Scaling a restaurant: the numbers that decide, and the ones that break groups

How much real margin does a restaurant leave before you open the second one?

Between 3% and 5% of sales in full service, and between 6% and 9% in limited service: that is the average net margin estimated by the National Restaurant Association, and with that figure in hand the decision to scale changes nature entirely.

At 4% net profit, a two-point drift in food cost inside the new location does not trim earnings, it wipes them out and leaves the unit working for free all quarter. That is why the threshold I defend before signing any second lease is prime cost at or below 60% of sales and operating contribution margin of 15% or better, measured across SIX consecutive months rather than during the best quarter of the year. One good month proves nothing; six straight months prove the margin belongs to the model and not to the season, the weather or a campaign that happened to land.

MTIE: what each location keeps after paying for everything it uses

At Masterestaurant we measure scaling with a starting indicator we call MTIE, Margin of Work Independent from Structure, and it answers a blunt question: how much stays in that location after paying absolutely everything that location consumes, without head office lending it a single peso. Diego F. Parra insists on calculating it per unit and never on the consolidated P&L, because the consolidated statement is the finest hiding place a sick unit will ever find. A three-location group in Bogotá billed 41% more than the previous year with 18% less cash in the bank; the founder was convinced somebody was stealing. Nobody was. He had opened two locations with the operating cash of the first, without its own reserve, and each opening ate four to seven months of cash flow while the sales curve matured. The constraint that stalls scaling is almost never money: it is repeatability of margin.

Repeatable margin, not available capital

A location delivering 18% EBITDA on the founder's charisma does not constitute a scalable model, only a very well paid job, and the proof arrives on opening day. When the founder splits in two, the mother unit usually loses 3 to 6 points of margin, and the drop gets blamed on the season because the consolidated P&L covers it with the new location's revenue. Here is my criterion, and it stings: before opening, pull the founder out of daily operations for thirty days and measure what happens to prime cost. If it climbs more than two points, the model is not documented, it is memorized by one person, and that knowledge does not replicate with capital. A badly costed recipe card gets fixed in fourteen days; a five-year lease at 16% of sales never gets fixed, and that asymmetry turns the site negotiation into the most irreversible decision of the whole expansion.

The lease condemns the unit from the day you sign

The ceiling I defend is 8% to 10% of PROJECTED sales in the base case, never in the optimistic case, which is exactly where nearly everybody runs the numbers because the broker shows Friday traffic and not a Tuesday in February. Run the counterfactual all the way through: if your base sales come in 20% under projection, which is routine in year one, a 10% rent becomes 12.5% and still breathes; a 16% rent jumps to 20% and swallows the entire contribution margin before the first payroll is paid. A well-located new site takes four to seven months to reach steady-state sales, and that period must be financed with RESERVED cash, never with the operating cash of the mother unit. Evidence from the large chains confirms the right sequence: Chick-fil-A added 179 net locations to reach 2,863 units in 2025, against 132 net the prior year according to QSR Magazine, and Wingstop opened 278 net restaurants between 2024 and 2025 according to the QSR 50.

The maturation curve nobody budgets for

Neither accelerated by improvising; they armored the unit-level model first and stepped on the gas afterwards. The small group that copies the pace without copying the reserve ends up funding the opening with the cash flow of the location that works, and the problem travels backwards: the healthy unit starts stretching supplier payments, loses its early-payment discount and raises food cost without anyone touching a recipe. Franchising moves enormous numbers and that is precisely why it seduces people too early. The International Franchise Association counted 821,000 franchised establishments in the United States during 2024, up 1.9%, and together with FRANdata projects 845,000 units for 2026; in Spain, the Asociación Española de la Franquicia recorded 269 restaurant brands billing more than 5.8 billion euros in 2024, while Tormo Franquicias counted 390 brands and 7,967 establishments. Those figures describe a mature market, not an escape route for a weak margin.

Franchising is not scaling: it is selling a system that already works

Franchising demands an operations manual, a closed recipe card, per-dish costing and a margin that survives being run by someone who is not you. If your second owned location still depends on you tasting the sauce every morning, selling the third to a stranger merely relocates your problem to another city with your name on the door. Benchmarks only earn their keep once you translate them into three concrete scenarios. Small single-unit restaurant: the target is prime cost under 60% and a reserve covering six months of fixed costs before you even look at a second site; without that reserve, the expansion conversation is premature. Mid-sized group of two to four locations: each unit gets measured separately with positive MTIE and rent inside 8-10% of base sales, and the weakest unit sets the pace for the whole group, not the average.

How to read these numbers in YOUR operation?

Large group of five or more:

central structure appears, and here the hard rule is that corporate cannot consume more than 4% to 6% of consolidated sales, because every extra point comes straight out of the 3% to 5% net margin the National Restaurant Association estimated. Pick your scenario and run the math on July's numbers, not December's. It is worth saying where the figures come from and how far they reach. The 3% to 5% and 6% to 9% margins come from National Restaurant Association sector estimates for the US market; the franchised unit counts, 821,000 in 2024 and 845,000 projected for 2026, belong to the International Franchise Association with FRANdata; the Spanish data comes from the AEF 2024 report and from Tormo Franquicias Consulting, which disagree with each other because they count different universes, and that gap of 269 versus 390 brands is informative on its own.

Where these benchmarks come from and what they cannot tell you?

None of those sources measures your kitchen. A national average knows nothing about your rent, your table turns or your menu mix, so treat them as a ceiling of expectation and as a warning light, never as a forecast.

The only number that decides your opening is yours, measured six months running. The difference is not available capital, it is margin repeatability. A location delivering 18% EBITDA because of the founder's charisma is not a scalable model, it is a well-paid job: split that founder in two and the flagship usually gives back three to six margin points that nobody blames on the opening, because the consolidated P&L hides it. Rent is the least reversible decision on the list. A badly costed recipe gets fixed in a fortnight; a five-year lease at 16% of sales condemns the unit from the signature onward. That is why the ceiling I defend is 8-10% of PROJECTED sales in the base scenario, never the optimistic one, which is exactly where most operators run the math.

Where scaling actually breaks?

The maturation curve gets ignored with remarkable consistency. A well-sited new location typically reaches stabilized sales between month four and month nine, and it burns cash through that stretch even when operations are flawless.

An operator who skipped that reserve does not have a sales problem, he has a treasury problem that surfaces precisely as sales begin to climb. And one paradox deserves resolving before anything gets signed: the moment the flagship performs best is also the moment the founder has the least time to prepare the second site. The fix runs against instinct — hire the second-in-command WHILE the room is full and cash is healthy, painful as that salary feels, because hiring once two locations are burning means hiring a firefighter rather than a director.