, and time is the binding constraint on this whole decision.

The bench must be in place, and performing, for a full year before you sign a second lease. Not hired. Not promising. In seat for twelve months. The reason is that a leader who has not been tested by a bad quarter has not been tested. Over twelve months a general manager will face:

• a slow February, when the schedule has to be cut without wrecking service
• a health inspection, and possibly a re-inspection
• a peak December, with private parties stacked against a full dining room
• at least one resignation from a key hourly position, at the worst possible moment
• one genuine crisis — equipment failure, a walk-in going down, a staff conflict, an incident with a guest

Anyone can run a good Saturday. What you are buying is judgment under load, and there is no way to observe it except to wait for load. A manager who has been in the seat four months and looks great has demonstrated that four months is not long enough to find out.

👨‍🍳 On the Line

How you actually build a bench, starting Monday.

Nobody builds a bench by announcing one. Here is the mechanism, and it is uncomfortable because it requires the owner to be worse at their job on purpose.

Step one: pick the seat, then the person. Write the scope first — "this person owns the schedule, the FOH labor number, the reservation book, and the Tuesday program" — and only then decide who fills it. Owners who pick the person first end up with a promotion and no job description, which is how you get a loyal employee who fails.

Step two: hand over the decision, not the task. Letting the AGM write the schedule while you approve it teaches nothing; they are still doing data entry for your judgment. Hand over the labor target and let them miss it. The first month they will run 34% against a 32.3% target, and it will cost you real money, and that money is tuition. Chapter 19's staffing guide is what makes the tuition affordable rather than catastrophic.

Step three: leave. Not for two weeks — start with a Tuesday. Then a Friday. Then a full week. The absence audit from §35.1 tells you exactly what breaks, and every break is a page of the manual.

Step four: pay them like the asset they are. A general manager who can run your building while you are unreachable is worth more than the difference between their salary and the next candidate's, because what they are actually holding is your ability to ever do anything else. Underpaying that person is the most expensive economy in this industry — they leave, and you are back to being the bench.

The thing nobody tells you: building a bench makes the restaurant less profitable for about eighteen months. You are paying for a layer of management that the business does not strictly need at one unit. That is the real cost of optionality, and it is why so few independents ever have it. It is also why the ones who do can sell the business, take a month off, or open a second one — three things that are, structurally, the same thing.

The bench is the gate for every option, not just the big one

I want to make a point here that most growth chapters miss, because it changes the whole plan.

The bench is not a requirement of the second location specifically. It is a requirement of every option on the ladder in Figure 35.2, including the cheap ones. Catering at 45 events needs a kitchen leader who can run production without the chef-owner. A delivery-only brand needs somebody who owns its specs and its packaging standard. Filling Tuesday needs a person who owns a marketing calendar. Chapter 19 found the salaried week full; adding any of these adds hours to a week that has none.

Which means the bench is not the thing standing between you and a second restaurant. The bench is the thing standing between you and growth of any kind — and the cheap options are how you build it. That reframing is the difference between a plan that says "not yet" and stops, and a plan that says "not yet, and here is what we do for the next twenty-four months." We will build that plan in the checkpoint.

35.6 Capital for growth: what changes when you're borrowing on a track record

Growth capital is money raised to expand an existing business rather than to start one. The distinction is not cosmetic, because the thing being evaluated changes completely.

When you financed the first restaurant, you were borrowing against a projection. You had a plan, an injection, a guaranty, some collateral, and industry experience. The document did the arguing.

When you finance the second one, you are borrowing against a history. There are real statements now. And here is the thing operators are not ready for: a track record can argue against you as easily as for you. A first unit that ran a 63% prime cost and a thin cash position for eighteen months is not neutral evidence; it is evidence. The plan for unit one could promise anything. The statements for unit one cannot.

What is different, structurally

Five things change, and they compound.

One: the analysis is now about the company, not the project. A second unit's ramp is funded by the first unit's cash flow. That makes the first unit's cushion part of the second unit's financing in an economic sense, whether or not it is pledged in a legal one. If you take nothing else from this section: the money that funds your second restaurant's first year is your first restaurant's profit, and you are spending it before you have it.

Two: the guaranty stacks. This is the one I most want you to feel. Bellwether's personal exposure today is \$1,367,600, and it decomposes:

A second unit adds a term note of \$425,000 and a second ten-year lease guaranty of roughly \$1,046,000 — about \$1,471,000** — for a combined **\$2,838,600. The partners' personal exposure roughly doubles on the day they sign, before the second building has served one guest.

And here is the structural point that almost nobody says out loud: a single-unit operator's downside is bounded by one building. A two-unit operator's is not. When the same two people guarantee both leases and both notes, a failure at unit two does not stay at unit two. It arrives at the partners personally, and the partners are the only thing holding unit one together. You have converted a bounded loss into an unbounded one, and no one will send you a statement about it.

