Franchising: the numbers before and after the Masterestaurant method

Sector size does not validate your model

Franchising inside an expanding sector does not make a business replicable when it isn't yet. The International Franchise Association projects franchised output of 921.4 billion dollars in the United States for 2026, up from 907.3 billion the previous year (International Franchise Association / FRANdata, Franchising Economic Outlook 2026), and that figure gets misused almost every time: the group pastes it onto slide three of the investor deck as if the tide lifted every boat equally. It does not. That projected 1.6% growth describes the aggregate of thousands of networks, including those closing units while opening others, and a new network without proven unit economics doesn't share in that average, it dilutes it. The concrete decision coming out of the number is simple: use it to size the appetite of capital, never as evidence that your model works. Before signing the first franchise, your own unit has to return the investment in under 36 months with EBITDA held above 12% by month 18.

How much unit EBITDA do you need before selling the first franchise?

That is the floor, and there is arithmetic behind it:

the National Restaurant Association places prime cost for a multi-unit operator between 55% and 65% of sales, so whatever room sits under that line has to absorb rent, utilities, royalty and still leave a return for the franchisee. A group running at 6% to 9% of unit EBITDA — the typical range of what reaches my desk before the Masterestaurant method — has nothing to pay a 5% royalty from without pushing the loss onto whoever bought in. Measure the last 24 closed months of each location separately, never the consolidated figure. A three-unit group in Mexico City arrived in January 2026 with an investor's letter of intent on the table and every intention of signing ten franchises within eighteen months. I asked for the P&L location by location across the last twenty-four months before saying a word, and the usual thing surfaced: the flagship returned 17% EBITDA and carried the other two, sitting at 3% and −1%.

The consolidated average hides the unit carrying the group

Franchising that system would have meant selling the average of a lie. On those numbers, the network would replicate two sick units for every healthy one, and the franchisor's royalty is calculated on sales rather than profit, so the franchisor would have collected all the same while the franchisee bled. The rule that came out of it: no unit enters the manual with EBITDA below 12%. Mature networks end up operated almost entirely by franchisees, and that proportion is a consequence rather than a starting point. Roughly 95% of McDonald's restaurants worldwide are franchisee-operated (McDonald's, Franchising Overview 2025), and that architecture arrived after decades of company units proving the mechanics. Domino's plans net expansion of 1,100 stores a year through 2028, 85% of it international, reaching some 26,200 (Quartr, 2025); McDonald's projects more than 8,000 new restaurants by 2027 to approach 50,000 (QSR Magazine, 2025).

Who actually operates the networks that did work?

Nobody gets there by franchising first.

Chipotle, by contrast, closed 2024 with just 85 international locations — 55 in Canadá, 27 in Europe, 3 in the Middle East — according to Restaurant Dive, and that deliberate slowness cost it share while sparing it a brand disaster. There is a tension almost nobody resolves out loud: the franchisor earns on the opening, because the initial fee lands whole and fast, while the franchisee only earns on operations, which are slow and measured in years. That misalignment explains why so many networks grow for two years and collapse in the third, when the earliest units start closing and fees from new ones no longer offset lost royalties. The bridge exists and it is BORING: royalty on sales with an audited operating floor, which forces the franchisor to make the unit work instead of merely opening it. If your contract doesn't penalize the franchisor when a unit falls below the floor, you designed a fee-selling machine.

The misalignment between whoever opens and whoever operates

Write that clause with the auditor in the room, not with the lawyer alone. These benchmarks read differently by size, and here are the three scenarios. A small operator with one or two units shouldn't be looking at the sector's 936.4 billion figure yet (IFA, 2025): their number is prime cost, and if it sits above the 65% the National Restaurant Association marks, franchising is premature, full stop. A mid-sized group of three to six locations applies the 36-month payback cut to the WORST location rather than the best, because the average franchisee will buy an average location. A large group, seven units and up, already plays in the logic of Domino's and its 1,100 net annual openings: what gets measured there is how many units clear 12% EBITDA by month 18, and that number sets the pace of expansion. Counting openings is the metric of a fee salesman; counting units above 12% EBITDA at month 18 is the metric of whoever collects royalties for ten years.

Openings versus healthy units: what counts as success

One example shows the gap: a group with twelve openings and four healthy units is worth less than one with six openings and six healthy ones, even though the first figure photographs better in a deck. Aggressive sector plans feed the confusion, since Firehouse Subs announced more than 500 restaurants in Brazil over the coming decade (The Brasilians, Franchising in Brazil 2025) and everyone reads the headline without asking how many survive year five. Spanish franchising illustrates the reverse: Portugal concentrates 176 networks with 2,632 establishments (AEF, 2025), density built slowly. Change your board's scoreboard this week. These figures come from public sector sources — International Franchise Association, FRANdata, National Restaurant Association, Restaurant Dive, QSR Magazine, AEF — crossed with the consulting reading we apply at Masterestaurant to expansion models. They are neither a sample nor a primary study: they are published data, read by someone who has stood inside the kitchen and inside the boardroom.

Where these benchmarks come from and what they can't tell you?

Their limits matter as much as their value.

The 936.4 billion in franchised output the IFA reports for 2025 aggregates dozens of industries besides restaurants, prime cost of 55% to 65% shifts with format and geography, and none of these sources publishes closure rates by cohort, which would be the genuinely useful number. Treat them as an order-of-magnitude reference, and your own P&L as the truth. Sequence. Almost everyone drafts the contract first and the unit economics later, so the contract ends up protecting a model that does not exist. We invert it: prove the unit returns its investment in under 36 months, then write the contract that protects that mechanic. A watertight contract over a bad model accelerates litigation instead of preventing it. What counts as success. The traditional franchisor counts OPENINGS, because the initial fee is the visible revenue. We count units above 12% EBITDA at month 18, the only thing that sustains a royalty stream over ten years.

The four differences that move the number

A group with twelve openings and four healthy units is worth less than one with six openings and six healthy units, however much better the first figure looks in an investor deck. Who carries the ramp risk. In the old model the franchisee puts up the capital and learns alone through the six months when most money burns. In ours the franchisor budgets USD 9,500 to 12,000 a year of support per unit and places a house operator inside the kitchen for the first eight weeks, precisely when the next three years of food cost get decided. The candidate filter. Accepting an investor with no operating partner is the decision that has killed most networks, and for years I defended it myself on the argument that capital rules. I was wrong. A restaurant is not run from a quarterly board meeting, and today the 18-point due diligence rejects the profile that brings only money and a calendar.