What is a dark kitchen: the BEFORE and the AFTER, and the five alternatives nobody puts on the table

What a dark kitchen actually is, minus the packaging?

A dark kitchen is a production kitchen with no dining room, no storefront and no walk-in customer, whose only sales channel is delivery apps plus, if the owner works at it, a direct ordering channel.

What follows from that design is not logistical, it is financial: every peso coming in passes through an intermediary charging 20% to 30% across Latin America, while New York capped commissions permanently at 15% per delivery plus 5% for other services (Restaurant Business 2023) and San Francisco held them at 15% starting in 2020 (Restaurant Dive 2020). Statista counted more than 20,000 ghost kitchen locations operating in the United States during 2023, and Rappi reported 35 million active users and 150 million downloads as of August 2024. The format exists, it moves volume, and it still guarantees nothing about margin. That distinction is the one almost nobody draws before signing a lease. The dark kitchen falls short the moment your net profit depends on a commission you do not negotiate.

When the original option falls short on you?

One number in your own books gives it away: if more than 70% of your orders are born inside an app, your business stopped being a restaurant and became an outsourced supplier to that app.

Take 1,400 monthly orders, a 38,000-peso average ticket and 3.1% net profit — the case that reached me last month — and a three-point commission increase wipes out the entire business, leaving the owner working for free. That increase does arrive: between 2020 and 2023 the platforms subsidized rates with investment capital, and by 2026 they no longer do. Add that the National Restaurant Association measured in 2025 that nearly 75% of restaurant traffic is off-premise, so demand is not the issue; what you lack is control over the price of reaching it. Launching a second brand inside the kitchen you already rent wins in eight of every ten cases I review, and it excites nobody who showed up wanting a big project.

Option 1 · A virtual brand inside the kitchen you already pay for

Entry cost runs from 2 to 6 million pesos — photography, product sheets, packaging, platform onboarding — the learning curve takes two weeks, and break-even lands between 90 and 140 monthly orders because rent, extraction and payroll are already covered by the main business. Who it suits: the owner whose kitchen sits idle from 2 p.m. to 6 p.m. and wants to test broasted chicken, bowls or wings without signing a new lease. The real risk lives in executing simultaneous peaks, not in the investment, which is exactly why I treat it as a reversible experiment rather than a bet. Renting an equipped module inside a third-party hub hands you extraction, grease traps and permits already solved for 2.5 to 5 million pesos a month, without one peso of build-out. The learning curve runs about three weeks, essentially learning the hub's rhythm and its pickup windows.

Option 2 · A host kitchen or rented module inside a hub

The profile that profits here is the operator with a brand already proven elsewhere and documented demand in a new zone, not the one still inventing a concept. With an average delivery ticket of USD 20 to 35 in the United States per Lightspeed 2025, and roughly USD 24 per online order in Spain per Ken Research 2025, a module sustains itself above some 700 monthly orders once platform commission is stripped out. Below that figure you are paying rent to work for somebody else. I got this wrong for years: I used to tell owners to look at cost per square meter first, when the variable that decides a delivery business is WHERE the order is born. DoorDash measured in 2024 that 46% of American diners prefer third-party apps and order nearly five times a month, which leaves half the market available to whoever owns direct demand. A properly worked Google Business Profile, twenty fresh reviews per quarter, WhatsApp with a menu and your own checkout cost under 3 million pesos a year and never pay a commission.

Option 3 · Win back your own channel before moving a single brick

The dark kitchen, by design, gives that engine up: no map pin, no storefront, no dining-room reviews. In the Masterestaurant method this option comes first because it changes the cost structure without touching operations, and because one point of commission saved beats two points of new sales. Sharing a certified kitchen in time blocks turns a fixed cost into a variable one, which is precisely what a business with uneven demand needs. You pay by hour or by shift, between 30,000 and 90,000 pesos depending on city and equipment, and the same space serves you for corporate catering, a frozen line or event orders, where the ticket rises and platform commission disappears. Who it suits: the entrepreneur with a strong recipe and low volume, or the seasonal restaurant that only needs extra capacity in December. The limit is obvious and deserves saying plainly: there is no permanent storage, no identity of place, and if your demand grows steadily you will end up paying more per hour than you would for your own lease.