Three: what "cheap capital" means changes. Operators evaluate financing on the interest rate. That is the least important of the three prices you pay.

(SBA program structures are real and are described here in general terms; terms, eligibility, and guaranty requirements change and vary by lender and borrower. Verify current program details before relying on any of this, and use an attorney and an accountant on anything you sign.)

The last row is why §35.7 and §35.8 matter so much. Retained cash is the only growth capital that costs nothing on any of the three axes — no rate, no guaranty, no control — and the way you generate it is by earning more inside the guaranty you have already signed.

Four: outside equity is not free money, it is the most expensive money. Equity has no payment, which makes it feel cheap in the year you take it and expensive for the next thirty. If an investor takes 30% for \$200,000 and the business eventually clears \$300,000 a year, that \$200,000 costs \$90,000 a year, forever, and you now have a partner in every decision including the one where you want to stop. Equity is the right answer for growth you genuinely cannot service with debt — but the honest test is whether the growth is worth a permanent partner, not whether the check clears.

Five: real estate changes the calculation, and it is the most underrated growth move in this business. Every dollar of rent is a dollar you will never see again. Buying the building — through SBA 504, which exists for exactly this — converts an operating expense into an amortizing asset, and it removes the lease guaranty, which is 76% of Bellwether's personal exposure. An operator who buys their building and runs one restaurant in it for twenty-five years frequently ends up wealthier than one who runs four leased units, and the reason is not the restaurants. The most reliable wealth in this industry is the real estate, and the restaurant is what pays for it. That is not cynical. It is the single most useful sentence in this section.

⚠️ Where the Money Leaks

The injection you thought you had.

Bellwether's second-unit stack calls for a \$120,000 owner injection. Where does it come from?

"From the business" is the answer everybody gives, and the business produces \$191,520 of cash after debt service. So \$120,000 is seven and a half months of it. Fine.

Except that \$191,520 is also:

• the partners' entire household income, for two households
• the source of every unbudgeted repair, and a hearth-driven kitchen generates them
• the reason the working-capital reserve stayed intact through the first slow February
• the only cushion between a bad quarter and a hard conversation

Take \$120,000 out of it and the partners have \$71,520 to live on and absorb shocks with, across two people, for a year — in the same year they are opening a restaurant, which is the most expensive and least predictable year an operator ever has.

The leak is not the \$120,000. The leak is treating cash flow as though it were surplus. Chapter 33 made this argument about weeks; it is exactly the same argument about years. A profitable business runs out of money by spending profit that was already committed, and "already committed" includes eating.

35.7 Line extensions: catering, retail products, licensing, and their real margins

A line extension is a new revenue stream built on an existing restaurant's brand, kitchen, or customer base — catering, a retail or consumer-packaged-goods (CPG) product, a delivery-only brand, a license — without opening a new dining room.

These are the options operators skip, and they skip them because line extensions are unglamorous. Nobody's friends congratulate them on a catering program. But look at the ladder in Figure 35.2 again: every one of them sits above the double line. No new lease guaranty. No ten-year signature. And their returns per dollar of capital are, in this business, extraordinary — because they use assets you have already bought and already guaranteed.

Catering: the best return in the chapter, and it has a ceiling

Bellwether already caters, a little. Chapter 29 found 14 events a year at 3.5 hours of coordination each — 49 hours a year coming out of a salaried week that Chapter 19 found fully committed. That is the current state: a program that runs on borrowed time and produces modest revenue.

What happens if you actually run it?

🧮 Run the Numbers

Catering at 45 events a year, costed honestly.

Line	Amount	% of catering revenue
Revenue: 45 events × \$2,850 average | **\$128,250**	100.0%
Food cost (better than the dining room: fixed menus, no waste on à la carte)	(\$33,345)	26.0%
Direct event labor: prep hours, event staff, delivery and setup	(\$28,215)	22.0%
Rentals, disposables, transport, and comped tasting product	(\$11,543)	9.0%
Contribution before coordination	\$55,147	43.0%
Events coordinator: 12 hrs/week × \$26 loaded × 52 weeks | (\$16,224)	12.7%
NET ANNUAL CONTRIBUTION	\$38,923	30.4%

Capital required:

Item	Amount
Used cargo van	\$22,000
Hot boxes, cambros, and transport gear	\$6,500
Chafers, serving pieces, and event smallwares	\$4,800
Walk-in shelving reconfiguration and a dedicated staging rack	\$3,200
Commercial auto insurance, catering permit, and endorsements	\$2,100
TOTAL CAPITAL	\$38,600

Return on capital: \$38,923 ÷ \$38,600 = 100.8 cents per dollar of capital.

