Ghost kitchens (dark kitchens): what they are, how they work and when they pay off

What is a ghost kitchen and how does it work inside?

A ghost kitchen runs like a restaurant kitchen without the other half of the restaurant. An order comes in through an app or the brand's own channel and lands on a tablet or an order management system;

the line cooks it from the standardized recipe, packs it and hands it to a platform courier or an in-house driver. The guest never sees the kitchen. They see a photo, a price, an estimated time and a rating. That's why the cost structure moves too. Prime-location rent, dining room furniture and front-of-house staff disappear; app commissions, packaging, in-app advertising and the cost of holding a good rank show up instead. Operators who understand that shift decide well. Those who only see the rent savings often learn too late that the commission weighs more than the rent they stopped paying. Ghost kitchen, dark kitchen and cloud kitchen describe the same thing: a production kitchen with no public-facing space that sells through delivery and pickup.

Ghost kitchen, dark kitchen, cloud kitchen, virtual restaurant: are they the same?

The difference is geography. Americans say «ghost kitchen»; «dark kitchen» is the British term, and «cloud kitchen» dominates in India. A virtual restaurant or virtual brand is not a synonym.

It names the brand that only exists in the apps, not the kitchen. One ghost kitchen can run five virtual brands, and a traditional restaurant can launch a virtual brand without becoming a ghost kitchen. Separating brand and kitchen makes the decision easier: validate the brand first, then decide where to cook it. Almost always for the same reason: the per-order math never worked, and volume hid the problem for a few months. The causes repeat. The app commission — 15% to 30% per order on DoorDash's and Uber Eats' U.S. plans — was treated as a minor expense instead of the business's second-largest cost. The menu was copied from the dining room, dishes didn't survive the trip, and ratings and rank fell.

Why did so many ghost kitchens close?

A multi-year lease was signed before demand was proven inside the delivery radius. And every sale depended on a single platform, with no direct channel for repeat guests.

The U.S. market shows it: IBISWorld measures an industry that shrank between 2020 and 2025, and Kitchen United said in November 2023 it would sell or close all of its physical kitchens. Where unit economics did work, the model kept growing. Ghost kitchen profitability is decided one order at a time. The formula is simple: app price minus commission, packaging, recipe cost and promotion. In an illustrative example, on a $12 ticket with a 25% commission, $0.60 of packaging, a 30% recipe cost and a 5% promotion, you keep $4.20 of contribution, 35% of the price. With $4,500 of monthly fixed costs, break-even is about 1,072 orders a month, or 36 a day.

The math that decides whether a ghost kitchen is profitable

Change one input and the answer changes: on a 30% plan contribution drops to $3.60 and you need 42 orders a day; with the same order coming through your own channel and a $2 delivery cost, contribution rises to $5.20. That's why costing with standardized recipes and spec sheets is the first job, not the last. Two points of food cost on every item is the difference between making money and subsidizing the app. Seven numbers are enough to run a ghost kitchen, and all of them are weekly, not monthly. Contribution per order, by channel and by brand, because one brand can quietly subsidize another. Actual versus theoretical food cost from your spec sheets, which exposes loose portions and waste. Daily orders versus break-even. Ticket time, because slow items slide down the app. Rating and the number of negative reviews per week. Cancelled or refunded orders, which the app deducts.

The weekly KPIs every ghost kitchen should track

And the share of sales through your own channel, the KPI that improves profitability most over time. A dashboard with those seven numbers, reviewed every Monday, prevents most closures: it lets you fix price, recipe or promotion before the problem reaches your bank account. Standalone ghost kitchen. You lease and build out a space for production only, running one brand or several. Full control of operations and costs, and the highest upfront investment: equipment, ventilation, health-code build-out and permits. It fits once demand is proven. Shared kitchen or commissary (ghost kitchen for rent). An operator rents out equipped stations inside one facility with shared services. It lowers upfront cost and speeds up opening, but adds monthly rent and the operator's rules: check the minimum term, what the rent includes and what happens to your business if the facility closes. Virtual brand inside an existing restaurant, also called a virtual restaurant.

The four ghost kitchen models, and how to choose

The kitchen that serves the dining room also produces a brand that only exists in delivery apps. It's the lowest-risk option — spare capacity, same crew, same suppliers — and its challenge is peak hours, when the dining room and the app compete for the same grill. Multi-brand kitchen run by an operator or a chain. Several brands ship from one kitchen to take up more shelf space in the app with the same overhead. It demands strict standardization — standardized recipes, spec sheets and ticket times per item — and control of the workload by hour.