How to open a restaurant step by step: the mistakes that burn capital vs the right method

Step 1: model ONE dish before you look at a single site

The first deliverable of an opening is not a lease or a menu: it is one sheet with the true cost of a dish and its contribution margin, and it is done when you can say, without hesitating, how much cash each unit leaves at the price you plan to charge. The math runs like this: ingredient cost including waste, divided by the pre-tax selling price, and that ratio must land at 28% as the target, with 32% as an absolute ceiling nobody recommends anymore. Payroll, rent and utilities do NOT belong here — they get paid later, out of the break-even point, and loading them onto the plate is the accounting mistake that sinks half the budgets I review at Masterestaurant. Verification is simple: if your signature dish leaves less than 68 cents of gross margin per dollar sold, the concept does not yet exist as a business.

Step 2: move from the dish to the shift and stress the mix

A profitable dish will not rescue a badly built shift, so the second step multiplies: expected covers times average check, distributed by the sales mix you actually project rather than the one you would like. The deliverable is a full-shift table showing gross sales, cost of goods and margin before fixed expenses. Here the founder's favourite trap shows up: the highest-margin plate is usually the slowest seller, while the item carrying volume tends to run 34-35% food cost, which drags the weighted average upward. Fix that with menu engineering before you open, moving prices and menu positions until the weighted food cost of the shift closes under 30%. If it will not close on paper, it will not close with real guests sitting in front of you either. Your maximum rent comes out of projected sales, never the other way around: a healthy project commits no more than 8-10% of monthly revenue to the lease, and that percentage stops being an opinion once the shift is modelled.

How much rent can I afford?

Multiply shift margin by the operable shifts in a month, subtract payroll and utilities, and whatever survives defines the rent ceiling your business tolerates without eating the profit.

Sign first and rent becomes a constant, forcing every other line of the plan to stretch until it balances — that is where the classic month-four price hike comes from, the one that scares off the clientele you just earned. Diego F. Parra hammers this at the start of every Masterestaurant project: the order of the steps is not method, it is 70% of the investment risk. Deliverable: one hard number you carry into the negotiation as a limit. Break-even is the exact monthly revenue that covers every fixed cost, and you get it by dividing those fixed costs by the contribution margin percentage you already calculated.

Step 3: compute break-even before the signature

At 70% margin and 21,000 dollars of monthly fixed costs, you must sell 30,000 a month to neither lose nor earn; translated into operations that is roughly 1,000 a day, which converts into concrete covers and shifts you can test against the foot traffic on the corner you are eyeing. The figure is verified when it fits inside the physical capacity of the room: tables times turns times check. If break-even demands filling the dining room twice every day of the week, the site is too big for the model, not the model too small for the site. Cut square metres before you cut wages. Initial investment has climbed faster than average check, and that gap devours copied budgets: commercial kitchen construction costs have grown at double digits since 2021 according to the Associated General Contractors of America indices, so a plan built on 2019 figures reaches the till with a hole near 30% before the first plate goes out.

Step 4: budget the build with 2026 prices, not 2019 ones

Get three current quotes for hood, refrigeration and hot line, then add 15% contingency for hidden work — wiring, drainage, permits that surface once the wall is already open. The deliverable is a signed budget dated this year with a named supplier per line item. Quick check: if any line in your file came from a downloaded template, that is not a budget, it is a wish formatted as a table. Before signing anything, set aside working capital for six full months of fixed costs, because a new location rarely matures on the schedule the pitch promised. At 21,000 dollars of monthly fixed costs that means 126,000 sitting idle, and the money is not a luxury: it is what separates negotiating with suppliers from begging them. Then run the model through a scenario of 25% lower sales for four consecutive months and watch what breaks. If under that drop you still cover rent, payroll and goods without draining the whole cushion, the project is financeable; if you run out of air by month three, shrink the size of the bet.

Step 5: secure the six-month cushion and stress the model

The tension here is real and worth naming: a conservative model kills good concepts, and founder enthusiasm rescues concepts the spreadsheet rejected. We settle it by giving the model veto power over SIZE, never over the concept. Some 43% of independent restaurants close before their third birthday according to the National Restaurant Association, and the autopsy points almost always to the same four places. First, signing the lease because it looked like an opportunity, then modelling afterwards. Second, budgeting construction with no contingency, so working capital ends up paying for the kitchen. Third, loading payroll and rent into plate cost, which inflates prices artificially and prices the room out of its own street. Fourth, opening with a long menu: every extra reference adds inventory, waste and station complexity, and during an opening waste spikes precisely when the team has not mastered portioning. One vaccine works against all four — put the number on paper before a legal obligation exists that forces you to defend the wrong number.

Closing checklist: how to know it all landed

You have finished the preparation phase when you can show six documents and none of them makes you uneasy: the dish spec with food cost under 30%, the shift table with its weighted mix, the rent ceiling stated as one figure, break-even translated into daily covers, the investment budget quoted this year with 15% contingency, and the statement proving six months of fixed costs are available. Add the 25% downside stress test resolved in writing. If all six exist and the stress test holds, sign the lease; if even one is missing, not yet. Franchising formalises this exact order and charges for it — 45,000 USD initial fee at McDonald's per the 2024 FDD reported by Franchise Chatter, against 15,000 USD at Subway per Upwise Capital. You can build that rigour without paying a fee; what you cannot do is skip it. The gap is not food quality: it is WHICH decision comes first.

Where the outcome is actually decided?

Sign the lease before modeling the business and rent becomes a constant while everything else turns into a variable that must stretch to fit;

model first and rent becomes a variable negotiated against a known ceiling, and in a healthy project that ceiling sits at 8-10% of projected sales. There is a genuine tension in this trade worth naming: conservative financial models kill good concepts, and founder enthusiasm rescues concepts the spreadsheet rejected. Both statements hold. We resolve it this way — the model holds veto power over the SIZE of the bet, never over the concept. If the numbers do not support 200 square meters on the avenue, they support 60 on the parallel street, and the concept survives with half the risk. Restaurant requirements (licenses, health permits, zoning) get treated as paperwork when they are really a measurable source of delay: every month of postponed opening consumes rent, core payroll and financing cost without a single dollar of revenue.

Where the outcome is actually decided — in practice?

Two months of delay on a mid-size project burns 6-9% of total investment before the first plate sells. Site due diligence is not the broker's walkthrough.

It is traffic counted by you, in three bands, across two weeks; it is asking for the previous tenant's electricity bill; it is confirming the exhaust duct can reach the roof without a fight with the building association. Each of those checks costs days and saves tens of thousands. And here is the point that stings investors for restaurants most: a group's profitability is not built in the first unit, it is built in the third, when overhead spreads. But you only reach the third if the first generated its own cash. Scaling on a unit that does not generate cash multiplies a negative number.