Against Bellwether's restaurant return of 42.0¢ per dollar of capital, that is two and a half times the return — and it adds zero new personal guaranty, because you already have the kitchen, the lease, and the license.

Note the coordinator line, and note that I did not skip it. The temptation is to leave it out, because the chef-owner "already does that." Chapter 29 measured what "already does that" costs: 3.5 hours an event. At 45 events that is 157.5 hours a year, up from 49 — an extra 108.5 hours out of a week that has none. Buying the time at \$16,224 is what makes the \$38,923 real rather than borrowed from a person who is already at capacity. A catering program that runs on unpaid owner hours is not a business; it is a loan against the owner, and it comes due.

Now the limit, because every method in this book gets its limits stated.

Catering has a hard ceiling set by the building, not by demand. Prep for a catered event competes with prep for service — for the hearth, the prep table, the walk-in, and the oven. Bellwether's usable catering-prep window is Monday, which is dark, plus Sunday after brunch: roughly 16 BOH hours a week that do not touch service prep. At about 8 hours of prep per event, that is two events a week, or about 100 a year, and that is a wall.

Push past it and you are prepping catering during Friday's service prep, which is precisely how Chapter 29's second case study ended: a restaurant that grew its events until they cannibalized the dining room. Its conclusion is the sentence I would put over the door of this entire chapter:

"Growth that consumes the thing that was working is not growth."

At 45 events you are at 45% of the building's capacity, which leaves room to be wrong. At 90 you are choosing between a wedding and a Friday, and you will choose wrong at least once.

Retail and CPG: the arithmetic that ends most of these projects

Bellwether's salsa verde is on every plate of Hearth Chicken and costs \$1.05 a portion. Guests ask for it. Putting it in a jar is the most natural idea in the world and it is where a great many operators lose \$15,000 and eighteen months.

Here is why.

