How to open a restaurant step by step: the opening as a portfolio decision, not an act of faith

1. Why does opening sequence matter more than available capital?

Sequence outweighs capital because the lease is irreversible and the business model still isn't.

The United States closed November 2025 with more than 860,000 restaurant locations, an all-time record (Datassential, 2025), and that figure rewrites the arithmetic of any opening: the territory is no longer empty and waiting. When an operator signs a five-year contract and then tries to make the unit economics fit inside it, he commits the heaviest asset in the operation against an assumption he never verified. The correct order is boring, which is precisely why almost nobody respects it: territory prefeasibility, unit economics on paper with a capped food cost, and only then the signature. With sector net margin running between 3% and 9% according to Statista, a location error cannot be repaired with better service or a more creative menu. It gets repaired by closing, and closing costs the deposit, the buildout and the owner's standing with the bank.

2. The under-500K band: one unit, one format, no soft debt

Below 500,000 USD in annual revenue the decision is singular and admits no nuance: one unit, one format, break-even reached before month nine. An operator in this band has no cushion for experiments, because a full-service restaurant runs between 3% and 5% net margin per Peppr POS's Restaurant Profit Margin Guide 2025, which means a four-point drift in food cost swallows the entire year. My threshold here is hard: food cost per dish at 32% as a MAXIMUM, never as a target, and rent under 8% of conservatively projected sales. Payroll, rent and utilities never get loaded onto the plate — they live in the break-even calculation. Anyone who cannot recite from memory how many covers a day he needs in order to stop losing money is not ready to sign anything yet. This is the most dangerous band of all, because the operation already bills enough to look healthy and still has no systems to prove it.

3. Between 500K and 1 million: where the owner stops cooking and starts measuring

Between 500,000 and one million dollars a year, the right call is to invest in measurement before expansion: blind weekly inventory, food cost variance under 1.5 points between theoretical and actual, prime cost below 65%. Latin America's fast food market moved 61.49 billion USD in 2025 and heads toward 94.98 billion by 2034 according to Market Data Forecast, so growth is real; the trouble is that owners in this band routinely confuse growth with cash. I got this wrong for years, recommending a second location while the first still produced no predictable flow. The rule I use today: twelve consecutive months of documented positive EBITDA, or unit two stays off the table. Crossing one million dollars unlocks the replication conversation, and it unlocks it on one condition: the manual exists and somebody other than the owner has already executed it. The International Franchise Association counted more than 20,000 new franchise units in the United States during 2025, a 2.5% rise to 851,000 total, alongside 210,000 new jobs that pushed the sector past nine million workers.

4. Above 1 million: replicate only what has already proven itself three times over

Those numbers describe a system replicating processes, not individual talent. An operator in this band should fix a cash threshold: reserves equal to six months of fixed costs before opening unit two, plus contribution margin per dish audited in the flagship across three consecutive quarters. A well-run fast casual sits between 6% and 9% net margin according to Peppr POS. If the first one never reaches six, the second one won't either. Past 5 million in revenue two distinct profiles appear, and both deserve naming without euphemism: the large-format themed venue — two hundred covers or more, scenographic production, high ticket — and the project backed by a media chef or a celebrity lending a name. Both share the same trap: year-one sales are inflated by curiosity, and the model gets evaluated against that spike. Today's restaurant market behaves in a K shape, with the top 250 chains growing sales 3% while the next 250 fall 6.2% (Technomic Top 500, 2025).

5. Above 5 million: the large-format spectacle and the celebrity-signature risk

Translated to an opening at this scale: the brand alone will not hold volume. The threshold I demand is a projected 35% decline between quarter one and quarter five, with break-even computed on the depressed level rather than on opening night. Beyond 10 million the owner stops opening restaurants and starts running a portfolio, which completely changes the question that must be answered before every opening. It is no longer whether the unit works, but which unit in the group funds which other, and for how long. Public chain targets illustrate the scale of that thinking: Chipotle aims at 7,000 restaurants across North America (Restaurant Dive, 2025), Wingstop declares 10,000 locations worldwide, and Raising Cane's sets 1,600 by decade's end. A group in this band needs three measurable things: contribution per unit reported monthly, an automatic closure threshold — eighteen months below target margin means it closes, with no sentimental debate — and a committee that approves capex against expected return.

6. Above 10 million: group or chain, where the decision turns into portfolio work

The high-end conversation belongs here; the faith-based one does not. A financeable project clears three gates before spending the first dollar of construction, and each one produces a document a credit officer can actually read. First comes territory prefeasibility: density, direct competitors within an eight-minute walk, and verified — not assumed — spending capacity. Second comes unit economics on paper, with food cost capped at 32% from menu engineering onward and break-even expressed in daily covers. Third comes the thirteen-week cash model, the instrument that reveals whether the business survives month four. At Masterestaurant we work with those three gates because they turn an opening file into verifiable productive infrastructure, with waste traceability under SDG target 12.3 and indicators that MSME banking can monitor. Diego F. Parra presses one point hard: without all three, credit for a new opening remains a lottery. Invert the order and the scenario unfolds with almost mechanical predictability.

7. What happens if the operator inverts the order and signs first?

The operator signs a five-year lease on a corner he liked, funds the buildout, opens with a menu designed around the kitchen he already built, and discovers in month three that the real average ticket sits twenty percent below projection.

He cuts portions, raises prices, loses frequency. By month eight prime cost passes 70% and payroll becomes the adjustment variable. Month fourteen arrives, the doors close, and the formal youth employment he created vanishes along with the asset — squarely SDG 8 territory, where restaurant openings rank among the fastest vehicles for first jobs in Latin America. The debt, by contrast, survives. With delivery-only concepts running between 10% and 30% net margin according to Peppr POS, an exit does exist; but it gets designed up front, not during the agony. The first difference is SEQUENCE. A bankable project validates territory and unit economics on paper before signing a lease; the project of faith commits the heaviest asset —a five-year lease— and then tries to fit the model inside it.

8. The four differences an investment committee actually decides on

With the American market stacking more than 860,000 locations at a historic record (Datassential, 2025), a territory error no longer gets corrected through better service: it gets corrected by closing. Second comes the CEILING. Food cost per dish caps at 32% under the Masterestaurant framework, and that ceiling is fixed during menu engineering before the menu goes to print, not at the third month's accounting close. Payroll, rent and utilities never load onto the dish: they live inside break-even, which is where the operator learns how many daily covers stand between the business and a loss. Third is REPLICABILITY, and the market already delivered its verdict here. Technomic measured a K-shaped market through 2025: the top 250 chains grew sales 3% while the next 250 fell 6.2%. What separates those two groups is not product, it is the manual — one runs on documented procedure and the other on whatever judgment the shift manager brings that night.

9. The four differences an investment committee actually decides on — in practice

The fourth difference is INTERPRETATION, and it is what makes this a public policy matter. A well-protocoled opening generates measurable formal employment —American franchising added 210,000 jobs in 2025, a 2.4% rise (International Franchise Association, 2025)— while an improvised one destroys capital and employment at once. For multilateral banking, financing protocol instead of financing enthusiasm is the cheapest risk mitigation available.