Option 4 · Shared kitchen by shift and volume catering

It is a bridge, not a destination, and as a bridge it beats almost every other option listed here. Licensing your brand to third-party kitchens flips the arithmetic completely: you stop paying rent and commission, and start collecting royalties of 4% to 8% on the licensee's sales. Entry cost is an operations manual, training and quality control, between 15 and 40 million pesos done properly, with a long curve of six to twelve months. It serves one very specific profile: the owner with a brand that already bills, with written processes and a dish that travels well. The U.S. QSR market moved US$ 289.68 billion in 2024 according to Business Research Insights, and growth at that scale did not come from operators owning every location, but from brands that learned to delegate production. The trade-off is harsh: you stake your reputation on a kitchen you do not control, and one food-safety incident in Cali damages the brand in Bogotá.

When NOT to change anything, said honestly?

Stay put if your dining room still produces commission-free orders and your kitchen has no idle capacity.

The owner about to sign that industrial-zone dark kitchen was saving four million pesos of rent a month while losing the 600 table orders that held up his margin, so the move cost him money while looking like financial discipline. Run four questions before you shift anything: what share of your orders is born outside an app, how many hours a day your kitchen sits cold, what margin survives commission on the worst dish of your menu, and whether you could hold the operation with commission three points higher. If all four answers calm you down, your format is not the problem. This week, pull the commission you paid over the last ninety days and divide it by your net profit for the same period: that ratio tells you who is running your business.

The five alternatives, without romance

ALTERNATIVE 1 · A virtual brand inside your current kitchen. Entry cost of USD 500 to 1,500, a two-week learning curve, and break-even between 90 and 140 monthly orders because the fixed cost is already paid. Who it suits: the owner whose kitchen sits idle from 2 to 6 p.m. and wants to test a second concept — fried chicken, bowls, wings — without signing anything. Verdict: it wins in eight of every ten cases I review, and it is the option that disappoints anyone who arrived wanting a big project. ALTERNATIVE 2 · Host kitchen, a module rented inside someone else's hub. You lease an equipped station with ventilation, grease traps and permits already solved, for USD 650 to 1,300 a month and no build-out capital. The curve is short, roughly three weeks, and it fits the owner who wants presence in ANOTHER part of the city to cut delivery times without committing capital.

The five alternatives, without romance — in practice

Verdict: excellent for testing geography, poor for building an asset — you never capitalize the build-out and the operator can raise the fee. ALTERNATIVE 3 · Reduced dining room with owned delivery. Instead of killing the room, shrink it to 20 or 30 seats and build a direct channel through WhatsApp Business, your own site and either salaried drivers or an outsourced fleet. Investment of USD 2,000 to 5,000, curve of three to six months since routing, thermal packaging and order handling all have to be learned. Who it suits: brands with strong reviews and a consolidated Maps listing, where customers already search by name. Verdict: the only alternative that structurally lowers commission, which makes it the most profitable at 24 months. ALTERNATIVE 4 · Stay put and fix the app channel. Zero bricks, USD 250 to 750 for professional photography, rewritten listings, channel pricing and geo-targeted in-app advertising.

The five alternatives, without romance — key points

One-month curve. Who it suits: every owner who has not done the basic work yet, and there are many — dark photos, two-line descriptions, a menu copied from the dining room with margins gutted by commission. Verdict: the best return per dollar invested, and the one almost nobody exhausts before jumping to a dark kitchen. ALTERNATIVE 5 · License or franchise your brand to an existing kitchen. You bring the recipe, the standard and the name; the operator brings kitchen, staff and working capital; fees run 5% to 8% of sales. Long curve, six to twelve months, because it demands manuals, quality control and remote auditing. Who it suits: brands with two or more years of stable operation and written processes. Verdict: it scales without your capital, yet hands experience control to a third party, and in delivery the experience IS the packaging and the punctuality.