⚠️ Where the Money Leaks

The jar that has to sell 27,802 times.

```text FIGURE 35.9 — "The salsa verde, in a jar" [constructed teaching example]

COST TO PRODUCE, PER 8 oz JAR (co-packed, minimum run 2,400 units) Ingredients, at co-packer scale $1.42 Jar, lid, and label 0.61 Co-pack run charge, allocated 0.88 Inbound freight to the restaurant 0.19 ───────────────────────────────────────────────────────────────── LANDED COST PER JAR $3.10 First production run: 2,400 x $3.10 $7,440

ONE-TIME COSTS BEFORE THE FIRST JAR EXISTS Recipe scale-up and shelf-stability work with a process authority (an acidified shelf-stable food) $3,800 Label design and a nutrition-facts panel 2,400 UPC assignment and a trademark search 1,100 ───────────────────────────────────────────────────────────────── $7,300 TOTAL TO GET TO A FIRST RUN ON A SHELF $14,740

MARGIN, BY CHANNEL Channel Your price Cost Margin/jar Margin % ───────────────────────────────────────────────────────────────── Sold in your own dining room $12.00 $3.10 $8.90 74.2% Wholesale direct to a grocer $6.00 $3.10 $2.90 48.3% Through a distributor $4.50 $3.10 $1.40 31.1%

JARS REQUIRED TO EARN WHAT THE CATERING PROGRAM EARNS ($38,923) In your own dining room: $38,923 / $8.90 = 4,373 jars/yr (84/week) Through a distributor: $38,923 / $1.40 = 27,802 jars/yr (535/week) ```

The same profit costs you 4,373 jars in your own dining room or 27,802 through a distributor — 6.4 times the volume for the same money.

And 84 jars a week is not modest. On roughly 685 covers a week, that is better than one jar per eight guests buying a \$12 retail item on their way out. It is achievable with a great product and a trained staff. It is not achievable by putting a stack of jars by the host stand.

The distributor row is where the projects die. Selling 535 jars a week means real grocery distribution, slotting, demos, a broker, and a marketing spend — none of which you have, and all of which are a different company with a different cost structure that has nothing to do with your restaurant. Your restaurant's advantage is that you have already paid for the rent, the kitchen, and the guest's attention. In a grocery aisle you have none of those advantages and you are competing with companies whose entire business is that aisle.

The second trap is the minimum run. 2,400 jars is set by the co-packer, not by your demand. At 84 a week you sell through in 29 weeks — fine. At 20 a week you have 120 weeks of inventory and a product with a 12-to-18-month shelf life, which means you will be dumping jars you paid \$3.10 for. The minimum run sets your inventory before you know your velocity, which is the exact opposite of how you buy anything else in a restaurant.

What a disciplined operator does instead: sell it in the dining room first, in jars you fill and label in house under whatever your jurisdiction permits, at whatever volume you can honestly move, for one full year. That measures velocity for a few hundred dollars. Only if the in-house number is genuinely large does a co-packed run make sense — and even then, the answer is usually "sell more of them here," not "get into grocery."

⚖️ Code and Compliance

Line extensions change your regulatory footprint, sometimes dramatically.

Every option in this section moves you into a rule set your restaurant license does not cover. In general terms, and with the standing warning that all of this varies by state, county, and city and changes over time:

• Catering and off-site service typically require a separate catering endorsement or permit, transport-temperature controls and logging, and often a commissary designation for the originating kitchen. Off-premise alcohol service is a separate question again, and in many jurisdictions your on-premise license does not travel. Chapter 29 covered the operational side; the licensing side is local and non-negotiable.
• A shelf-stable retail food product is regulated on an entirely different track from restaurant food. In the United States, a packaged food entering commerce generally brings food-facility registration, allergen labeling, and a nutrition-facts panel into play, and an acidified or low-acid shelf-stable product brings process-authority review and filed processes. Your restaurant kitchen is very often not a lawful place to produce it. Cottage-food exemptions exist in most states but are narrow, usually exclude acidified and refrigerated products, and generally do not permit interstate sale. Talk to a process authority and to counsel before you talk to a co-packer.
• Licensing your concept carries a risk that catches operators badly. In the United States, the FTC Franchise Rule defines a franchise by what the arrangement does, not by what you call it. Broadly, where there is a trademark license, significant control over or assistance with the licensee's operations, and a required payment, you may have sold a franchise — with the disclosure obligations that entails, and in a number of states a registration requirement as well. Calling the document a "license agreement" does not decide the question. Chapter 36 covers this properly; the only thing to take from here is do not paper a licensing deal without a franchise attorney.
• A second unit or a delivery brand means a second set of everything: permits, licenses, food-handler and manager certifications, a separate health-department file, and in most jurisdictions a separate liquor license that is not transferable from your first one and may be quota-limited.

None of the above is legal advice, and all of it varies. Verify locally, in writing, before you spend money.

Licensing: small money, real risk

Licensing is granting another operator the right to use your name, recipes, and specifications for a fee, without the ongoing system-support obligations that define franchising.

The pitch is seductive: a hotel or a ballpark wants your name on a stall, you take a royalty, and you do nothing. The reality has three problems.

The money is small. A licensee doing \$1,200,000 at a 4% royalty pays \$48,000 a year — real money, but it arrives with an obligation to support them, and support costs owner-hours you do not have.

