Standardizing to Grow: the fat-manual myth and what actually replicates in a restaurant

Step 1: list the decisions that move cash, not the tasks people can see

Start by writing down the 12 to 18 operating decisions that hit food cost, waste and service time directly, and leave everything else out. In that Bogotá group I mentioned, the whole list fit on one sheet: protein portion weight, grill temperature and timing, yield per cut, purchase order by supplier, counting hour, discard criteria. What you must produce here is a numbered list with an owner and a frequency for each decision, and the test is brutal: delete decision number 7 and if nobody in the restaurant notices a difference in the monthly cash, that decision does not belong in the manual. The US industry bills 1.5 trillion dollars across more than one million establishments according to the National Restaurant Association, and six out of ten still close before year five; almost never because of the logo. A plate spec sheet works when it carries grams, minutes, degrees and unit cost; if it says «to taste» or «as needed», you just wrote a poem.

Step 2: turn every decision into a spec sheet with numbers, not adjectives

Draft each one in half a page: input, exact gram weight, expected waste as a percentage, yield per purchase unit, cost per portion and target selling price, with a food cost ceiling of 32% as a MAXIMUM you tolerate, never as a goal. Those 18 extra grams of protein the third location gave away on every plate cost 4.1 points of food cost, and no leadership course brings them back. The deliverable is a costed recipe book signed by the chef and by finance; you verify it by weighing three random portions of the same dish during the dinner shift and comparing against the sheet, accepting ±5 grams of deviation. A replicable operations manual sitting in a shared folder changes nothing; the standard holds when the workstation itself makes mistakes hard. Buy thirty-dollar scales per station, numbered portion ladles, containers marked with a maximum level, and hang the laminated sheet at the cook's eye level.

Step 3: fix the standard in the furniture, not on paper

Domino's added 776 net stores in fiscal 2025 according to its published results, and that pace does not survive on individual judgement: it survives on utensils that allow only one quantity. I got this wrong for years, pushing training when the problem sat in the furniture. The deliverable is a photo of the standard setup for each station, posted right there; you verify it with an opening round where every station matches its photo, with no exceptions negotiated because of a cook's seniority. Save a full page for the question that sinks young groups: what does the shift do when something happens that nobody wrote down. The answer has to be an explicit rule, along the lines of «if the shortage affects less than 10% of shift sales, the head cook substitutes following this replacement hierarchy and reports it; above that, call the operations director». That page is the difference between a replicable business and a good restaurant where the owner solves things over WhatsApp at eight at night.

Step 4: write the judgement rule for everything you will NOT document

US franchising projects close to 8.9 million jobs for 2026 according to FRANdata and the IFA, and none of those chains runs because the founder answers the phone. The deliverable is the written substitution hierarchy; you verify it with a stopwatch: the decision must take under 40 seconds with no phone call. Do not open the second location until the first shows three consecutive monthly closings with the same cost structure and a food cost swing under one point month to month. If location one leaves you a 9% operating margin, location two will not leave 15% because you have a manual: it will leave 9% minus the learning curve, which in my experience eats six to ten months. The entry figures are public: a restaurant franchise fee usually runs from 10,000 to 50,000 dollars according to Toast, with McDonald's at 45,000 and Subway between 15,000 and 25,000 dollars.

Step 5: measure before replicating, using unit economics from location one

That number is the toll, not the business. What you produce here is a P&L covering three comparable months per location; you verify it when two different people rebuild it from the same POS and land on the same margin. The most common failure is not documenting too little, it is documenting everything: three hundred procedures nobody memorizes give you a cook who stops thinking and starts hunting for the page, and service time blows up. The second mistake is brand theater, starting with uniforms and menu design because those photograph well while inventory counting waits. The third and priciest is auditing with affection: if the supervisor announces the visit, you are not measuring the standard, you are measuring how well the team prepares for a visit. In Spanish franchised foodservice, 92,109 direct jobs represent 24% of the system's employment according to Tormo Franquicias, people executing written procedures daily.

The mistakes that sink standardization: excess, theater and friendly audits

Avoid all three like this: a hard cap of 18 spec sheets, one unannounced audit per week, and a retirement rule forcing you to remove a procedure every time a new one comes in. Before you sign the lease on the second site, run the test Diego F. Parra uses in Masterestaurant audits: send a cook from another shift, with no briefing, to run a full station using only the manual. If the plate comes out within spec, if the gram weight holds and if service time stays within two minutes of standard, the document works; if the cook asks three questions, the document is expensive decoration. Consider the counterfactual: you open anyway, skipping that test, and location two starts running on the new head cook's judgement instead of yours, which means it replicates a personality rather than a margin, and eight months later you own two different businesses behind the same storefront.

Step 6: the blind kitchen test, run by a cook who has never worked there

The deliverable is the signed test record with every deviation noted. You will know standardization is done when you can tick six concrete boxes, not when the place feels orderly. One: the 12 to 18 spec sheets exist, costed and signed. Two: three random portions per dish land within ±5 grams. Three: the judgement rule solves a shortage in under 40 seconds without calling the owner. Four: last quarter's food cost moved less than one point between months. Five: the blind cook test passed with no questions. Six: one procedure left for every one that entered. The market rewards whoever has this ready, with Brazilian foodservice growing near 7% a year through 2028 according to ABRASEL and US franchised output rising 8.5% in the Southwest according to the IFA. Your next move is to weigh three plates tonight and write down the deviation. The difference between a replicable business and a good restaurant is who decides when an ingredient runs out.

Where the promise to replicate breaks?

In a replicable business the answer is written and takes forty seconds; in a good restaurant the owner answers by WhatsApp at eight in the evening, and that cannot be cloned or purchased.

Standardizing to grow usually fails through excess, not scarcity. A repeatable operations manual that tries to cover every scenario creates paralysis: the cook stops thinking and starts hunting for the page. I would rather have 18 closed decisions plus one explicit judgment rule for everything else than 300 procedures nobody memorizes. Unit economics outrank design. If your first location returns a 9% operating margin, the second will not deliver 15% because it has a manual: it will deliver 9% minus ramp-up errors, maybe 5% through the first half-year. Standardization improves the CONSISTENCY of a model, it does not change the model. This is exactly where groups burn expansion CapEx chasing profitability their menu never had.

Where the promise to replicate breaks — in practice?

There is a real tension few owners resolve out loud: standardizing does kill part of the original location's magic. True. The answer is not denial, it is deciding WHICH part of that magic is functional and which is folklore.

Bread coming out hot at seven in the evening is functional and gets standardized with a timer; a server remembering a regular's name is valuable folklore, you train it, you do not manualize it. According to Danny Meyer, founder of Union Square Hospitality Group, hospitality is what happens when service runs correctly and the guest also feels someone is on their side; his public thesis holds that process gets standardized while warmth gets hired. That line is precisely the one a group must draw before signing its first franchising agreement.