Standardize before you scale: the before vs after that decides your 2026 expansion

Investment committees stopped asking about revenue and started asking about variance

The measurable signal of 2026 is food cost variance across units, and that number sets valuation before revenue does. A group with two locations and twelve points of deviation between them is worth less than one with three locations and two points, even if the first sells more, because the second proved the result does not depend on who happens to be in the kitchen that night. Scale explains the pressure: franchised restaurants in Spain account for 390 brands and 7,967 establishments (Tormo Franquicias Consulting 2024), and none of those brands grew with portions decided by eye. What to do by size: if you run one location, measure the deviation between shifts, morning against night, for eight weeks; if you run three or more, publish variance per unit in the monthly committee with the name of whoever owns it next to each figure. What is not published never gets corrected.

The manual is no longer the deliverable: the deliverable is an auditor with a first and last name

A manual with no assigned auditor and no biweekly cadence turns into corporate decoration in less than a quarter. I have watched groups pay expensive consultancies for hundred-and-twenty-page binders nobody opened again, and the diagnosis was always identical: the document existed, the consequence did not. What moves the result is one specific person weighing the technical sheet against the actual plate twice a month, with that review touching the head chef's bonus rather than a courtesy report. Sector scale forces the point: U.S. franchises project close to 8.9 million jobs for 2026, with 150,000 new positions and 1.8% growth (FRANdata / IFA, Franchising Economic Outlook 2026), and that volume of people rotating only holds up under live auditing. With one location, the auditor is you on a fixed calendar; with four or more, the role is full time and cannot belong to whoever cooks.

Territorial prefeasibility comes before signing the lease

The order of decisions weighs more than their quality, and signing a five-year contract before knowing whether the model replicates inverts the method. Territorial prefeasibility costs under 1% of expansion CapEx and prevents 100% of the irreversible mistake, because bad operations get fixed in six months and a badly signed lease never does. I got this wrong for years: I believed the second unit was chosen on a hunch about foot traffic, until I reviewed the pattern behind the closures and every one of them shared the same sequence, contract first, model afterward. With more than 830,000 franchised establishments in the United States (International Franchise Association, Franchising Economic Outlook 2026), the market already proved location is a calculation. For one or two locations, demand three sales scenarios and a break-even per zone before negotiating; for groups, make prefeasibility a committee requirement, with no exception for a good opportunity.

The costed recipe is the asset, not the menu

No serious fund signs in 2026 without a technical sheet costed plate by plate, with yield, waste and an alternate supplier documented. The reason is arithmetic: when the recipe lives in the chef's head, every resignation erases margin, and food cost per plate, which in the Masterestaurant method carries 32% as a MAXIMUM rather than a target, drifts without anyone knowing which week it started. Diego F. Parra insists on separating what loads onto the plate from what loads onto the break-even point, because pushing payroll and rent into recipe cost produces inflated prices and blind decisions. Labor pressure is real: the U.S. restaurant sector employs 15.9 million people after adding 200,000 positions (National Restaurant Association 2025), and that turnover punishes whoever failed to document. Start with the ten dishes that drive 60% of your sales and cost them this week with a scale, not from memory.

Auditable unit economics per location: the metric that replaced consolidated EBITDA

Consolidated numbers hide things, and 2026 capital stopped buying them. A group can show positive EBITDA while two of its five units burn cash, so the committee that once looked at the total now asks for P&L per location, prime cost per location and contribution margin per location, closed by the fifteenth of the following month. Transparency per unit separates the operator from the amateur investor, because it forces you to admit the flagship subsidizes the rest. In Brazil, food service employs 4.9 million people, 7.9% of the country's formal employment (ABRASEL 2025); that weight turned the sector into an object of serious financial scrutiny rather than a business of intuition. Action by size: with fewer than three locations, close each unit as if you were selling it separately; with more, audit that central-cost allocation is traceable and not a percentage invented to dress up the weak unit.

Small format and narrow operations: less menu, more replicability

Whoever cuts scales better. The trend that consolidated is trimming the menu and narrowing the format so a unit can be staffed with fewer trained people, and Dutch Bros took it to the extreme with 90% of its revenue depending on the drive-thru (QSR Magazine), an operation so narrow that training fits into days. Here comes the tension almost nobody resolves: the owner wants a wide menu because it feels like pride and looks like more sales, while every extra reference multiplies inventory, waste and execution variants. The bridge is accounting, cut the references that fail to reach 2% of sales and measure food cost ninety days later. If you run one location, start by removing four dishes this month; if you are preparing a second unit, freeze the menu six months before opening and leave it alone until the new kitchen replicates the standard without calling you.

What to adopt now and what to merely watch?

Adopt three things now and skip the debate: a costed technical sheet for the dishes that drive 60% of sales, a financial close per unit on a fixed date, and a named auditor reviewing the standard every fifteen days.

That gets implemented in a quarter and is exactly what a committee wants before looking at your projection. Watch, without investing yet, kitchen automation, kiosks and dynamic pricing models: they work once the process is already stable, and over an unstable process they only accelerate the error. Watch the sector's hiring pace too, since it sets the competition for talent: franchises added 210,000 positions in 2025, up 2.4% (IFA Economic Outlook 2025), and franchised QSR passed 4 million jobs with a 2.6% rise (International Franchise Association 2025). The decision criterion is simple: standard first, then the technology that multiplies it. Franchising does not repair a model that is not standardized; it publishes it.

The overrated trend: franchising as a shortcut to expansion

Selling units to fund growth seduces because it looks like money without CapEx, and the fine print says something else: the franchisee buys your manual, your training and your supply chain, and if those three pieces do not exist, you sold an expectation you will have to sustain with your own hours for years. In Spain, franchised restaurants group 269 brands billing more than 5.8 billion euros (Asociación Española de la Franquicia, La Franquicia en España 2024), with 92,109 direct jobs representing 24% of the system's employment (Tormo Franquicias Consulting 2024): those are the figures of a mature model that demands real support. What happens if you franchise with twelve points of food cost variance? The franchisee replicates your disorder, blames the brand, reputation drops in reviews and the next candidate asks for a discount. Standardize first. The difference is not the manual: it is who audits it.

What truly separates the two scenarios?

A hundred-and-twenty-page document with no assigned auditor and no biweekly cadence turns into corporate decoration in under a quarter, and groups pay expensive consultancies for binders nobody opens.

What changes the outcome is a named person checking the technical sheet against the actual plate, with that review carrying consequences for the head chef's bonus. Decision order matters more than decision quality. Signing a lease and then wondering whether the model replicates inverts the method: territorial prefeasibility costs under 1% of expansion CapEx and prevents 100% of the irreversible error, because a badly signed five-year contract cannot be fixed with better operations. I got this wrong for years, pushing clients into good sites with models that were not yet sealed. Standardizing does not mean uniforming the menu. It means fixing the SKELETON —costing, critical processes, service standards, indicators— and leaving room in the local menu, because a restaurant franchise that freezes 100% of its offer loses relevance in markets moving at another speed.

What truly separates the two scenarios — in practice?

The tension between control and adaptation resolves with one rule: standardize what drives cost and risk, release what drives taste. Per-location unit economics are the language of capital in 2026.

A group average hides the bleeding unit, and any decent analyst finds it in the second due diligence meeting; showing the disaggregated table from day one buys credibility and usually moves valuation more than any polished deck.