The quality risk is total and asymmetric. A licensed unit that runs a 46°F walk-in is your brand's incident. You carry all of the reputational downside and 4% of the revenue.

The legal line is not where you think it is. See the compliance callout above; this is Chapter 36's subject and it is genuinely complicated.

For an operator at Bellwether's stage, licensing is best understood as something to be offered and politely declined until there is a documented system worth licensing — which is, again, Chapter 37's manual.

The small-format second unit: the second location that actually works

If you are going to open a second building, this is the one to open.

(Small-format P&L: revenue \$586,872; COGS 30.0% = \$176,062; labor 27.0% = \$158,455; prime \$334,517; occupancy \$47,600 on 1,400 sq ft at \$34 all-in = 8.1%; other operating 13.0% = \$76,293; G&A 3.0% = \$17,606; total costs \$476,016; operating profit **\$110,856.)

The small format wins on every axis that matters, and it wins for a reason worth internalizing: it needs less of you. A counter-service format with a short menu and no hearth can be run by a strong manager. A 64-seat full-service restaurant with a wood-fired hearth needs a chef, and there is only one of those in this company.

Add it to the existing unit and the group's numbers actually improve: \$261,020 + \$110,856 = \$371,876** of operating profit on \$910,000 of combined capital (40.9¢ per dollar) and \$1,912,600 of exposure (19.4¢** per dollar) — slightly better than one unit alone, on the same order of guaranty risk per dollar earned.

It still fails the gates in Figure 35.1. It is still a lease guaranty. But it fails them by much less, and if the partners ever build the bench, this is the row of the ladder to start on.

🔍 Check Your Understanding

1. Catering returns 100.8¢ per dollar of capital and the restaurant returns 42.0¢. Why doesn't every restaurant just cater?
2. Why is the same jar of salsa worth \$8.90 to you in your dining room and \$1.40 through a distributor, when the product is identical?
3. An operator says "the license is not a franchise, we just take a fee for the name." What is the specific risk in that sentence?

(1: Because the ceiling is set by the building — about 16 usable prep hours a week outside service prep, or roughly 100 events a year — and because past that point catering starts consuming the dining room's kitchen, which is Chapter 29's failure case. High return on a small base is not the same as a scalable business. 2: Because in your dining room you have already paid for the rent, the staff, and the guest's attention; through a distributor you are buying all three back at market, plus the distributor's and the grocer's margins. The product is the same; the cost structure around it is not. 3: Under the FTC Franchise Rule the arrangement is classified by what it does — a trademark license plus significant control or assistance plus a required payment — not by what the document is titled. Calling it a license does not make it one, and the disclosure and registration consequences are serious. Use a franchise attorney.)

35.8 When not to grow — and how to make one great restaurant a career

I have spent seven sections making a negative case. Now let me make the positive one, because "don't grow" is not a consolation prize and it is not what I am arguing. I am arguing for a different definition of growth.

There are 37 covers a night inside a building you have already guaranteed

Start here, because it reframes everything.

Bellwether's hearth caps the kitchen at about 132 covers. The plan is 95. Cash break-even is 77. The room is demand-constrained Tuesday and Wednesday — not capacity-constrained, demand-constrained, which is a completely different problem with a completely different solution.

There are 37 covers a night of unused capacity inside a lease the partners have already personally guaranteed for \$1,032,600. Selling those covers requires no new capital, no new guaranty, no second set of permits, and no general manager for a building that does not exist.

The theoretical maximum is worth computing once, just to see the size of it: 37 covers × \$46 × 5 nights × 52 weeks = **\$442,520 of additional dinner sales, at a 57% contribution margin = \$252,236**. That is very nearly a second restaurant's worth of profit, from a building that is already paid for and already signed.

You will not get that, and I am not going to pretend otherwise. Tuesday will never do 132 covers. But look at what happens if you get half the gap on the three soft nights:

🧮 Run the Numbers

Filling Tuesday, Wednesday, and Thursday.

Night	Now	Target	Gain
Tuesday	66	85	+19
Wednesday	72	90	+18
Thursday	92	105	+13
Friday	118	118	—
Saturday	127	127	—
Weekly total	475	525	+50

50 covers a week × 52 weeks = 2,600 covers × \$46 = **\$119,600 of additional sales. At a 57% incremental contribution margin: \$68,172** of additional contribution. Less a program cost of \$18,000 a year — a Tuesday offer, a small paid-media budget, a loyalty push, a private-dining sales effort, and the salaried hours to own it.

Net: \$50,172 a year. Zero new capital. Zero new personal guaranty.

And look at what it does to the thing that actually protects you: the weekly average goes from 95 to 105 covers, and the cushion above cash break-even goes from 18 covers to 28 — a 56% increase in the only margin of safety this business has.

Compare that to the second restaurant, which reduced the cushion from 18 covers to 12.4.

One of these two options makes the business safer and the other makes it more fragile, and the fragile one costs \$680,000.

This book's fourth theme is the whole argument here: every seat-hour is inventory you cannot store. An empty chair on Tuesday at 7:15 is revenue that has expired. You have already paid the rent on it, the insurance on it, the manager standing near it, and the light above it. The marginal cost of filling it is 43 cents on the dollar. There is no growth investment in this entire chapter with a better return than selling the inventory you have already bought.

The full inside-the-walls program

Stack the three options that require no new guaranty:

\$128,336 of new annual contribution for \$50,600 of capital — 253.6 cents per dollar — with no new personal guaranty and nothing signed that cannot be unwound in a season.

