Dark kitchen: before vs after with Masterestaurant

What a ghost kitchen is: a precise definition?

A ghost kitchen —also known as a dark kitchen, cloud kitchen, or virtual kitchen— is a food production facility with no public-facing dining area that operates exclusively for delivery orders through third-party apps.

It has no dining room, no waitstaff, and no need for a commercial storefront on a high-traffic street. Its only showcase is a profile inside the delivery aggregator. Diego F. Parra, from Masterestaurant, has audited many of these operations across Latin America and summarizes the key distinction this way: a traditional restaurant sells both a dining experience and food; a ghost kitchen sells only the product and the logistics. That shift moves costs away from rent and front-of-house payroll toward the aggregator's commission per order, and toward digital ranking within the platform. Without understanding that cost transfer, the model appears cheaper than it actually is.

What a ghost kitchen is NOT?

A ghost kitchen is not simply a restaurant that also offers delivery. The distinction is operational and financial. A hybrid restaurant keeps the dining room, absorbs the cost of prime-location rent, and treats delivery as an additional channel with compressed margins.

A ghost kitchen eliminates the dining room from day one: the average footprint shrinks from 120 m²—typical for a neighborhood restaurant—to 35 m², with a single production line optimized for speed. It is also not an industrial catering kitchen: catering produces large volumes for one-off events, while a ghost kitchen handles individual tickets, with peaks around lunch and dinner service. And it is not a new concept: the term reached Latin America in 2019, and by 2026 a growing share of new restaurant registrations in Bogotá, Mexico City, and Lima no longer open a physical point of sale, confirming the model has moved from experiment to standard practice.

The four essential components of the model

Every viable ghost kitchen relies on four components that must work together: production space, management technology, delivery logistics, and digital positioning. The space must meet health code requirements—exhaust hoods, epoxy floors, certified hot water—even if it is only 25 m², because inspectors do not grade on a curve for small kitchens. Technology means a POS linked to a KDS (Kitchen Display System) and to the aggregator platforms so orders flow in without friction; losing an order to a disconnection carries a real cost per ticket and, worse, drives down the acceptance rate in the algorithm. Logistics depend on the aggregator's delivery fleet or a proprietary one; route time must stay under 25 minutes or product quality collapses. Digital positioning—ratings, product photography, response time—is the equivalent of the storefront that no longer exists.

The cost structure that changes everything

Opening a traditional restaurant in Latin America once required a large sum in buildout, furniture, and front-of-house staffing. With the ghost kitchen model, that investment drops several times over. The savings are real, but they do not flow to the operator's pocket: they get redistributed. Prime-location rent—which could represent a meaningful share of revenue—disappears, but the aggregator's commission appears in its place, taking a real cut of every ticket. If the average ticket is below $9 USD, the operation is mathematically unviable before accounting for food cost, payroll, and packaging. Diego F. That is the recurring mistake: confusing lower upfront investment with lower operational risk. The risk does not disappear; it changes shape.

The menu: from 40 dishes down to 10–16 items

A ghost kitchen's menu is an engineering tool, not a traditional restaurant card. A large multi-item menu—standard for a full-service restaurant—becomes a liability when every item must survive transit time without losing texture, temperature, or presentation. Masterestaurant recommends cutting down to a handful of proven items, selected against three criteria: food cost under control, short preparation time, and packaging resilience. A brothy rice dish does not travel well; a protein over compact rice does. A ceviche takes 3 minutes to assemble and arrives fresh if the delivery driver leaves within 5 minutes; an 18-minute lasagna destroys the algorithm's acceptance-time metric. Reducing the menu also cuts waste: with 12 well-calibrated items, inventory turns in 2–3 days compared to 5–7 days for an extended menu, which can lower raw-material costs by 8% to 14% per month.

Payroll, technology, and the real break-even

Payroll at a ghost kitchen can run much lower than at a traditional restaurant with comparable revenue, because front-of-house positions—host, servers, sommelier, dining-room cleaning staff—simply do not exist. What remains is the kitchen crew and an order coordinator. But that payroll reduction comes with proportional technology dependency: the POS, KDS, aggregator connectivity, and digital reputation system are now critical assets. An internet outage during peak hours can cost real money in lost orders and algorithmic penalties. That acceleration is real, but it is conditional on financial discipline from the first month of operation.

The most common mistake that destroys the model

Diego F. Parra has seen it across dozens of ghost kitchens in the region: the operator signs the lease excited by the low upfront investment, lists on two or three aggregators without modeling the cumulative commission impact, and 90 days later discovers solid sales volume but no margin left. The mistake has a name: failing to separate channel cost from product cost. If food cost and aggregator commission both run high at the same time, a large share of revenue is already committed before paying rent, payroll, packaging, and utilities. For the equation to work, food cost cannot exceed 32%—that is the hard ceiling in the Masterestaurant methodology, not a suggestion—and the average ticket must be above $12 USD if the commission exceeds 25%. Quality control is no longer the waiter's job at the table: it is the packaging, the temperature at container seal, and the route time. If any one of those three fails, the rating drops, the algorithm penalizes, and sponsored-order costs rise to recover lost visibility.

When opening a ghost kitchen makes sense in 2026?

The ghost kitchen model makes sense under specific conditions, not as a universal solution. First: the concept must travel well—proteins, bowls, sushi, pizza, individual desserts—and the average ticket must clear the platform commission by a healthy margin.

Second: the city must already have a large share of out-of-home food orders moving through delivery; in Bogotá, Mexico City, and Lima that threshold was already crossed in 2024. Third: the operator must have genuine digital management capacity—responding to reviews quickly, updating the menu to match real-time availability, keeping the acceptance rate high—or must hire someone who does. For an existing restaurant looking for a second channel without a second physical location, the ghost kitchen can be the right lever. For someone who has never run a kitchen and sees the model as a cheap entry into the business, the risk is high: without food-cost discipline and a break-even calculation from day one, the model burns through the same cash it was supposed to save.

The 5 differences that hit your margin hardest

A dark kitchen shifts premium-rent spending into aggregator commission, which can take a meaningful share of every order. Quality control no longer happens at the table: it happens in the packaging and the route time, which shouldn't exceed 25 minutes. The menu shrinks from 35-40 dishes down to a much shorter list of references optimized to travel without losing texture. Payroll drops substantially, but tech dependence on POS, aggregator and KDS rises proportionally. Break-even accelerates when food cost stays under control and the operation is sized to the ghost-kitchen format.