How to start a dark kitchen from scratch: the 40,000 USD mistake and the five alternatives that actually balance the books

Forty-one orders a day before you earn a cent

Above 3,500 monthly orders the buildout holds; below it, nothing does, and the mistake starts when the math gets run in money instead of orders, which is the unit this business actually runs on. A 45-square-meter hidden kitchen in Bogotá, Mexico City or Medellín carries 4,200 to 7,800 USD a month in fixed structure (rent, three cooks, utilities, software, licenses) before the first plate leaves the pass. Put a 9 USD average ticket against a 38% contribution margin after platform commission and the arithmetic asks for 1,230 to 2,280 orders a month just to break even. Translate that into the cook's calendar: forty-one orders a day, every day, rainy Tuesdays included, when the app sends nothing and the rent charges you all the same.

Commission never touches food cost: it takes the selling price

A dish costed at 30% on the menu operates at an effective 41% once the platform keeps 27%, and no amount of buying chicken better closes that eleven-point gap, because commission bites the selling price BEFORE the money reaches your account, not the raw material cost. Rappi, iFood, Didi Food and Uber Eats take 18% to 30% of order value depending on city, category and who supplies the courier. Ghost kitchens die right there, in the app's settlement, not on the hot line. And the platforms will not bend, because the size of the business backs them: the platform-to-consumer segment moved USD 96,864 million in 2024 according to Statista, while Delivery Hero reported €12.8 billion in segment revenue that same year, up 22% on the prior one.

When building from scratch falls short for you?

Month three is when the signal arrives, quietly: the virtual brand has no traffic, the algorithm refuses to surface it, and the bank account registers it before you do.

A virtual restaurant lives entirely off the app's ranking, which rewards volume, rating and prep time, none of which exists in week one; either you buy the first 60 days with in-app advertising, an extra 8% to 15% of sales, or you accept the empty ones. Stack those points on a 27% commission: your 38% contribution margin sinks below 25% in the exact quarter when cash runs thinnest and the lease is still fresh. A projection missing that line from day one is not a projection, it is a wish in spreadsheet form.

Option 1: start inside the kitchen you already run

Three hundred to nine hundred dollars is what it costs to launch a virtual brand inside the kitchen you already pay for: packaging, menu photography, platform onboarding, and the investment ends there. Since the dine-in business already carries the fixed structure, break-even drops from 1,230 monthly orders to under 300, ten a day, a volume any working kitchen ships between three and six in the afternoon without a single new hire. The downside deserves saying in full: if the brand lands, it will collide with the dining room at peak, and you will have to choose whom to fail, because one cook cannot work two tickets at once. This route suits an operator who owns the venue and runs kitchen occupancy under 60%.

Option 2: shared kitchen by the hour or by station

A station in a shared kitchen runs 600 to 1,800 USD a month depending on city and shift, against 4,200 to 7,800 for a structure of your own, and the saving matters least. What changes is the NATURE of the cost: delivery demand swings up to 40% between the first and the third week of the month across Latin American markets, so paying by the hour lets you spend little in the slow stretch, while a buildout charges the same on the 8th as on the 25th. Working against you is everything you stop controlling, from the cleaning to the hood schedule to the neighbor frying fish beside your dessert. It fits the founder without assets who wants to test two or three concepts before committing to one.

Option 3: host kitchen under a co-packing contract

Hiring an established restaurant to produce your menu costs a fee of 12% to 20% of sales, and it wipes out rent, payroll and equipment in one stroke. You supply brand, technical sheets and quality control; the host supplies a kitchen already fired up. It is the least used route and the one I recommend most when the recipe book is closed, though the switching cost, near zero in money, comes due in management: skip the weekly audit of portion weights and margin evaporates into generous plating. Diego F. Parra brought it into the Masterestaurant methodology for a cash reason: fixed turns into pure variable, and pure variable is what keeps a new brand alive when the algorithm punishes it two months running. It asks for standardized recipes and no operating team of your own.

The cost of leaving weighs more than the cost of entering

The figure that decides is the EXIT, not the entry, and for years I told owners the opposite: look at the initial investment. Closing a dark kitchen with a buildout means paying the penalty on the remaining term, dismantling the hood and dumping used equipment that the secondary market returns at 30% to 40% of the original invoice; 35,000 USD in equipment comes back as 12,000 when a buyer shows up. Without construction, the close fits in an email and thirty days' notice. Take the counterfactual all the way: at month eight your brand still sits under 900 monthly orders and you decide to shut it. A buildout costs you the sunk capital plus the penalty; with a shared kitchen you lose the current month and the recipe walks out with you.

When NOT to switch and keep the buildout?

One scenario makes building from scratch the right call, and hiding it would be dishonest:

you already sell more than 3,500 orders a month, you negotiated commission under 22% on volume, and at least 35% of sales move through your own channel, web, WhatsApp or phone, where nobody takes a cut. At that volume a 7,800 USD fixed structure dilutes to 2.2 USD per order, while a shared kitchen would keep charging a variable that never drops again. The mature market confirms it: China holds more than 3,200 ghost kitchen facilities according to Coherent Market Insights, and they went up where the volume already existed, not the reverse. Before you sign five years, work out your break-even in daily orders and sit with that number on a rainy Tuesday.

Four differences that decide the cash

Turning a variable cost into a fixed one sounds like corporate maturity and in delivery it is a trap, because sales move and rent does not: across Latin American markets orders fall close to 40% between the first week of the month and the third, once payday money is gone. Pay by the hour and your spending follows that curve; sign for construction and the dead week costs you the same as the busy one. Look at it from the end rather than the beginning: everyone pays the entry, only the one who got it wrong pays the exit. With construction, closing means a contractual penalty, hood removal and used equipment the secondary market buys back at 30% or 40% of invoice. Without it, a single email a month ahead ends the matter.

Four differences that decide the cash — in practice

Format governs food cost: buying for a single brand at 400 orders a month puts you on retail prices and pushes the real number toward 32%, which the Masterestaurant costing contract treats as the ceiling and never as the goal. Split that same input across three virtual brands on one line and it falls to 27%. Scale here is not vanity, it is margin. Diego F. Parra insists on something almost nobody models: selling through an app does not retire the printed menu. When the dark kitchen is born inside a restaurant with a dining room, print still rules the table, because it sets the pace of service and holds up suggestive selling, while the QR menu handles delivery, analytics and price moves that need no reprint. The Masterestaurant verdict is BOTH, each in its role, never QR alone.