Set that against the second restaurant's first year, which costs \$163,996. **The swing between the two paths is \$292,332 in year one**, and one of them can be stopped in thirty days.

Note also what these three do to the ratio that matters most:

FIGURE 35.10 — Return per dollar of personal exposure, by path      [constructed]

  Ch. 30's illustrative food truck            84.4¢  ██████████████████████████████████
  ONE UNIT + catering + a delivery brand      24.8¢  ██████████
  ONE UNIT + a small format                   19.4¢  ████████
  ONE UNIT TODAY                              19.0¢  ████████
  Two restaurants, stabilized (unit two yr 3) 10.3¢  ████
  Two restaurants, year one                    6.4¢  ███

  (Bellwether today: $261,020 of operating profit on $1,367,600 of personal
   exposure. With catering and a delivery brand: $339,184 on the same
   $1,367,600, because neither one adds a guaranty.)

Growing inside the four walls raises the return on the exposure you have already signed from 19.0¢ to 24.8¢. Growing by building a second restaurant cuts it to 6.4¢ in year one and, even at full maturity, only reaches 10.3¢.

And note what the figure also shows, honestly: nothing inside the four walls gets near the truck's 84.4¢. That is not a failure of the line extensions. It is the arithmetic of the denominator. The truck's advantage is not that it earns more; Chapter 30 was clear that it earns much less in absolute dollars — 3.43 of them are needed to match one Bellwether, which means each one produces under 30% of what the restaurant produces. Its advantage is that it never signed a ten-year lease. You cannot undo a guaranty by earning more inside it. You can only stop adding to it.

But there is a catch, and I am not going to hide it. Those three programs together add roughly nine to twelve hours a week of owner and salaried time, to a week Chapter 19 found full. Which means the answer here is the same as the answer to the second restaurant: you cannot do all three either — not without the bench. The difference is that these three build the bench while they earn, and a second restaurant consumes it before it exists.

The honest case for one restaurant, held

Now the part this industry never says out loud.

A single restaurant, run well for thirty years, is a genuinely excellent financial instrument, and it is not a failure of ambition. Here is the arithmetic.

It de-leverages itself. Bellwether's \$60,000 equipment lease retires in year five; debt service falls from \$69,500 to \$54,250 and cash after debt service rises to **\$206,770. The SBA note retires in year ten; debt service goes to zero and cash after debt service becomes the whole operating profit — \$261,020** in today's dollars, on a business the partners have already paid for.

Years eleven through thirty are 20 × \$261,020 = \$5,220,400 of pre-tax cash, in today's dollars, from a business that is finished being bought. That is before any growth in check average, any of the line extensions above, and any of the soft-night covers.

And the exposure falls every single year. The note amortizes. The lease term runs down. By year ten the note is gone and the guaranty is only the remaining lease term. A single unit is a self-de-leveraging asset. The growth path is the exact opposite: every new unit resets the exposure clock to ten years and adds a fresh note at the top of its amortization.

FIGURE 35.11 — Two paths, in personal exposure over time              [constructed]

  YEAR       1      3      5      7      9     11     13     15
  ────────────────────────────────────────────────────────────────
  ONE UNIT, HELD
    exposure ███████████████████████████████████████████
             $1.37M         $1.09M        $0.71M       $0.31M   → 0
             note amortizes; lease term runs down; nothing resets

  GROW AT YEAR 3, AGAIN AT YEAR 8
    exposure ██████████████████████████████████████████████████████████
             $1.37M  →  $2.84M  ...  $2.51M  →  $3.90M  ...  $3.5M
             every unit resets a ten-year clock at the top of a new note
  ────────────────────────────────────────────────────────────────
  The single-unit line is the only one on this page that reaches zero.

And there is an exit. A well-run single unit with clean books, a transferable lease, a documented system, and a manager who stays is a sellable business. I am not going to quote you a multiple, because independent restaurants trade across a very wide range and it is market-specific, size-specific, and deal-specific. What I will tell you is the thing that actually determines which end of that range you land on: the books determine the multiple far more than the food does, and a business that cannot run without its owner sells for the value of its equipment. The bench you build in §35.5 is not only what lets you grow. It is what makes the business worth something on the day you want to stop.

What "one restaurant" requires instead of growth

Choosing one unit is not choosing to stand still. It has its own program, and it is demanding:

• Raise the check deliberately, every year, through beverage attachment, menu engineering, and pricing discipline. Two dollars on a \$46 check across 95 covers a night is \$49,400 a year (Chapter 1 worked this).
• Fill the soft nights, permanently — the \$50,172 above, and then keep going.
• Hold prime cost. Every point you let drift is \$15,500 a year at Bellwether's volume, and the drift is invisible until it isn't.
• Build the bench anyway. Not to grow — to make the business saleable, to take a month off, and because Chapter 21's turnover math says a place people stay costs less to run.
• Buy the building if you ever can. It is the single highest-value move available to a one-unit operator, and it removes 76% of the personal exposure.
• Reinvest in the room every seven years or so. A restaurant that looks fifteen years old charges what a fifteen-year-old restaurant charges.

And the honest cost of not growing

I would be doing exactly what I criticized if I sold you this without its limits. Choosing one restaurant costs you real things:

• Your income is capped at one building's profit. There is a ceiling, and it is 132 covers a night.
