How to open a restaurant step by step: the classic route, its limits and four honest alternatives

When the classic route falls short

The classic opening route falls short at the exact moment your CapEx crosses 400,000 USD without the MTIE —months to break-even— dropping below twelve, and the number that gives it away sits not in the business plan but in the heat map for «restaurant near me» across your 1.5 to 3 km radius. A site on an avenue with 14,000 vehicles a day and zero local searches for your category is an expensive, empty site, because in 2026 foot traffic no longer rules: what rules is the digital discovery radius Google Maps resolves at 12:40 p.m. The industry employs 15.9 million people in the United States per the National Restaurant Association 2025, and 2.1 million directly in Mexico per CANIRAC 2024; demand is abundant. What is scarce is GEOLOCATED demand on your particular block.

The dark kitchen: cheap speed paid for in commission

A dark kitchen hands you speed for 25,000 to 60,000 USD in CapEx against the 120,000 to 400,000 USD of a full dining room, and that 70 to 80 % discount gets billed back by the marketplace as commissions of 18 to 30 % on every single order, month after month, with no expiry date. Who it suits: an operator with a proven recipe, an average ticket above 12 USD and a category not yet saturated in the app in that city. Switching cost runs about sixty days between the shared-kitchen contract, health registration and menu photography. Here is the arithmetic trap almost nobody runs: a 200,000 USD site that hits break-even in month 8 outperforms a 25,000 USD hidden kitchen stuck at 60 % kitchen occupancy, because CapEx amortizes once and commission gets paid forever. The number that matters is not the investment. It is the MTIE.

A franchise buys demand that already exists

Buying a franchise means buying pre-existing demand, and that is the asset almost nobody prices properly on the spreadsheet: when someone types the brand by name into Maps, that query already carries settled purchase intent and you skip the 45 to 90 days Google takes to establish relevance for a brand-new business. Entry prices swing brutally by banner: a 15,000 USD initial fee at Subway against 90,000 USD at Dunkin' per GrowthFactor's FDD 2025-2026 analysis, and 45,000 USD at McDonald's per the 2024 FDD reported by Franchise Chatter. Who it suits: the group leader chasing a second or third location without rebuilding brand equity from scratch. The U.S. franchise system projects close to 8.9 million jobs in 2026, up 1.8 %, per FRANdata and IFA. You are renting the demand machine. You do not own it.

Drive-thru and the single-channel dependency

Building a drive-thru instead of a full dining room shrinks leased square footage and lifts turnover, yet it chains you to one channel with a force worth measuring before signing: Chick-fil-A books 60 % of its sales through the window and Dutch Bros reaches 90 % of revenue, per QSR Magazine. That concentration works beautifully as long as the traffic pattern holds, and turns into a bomb the day the city flips a one-way street or opens a nine-month roadworks on the corner. Who it suits: a coffee, beverage or handheld-food operator with owned land or a long lease on a stable traffic artery. Switching cost: high, because the lane's civil works cannot be undone. And here comes the uncomfortable question I suggest asking before the contract: if 60 % of your sales vanish tomorrow through a traffic decision you do not control, how many months of cash do you have left?

Guest kitchen: proving the concept under 15,000 USD

The cheapest alternative of all is not opening yet: running as a guest kitchen inside a restaurant that already bills, during the hours its dining room sits dead, on an investment that rarely exceeds 10,000 to 15,000 USD across smallwares, permit and menu. You pay variable rent on sales, typically 12 to 20 %, and you buy the one thing no business plan gives you: demand measured in your own neighborhood before committing 300,000 USD and a five-year lease. Who it suits: a founder with no prior operation, or a group validating a second brand without diluting the first. Switching cost is near zero, and that is precisely its virtue, because closing a pilot that failed costs one conversation while closing a site signed for five years costs the deposit, the build-out and the reputation. Three months of real data beat any feasibility study.

The right order: the digital engine fires before the build-out

The sequence almost every manual teaches —idea, plan, financing, site, licenses, build-out, staff, opening— carries a serious design flaw: it puts digital discovery at the end, as if it were launch marketing, when Google takes 45 to 90 days to establish relevance for a new business. I got this wrong for years, because I recommended locking down operations first and the digital channel second, and that misplaced honesty gets paid in months one through three with food cost under control, service dialed in and a half-empty room. Diego F. Parra and the Masterestaurant team flipped the order in the method: the Google Business Profile listing should exist in «coming soon» status from the week the lease gets signed, with build-out photos, a defined category and declared hours. Firing the engine during construction gifts you ninety days of indexing you would otherwise burn with payroll running. Sixty percent of restaurants that close do so with the kitchen running well, and that fact ought to reshuffle your investment priorities more than any menu advice.

Why the ones who cook well still close?

They do not die from the food.

They die from unit economics that never closed —rent too high for the average ticket, CapEx financed over 24 months when the real return lands in month 36— or from leaning on a single demand channel that an algorithm switched off without notice. The arithmetic is stubborn: at 30 % food cost and 12 % rent on sales, an 18 USD ticket needs roughly 190 covers a day to carry a 22 % payroll, and if your digital discovery radius only reaches 40 daily searches for your category, those covers do not exist. Spanish franchised food service generates 92,109 direct jobs, 24 % of the system's employment, per Tormo Franquicias Consulting 2024. The operator who measures survives, not the one who cooks best. Stay with the classic route without hesitation when three conditions hold at once: your own CapEx of 120,000 to 400,000 USD with no expensive debt, twelve months of operating cash, and a 1.5 to 3 km radius where the «restaurant near me» search for your category shows real volume against competitors with weak reviews.

When NOT to switch routes?

In that scenario, your own dining room buys CONTROL, which is what no alternative sells: you decide the menu, the price, the music and whether suggestive selling happens tableside with a printed card.

The dark kitchen delivers speed and charges dependency. The franchise delivers demand and charges sovereignty. The drive-thru delivers turnover and charges concentration. None of the four returns the margin a well-sited owned location earns with food cost under 32 %. If you have the three conditions, opening is the right call and everything else is noise. If even one is missing, start with a guest kitchen this quarter. CapEx is not the number that matters; months to break-even is. A 200,000 USD venue that breaks even at month 8 beats a 25,000 USD dark kitchen that never clears 60 % kitchen utilization because its category is saturated inside the app. The classic route buys CONTROL.

Five differences that decide which route fits you

You set the menu, the price, the music and whether suggestive selling happens with a printed menu on the table. The dark kitchen buys SPEED and pays with dependence: 18 to 30 % commission on every order and a delivery radius the marketplace adjusts whenever it wants. A franchise sells something few people price correctly on paper: pre-existing demand. When someone searches the brand by name in Maps, that query already carries resolved purchase intent. Brand traffic outweighs the 6 % royalty during the first eighteen months, and stops outweighing it around year four, once you know how to operate. Takeover is the only route handing you a live digital asset. A profile with 340 reviews and 4.3 stars carries an age signal Google uses, and rebuilding that from scratch costs 14 to 20 months of clean trading. Due diligence changes nature, though: it stops checking zoning and starts checking labor liabilities and the inherited lease.

Five differences that decide which route fits you — in practice

The hybrid resolves the tension that breaks the other models. A dark kitchen builds no brand because nobody sees the venue; a classic restaurant pays rent on square meters that only fill on Friday and Saturday. A 45 m² counter with twenty-two seats and a kitchen sized for 180 daily delivery orders uses the same square meter twice: physical presence feeding reviews, digital volume paying rent.