What is a dark kitchen: the mistakes that make it expensive and the method that makes it pay

What exactly is a dark kitchen, and what does your business stop being when you open one?

A dark kitchen is a production kitchen with no dining room, no servers and no storefront, where every sale arrives through delivery aggregators, which means you stop being a restaurant and become a SUPPLIER for somebody else's channel.

The name misleads because it sounds like technology, when the decision is purely about channel: the app sets visibility, listing order, promotions and much of the final price, not you. Best for operations born fully digital, with no dining room to sacrifice.

If you already have a dining room and 15% to 35% of sales come from apps, a virtual brand beats a dark kitchen

Build a VIRTUAL BRAND inside the kitchen that already pays rent: it is the better option for an owner with an active dining room, and the arithmetic holds without decoration. A Medellín operator showed me his plan in March, closing 22 tables to move into a kitchen park; he billed USD 41,000 monthly with USD 12,300 arriving through Rappi, and his math celebrated the rent savings while ignoring that the remaining 70%, the portion paying 0% commission, lived precisely in those tables. We shifted that 70% to delivery at 26% average commission and projected operating profit fell from USD 4,900 to USD 1,100 per month. He opened nothing. A second brand in the same kitchen shares rent, hood, gas and line cook, so every incremental order arrives with fixed cost already covered.

When NOT to choose the popular option: three scenarios where a dark kitchen destroys margin?

Rule out the dark kitchen if your delivery polygon is already saturated, if your average ticket cannot absorb the commission, or if your product travels badly.

First, the profitable radius runs 3 to 5 kilometers and orders inside it are finite by category: with 40 burger brands competing, yours creates no demand, it splits what exists. Second, an USD 8 ticket carrying 26% commission leaves USD 5.92 to cover ingredients, packaging and labor, and a 32% food cost simply will not fit there. Third, pasta, delicate fried items and ice cream lose quality across 18 minutes on a motorbike, and that loss shows up as 3.4-star ratings that sink your placement. In those three cases, your better move is strengthening the dining room and treating apps as a complementary channel, never as the backbone.

Red flags when comparing kitchen parks and operator contracts

Four signals in this trade predict a bad contract, and recognizing them before signing saves years. First: variable rent tied to a percentage of sales with NO cap, because the operator collects his cut even when you lose money. Second: category exclusivity that does not really belong to you, meaning the park bans you from competing while renting the neighboring module to another brand serving the same dish. Third: when the operator holds the aggregator account and you cannot reach the merchant dashboard, you lose order-by-polygon data, which is the only genuine asset this model produces. And fourth: shared hoods and extraction without independent gas metering, a line item weighing between 2% and 4% of sales in Latin American kitchens. Best for whoever can negotiate fixed rent plus dedicated extraction capacity.

A dark kitchen does not cut costs: it moves them to another line, and the move has a price

Front-of-house payroll leaves and aggregator commission arrives, so the savings you celebrate on the expense sheet reappear, more expensive, on the net sales line. A restaurant with 40% of billing at the table carries a blended effective commission far healthier than an operation depending entirely on apps, and that comparison almost never happens before the move. Run the full counterfactual: if tomorrow the app raises its commission plan, something that has already happened across several cities in the region, a balanced operation loses margin while the fully digital one loses far more and enters closing territory. Mexico's delivery market added USD 9.22 billion in 2024, growing 14.66% annually according to Statista, and that growth hands aggregators pricing power you do not have.

The asset each model buys you: a customer who returns, or a position you rent

With a dining room you build a customer who comes back because the treatment was good; with a dark kitchen you rent a POSITION inside the app's search engine, month after month, and the difference in equity is brutal. The aggregators' internal engine ranks by proximity, preparation time, rating and money spent on ads, exactly like Google Maps, so your visibility evaporates the day you stop paying. I got this wrong for years: I recommended judging the model by its low opening cost, seductive and deceptive, when the deciding variable is geolocated DEMAND for your category inside the radius. From Masterestaurant, Diego F. Parra frames it plainly in channel audits: whoever lacks the stomach to compete in a ranking that shifts on its own should stay out. Best for brands with sustained local ad budgets.

How to measure your polygon before signing, using data you already have for free?

Open the app's merchant dashboard and measure how many monthly orders your category moves within a 3 to 5 kilometer radius: that number decides more than any spreadsheet projection.

Count the brands already serving that dish inside the polygon, divide orders among them and compare the result against your kitchen break-even; if the average per brand fails to cover fixed cost, no recipe will rescue you. Asia-Pacific shows where this is heading, at USD 21.73 billion in 2024 projected toward USD 60.59 billion by 2032, a 12.8% annual pace according to Coherent Market Insights, and that rising volume also brings fresh competitors every quarter. With a dining room running and a virtual brand layered on top, you test the same polygon risking the cost of different packaging, not the cost of a full relocation.

The model's paradox and the decision worth making this week

The cheapest model to open is the most expensive one to sustain, and that tension settles the question for most owners. A dining room costs when you open and returns clientele; a ghost kitchen costs when you sell and returns dependency. The virtual restaurant and ghost kitchen market stood near USD 65.3 billion in 2023 according to Next Move Strategy Consulting, proof the format works for operators who choose it with numbers in hand. This week, pull ninety days of sales by channel and calculate your blended effective commission: if it lands below 12%, your dining room is worth more than you think.

The differences that decide whether it makes money

A dark kitchen does not lower your costs: it MOVES them. Floor payroll leaves, aggregator commission arrives, charged on every ticket depending on plan and city. A restaurant with 40% dine-in carries a far healthier blended commission than a delivery-only operation, and almost nobody runs that number before moving. The asset of a restaurant with a dining room is the guest who returns because of how she was treated; the asset of a dark kitchen is POSITION inside the app's search. The first is defended with hospitality, the second is rented monthly with geotargeted ads and rating. If you lack the stomach for a ranking that shifts on its own, stay out. A virtual brand inside your current kitchen shares rent, hood, walk-ins, staff and purchasing; a new dark kitchen duplicates every one of those fixed costs.

The differences that decide whether it makes money — in practice

So the correct sequence is virtual brand first, own kitchen later — the opposite of what gets sold at foodtech expos. A delivery menu is not the dine-in menu photographed again. Dishes that survive 25 minutes in a thermal bag are different ones: braises, rice, short pasta, bowls, breads assembled last. If your signature plate dies in transit, a dark kitchen wrecks your reputation faster than it builds cash. In the dining room you own the customer record; on the aggregator the customer belongs to the platform. Any serious dark kitchen plan includes a migration path to direct ordering, with a worked Google Business Profile and your own WhatsApp or web channel, because a direct order carries roughly twice the margin of a commissioned one.