• Your risk is concentrated. One lease, one neighborhood, one hood system, one road-construction project. Diversification is a genuine argument for a second unit and I am not dismissing it — I am saying it is not free, and the honest version costs \$680,000.
• You will train people who leave to open their own places. Some of them will do well. That will be harder than you expect.
• Your identity gets welded to a room. Twenty years in, "what else could I do" is a real question with an uncomfortable answer, and Chapter 40 takes it seriously.

Name those. Do not pretend the choice is costless. Then choose on the arithmetic rather than on whichever story you find more flattering.

🍽️ The Business Plan

Checkpoint 35 of 40 — the Growth section.

Every previous checkpoint added a section that made the plan more fundable. This one adds a section that says no, which is the hardest page in any business plan to write and the most persuasive one to read.

Section 12 — Growth

The second-location test, applied to Bellwether at the end of year one.

Answer: not yet.

Bellwether clears none of the seven gates in Figure 35.1 cleanly.

Gate	Status	Evidence
1. Absence — both owners out 14 days	✗	Ch. 21: no management bench exists
2. Replacement — profitable at market-rate management	?	never computed; possibly 5.5%
3. Repeatability — the system is written down	✗	no operations manual (Ch. 37's standard)
4. Bench — the leaders are already employed	✗	4 salaried, 2 of them owners; 5 leaders needed
5. Slack — room in the salaried week	✗	Ch. 19: fully committed; Ch. 29: 49 hrs already gone
6. Cash — the first-year hole is fundable	marginal	\$163,996 hole vs. \$191,520 that is also income
7. Audit — who audits the owners	✗	Ch. 34's question, still open

What a second location would actually do, in year one: reduce the partners' cash from \$191,520 to \$27,524 — a \$163,996** decline — while roughly doubling personal exposure from **\$1,367,600 to \$2,838,600**. At full maturity in unit two's third year, group cash reaches **\$138,000, which is still \$53,520 less than the single unit produces today. For the group merely to match today, unit two would need an 18.5% operating margin — better than Bellwether's own plan.

What the plan commits to instead, in order, over the next twenty-four months:

#	Program	Capital	New guaranty	Target contribution	Why it is first
1	Fill Tuesday, Wednesday, Thursday	\$0 *(\$18,000/yr program)*	none	\$50,172	it raises the cushion from 18 covers to 28
2	Catering to 45 events, with a paid coordinator	\$38,600 | none | \$38,923	it builds a kitchen leader who runs production without the chef-owner
3	Delivery-only second brand (Ch. 30)	~\$12,000 | none | \$39,241	it forces written specs — page one of the operations manual
	Total	\$50,600** | **none** | **\$128,336

Each of the three is chosen as much for the bench it builds as for the money it makes. That is the plan's actual thesis: growth inside the four walls is how you build the bench that makes growth outside them possible.

The conditions that would change the answer, stated as testable milestones

The plan does not say "someday." It says what would have to be true, in terms that can be measured and dated:

The binding constraint is Milestone 2, and it sets the date. Hiring the leaders takes three to six months. Twelve months in seat follows. Milestone 1 cannot even begin until the bench exists, and it needs two quarters. Milestone 6 needs twelve months of accumulation. Stacked honestly, the earliest defensible date for reconsidering a second location is roughly twenty-four months out — and when it comes, the first row of Figure 35.2 below the line to consider is the **small format at \$290,000**, not the \$680,000 restaurant.

What this checkpoint does not settle. Whether the concept works outside Rivermill — Chapter 30's ten-Monday residency, at \$430 of downside, is the cheapest available answer and the plan should schedule it. Whether the partners will accept a plan whose headline is "no." And whether the delivery brand and the catering program can genuinely be run inside a salaried week that Chapter 19 already found full, which is a question about the bench and therefore the same question as everything else in this chapter.

Open questions carried forward:

1. What do the partners actually draw, and therefore what is the real owner-adjusted margin? (Compute this month; it changes Milestone 3.)
2. Does the concept travel? (Chapter 30's residency test — schedule ten Mondays.)
3. Who audits the owners? (Chapter 34 asked. Milestone 7 is the answer, and it is overdue.)
4. Is a franchise — bought or sold — a better route to a second unit than building one? (Chapter 36.)
5. What does the operations manual actually have to contain for a standard to survive the owner's absence? (Chapter 37.)

Conclusion

The second restaurant kills more successful operators than the first one does, and the mechanism is not mysterious. Success at a single unit is very often the owner, and when you open a second building you do not duplicate the owner. You divide them, and you divide them at the exact moment you have doubled the number of buildings that need one.

For Bellwether the arithmetic is unambiguous and it does not depend on the second restaurant failing. A second unit that opens well — \$1,240,000 in year one, 10.2% operating profit, a solid ramp — still reduces the partners' cash from \$191,520 to \$27,524, because it adds \$80,520 of management to the first building, \$87,000 of company overhead that did not exist, \$85,300 of debt service, and takes \$38,176 of contribution out of the restaurant that is paying for everything. At full maturity, in unit two's third year, the group still produces less cash than one unit does today, on 89% more revenue and 108% more personal exposure. Unit two would need an 18.5% operating margin — better than Bellwether has ever planned for — merely to leave the partners where they already are.

Meanwhile there are thirty-seven covers a night of unused capacity inside a lease the partners have already guaranteed for \$1,032,600, and the marginal cost of filling them is forty-three cents on the dollar. Filling the soft nights, running catering properly, and launching Chapter 30's delivery brand produce \$128,336 of new annual contribution for \$50,600 of capital and no new signature — and every one of them can be stopped in a season.

That is the chapter's argument, and it is not an argument against growth. It is an argument about which growth, in what order, and against which denominator. The four axes are capital, guaranty, owner-hours, and reversibility, and the option operators reach for first is worst on all four.

The plan says not yet, with eight milestones and a date about twenty-four months out. And the thing that gates every one of them is the same thing: a management bench that does not exist, and that has to exist for a year before it counts. Building it is not a detour on the way to growth. It is the growth, and it is what makes the business saleable, survivable, and worth having whether or not there is ever a second building.

Chapter 36 takes up the option this chapter deliberately set aside. Franchising is two entirely different businesses wearing the same word — buying one is purchasing a job with a system attached; selling one means you are no longer in the restaurant business at all, you are in the business of selling systems. It reads the Franchise Disclosure Document, works the franchisee's real return after royalties and required spend, and then asks the question that belongs squarely next to this chapter's: what would have to be true for your concept to be franchisable? Most concepts fail that test. Chapter 37 then builds the thing both chapters keep pointing at — the operations manual and the standards that let "good" happen in a building you are not standing in.

Key Terms

Second-location test — a gate, not a scorecard: seven conditions (absence, replacement, repeatability, bench, slack, cash, audit) that must all be true before an operator signs for a second unit, because in a business with a small cover cushion each failure is independently capable of consuming it. (Ch. 35)

Owner dependence — the share of a business's results produced by the owner's personal presence, judgment, and relationships rather than by systems that would operate without them; measured with an absence audit that sorts eight weeks of owner-reaching decisions into standards-exist, standards-should-exist, and genuinely-owner bins. (Ch. 35)

Management bench — the people already employed and already performing who can take over a unit's leadership in place, without a search; measured by how long the business runs correctly with each of them in charge and the owner unreachable. Two units need five leaders: one per seat plus one in development. (Ch. 35)

Unit economics — the revenue, cost, and capital of a single operating location measured independently of the company that owns it, answering four questions: what one unit costs to build, what it earns at maturity with paid management in it, how long the ramp takes, and what the return is on both capital and personal guaranty. (Ch. 35)

Owner-adjusted unit profit — a unit's operating profit restated with market-rate management in the labor line in place of the owner's uncompensated or under-compensated work; the figure that actually repeats when a model is duplicated, and frequently far below the reported margin. (Ch. 35)

Cannibalization — revenue at a new unit transferred from an existing unit rather than newly created, plus the second-order costs of the transfer: the lost waitlist, the loss of scarcity, and the permanence of switched guests. It concentrates on weak nights, because strong nights backfill. (Ch. 35)

Trade area — the geography from which a restaurant actually draws its guests, measured from reservation and list data rather than estimated from a radius; overlap between an existing unit's trade area and a proposed site's is the direct driver of cannibalization. (Ch. 35)

Above-unit overhead — the costs of running the company that owns the restaurants rather than the restaurants themselves: group bookkeeping, accounting and legal, entity insurance, technology seats, vehicles, and eventually a director of operations. Zero at one unit because the owner absorbs it; the reason two units is the most structurally disadvantaged size. (Ch. 35)

Line extension — a new revenue stream built on an existing restaurant's brand, kitchen, or customer base — catering, a retail or consumer-packaged-goods (CPG) product, a delivery-only brand — without opening a new dining room, and typically without a new personal guaranty. (Ch. 35)

Licensing — granting another operator the right to use your name, recipes, and specifications for a fee, without the ongoing system-support obligations that define franchising; a legally fraught category, because an arrangement is classified by what it does rather than by what the document is titled. (Ch. 35)

Growth capital — money raised to expand an existing business rather than to start one, evaluated on three prices rather than one: the rate, the personal guaranty, and the control given up. Retained cash is the only source that costs nothing on all three. (Ch. 35)

Spaced Review

1. Without looking back: what are the four axes on which every growth option should be ranked, and which one do operators most often skip?
2. A restaurant reports 16.8% operating profit. Its two owner-partners take whatever is left after the vendors. What is the first calculation you would run before considering a second location, and roughly what would you expect it to do to the reported margin?
3. From Chapter 32: a business runs 95 covers a night, breaks even in cash at 77, and is capped by its kitchen at 132. State the cushion, state the headroom, and explain why confusing them is dangerous.
4. From Chapter 1 and Chapter 13: an operator plans to launch a catering program and says the food cost will be better than the dining room's. Give two reasons that is usually true, and one reason it might not be.
5. The recurring question: an operator has \$50,000 of retained cash and is choosing between a marketing program to fill Tuesday and Wednesday, a catering build-out, and a down payment toward a second restaurant. Rank the three, and state what single piece of information would most change your ranking.