Ghost kitchen: the mistakes that burn cash vs the right method

Chapter 1 — Why does a ghost kitchen fail when orders keep coming in?

A profitable ghost kitchen is defined by its contribution margin per order, not by order volume. The root error I see over and over is running it like a physical restaurant without a dining room:

signing with two or three aggregators at 28-32% commission and leaving demand in the hands of the algorithm. With no dining room there is no ticket to subsidize the expensive order and no tip to cover service; every dollar passes through commission and packaging. I run the dirty math with the owner in the meeting: an $18 order with 38% food cost, 30% commission and $1.10 packaging leaves contribution margin near zero. That same order, served through your own channel at 30% food cost, leaves $4-$5 clean. A ghost kitchen does not go under from lack of demand; it goes under from its unit economics. Treat aggregator commission as a variable cost to manage channel by channel, never as a fixed business expense.

Chapter 2 — Aggregator commission is a variable cost, not a fixed figure

Here is where the money is: a single commission point on $40,000 in monthly sales is $400 a month going straight against margin. At Masterestaurant we break that 28-32% apart by platform and treat it as a lever. If you move 20% of orders from the aggregator to your own channel, you recover 6 to 8 commission points on that slice of sales. The operator burning cash looks at total orders in the Rappi panel and congratulates himself; the one who wins looks at how much contribution margin each channel leaves. The gap between 30% and 15% effective blended commission, on $40,000 a month, is $6,000 monthly, $72,000 a year that decides whether the virtual brand survives its second year or not. Local SEO and the Google Business Profile are what separate a cheap organic order from an expensive one bought with commission plus ad spend.

Chapter 3 — The local digital engine decides if the order arrives cheap or expensive

At restaurantecercademi we measure it this way: an organic order —someone who searched you on Google Maps, saw your 5★ reviews and ordered through your own link— costs almost zero to acquire. An aggregator order with geo-targeted ads on top can carry 30% commission plus $2-$4 in ad spend, up to 40% of the ticket in channel cost. The error's demand is hostage to the algorithm: if Uber Eats changes the ranking or DiDi raises commission, the cash register collapses that same week. The method builds an owned demand base that does not depend on delivery. Optimizing the Google listing, answering 100% of reviews and winning positions for «restaurant near me» is margin infrastructure, not decorative marketing. A healthy ghost kitchen must survive a 20% input jump without turning to loss, and that is tested before opening, not after.

Chapter 4 — Input stress scenarios: 5%, 12% and 20%

We model three stress scenarios on a 30% base food cost: at +5% food cost rises to 31.5% and margin holds; at +12% it hits 33.6% and already breaches the 32% ceiling we set at Masterestaurant as the per-dish maximum; at +20% it jumps to 36% and contribution margin per order evaporates if you keep 30% commission. The cash lesson is hard: if your only lever is raising the price on the aggregator, the algorithm punishes you with lower conversion. That is why the owned channel is not a luxury; it is the cushion that lets you absorb input stress without losing the customer. Every commission point recovered buys a point of tolerance to pantry inflation. The metric that governs a ghost kitchen is contribution margin per order, not the raw order count on the dashboard. I break it down in the board meeting with round numbers: an $18 ticket, minus 30% food cost ($5.40), minus $1.10 packaging, minus 30% commission ($5.40), leaves $6.10 of contribution before variable labor.

Chapter 5 — Contribution margin per order, not total orders

That same order at 15% commission through your own channel leaves $9.10, almost 50% more margin on an identical sale. Multiply by 1,200 monthly orders: the difference is $3,600 a month from channel alone. The error measures volume because volume feeds the ego; the method measures contribution because contribution pays the rent. An operator with 800 profitable orders earns more than one with 1,500 orders that leave zero margin. Count the margin, not the tickets. The trap of the ghost kitchen is that its low CapEx seduces the owner while its variable OpEx eats him alive month after month. Opening costs little: no dining room, no waiter, no dining design, the initial investment in a virtual brand can be a fraction of a traditional restaurant. But that is where the good news ends. Variable OpEx —28-32% commission, $0.90-$1.30 packaging per order, rising geo-targeted ad spend— scales with every order and does not fall over time, it rises.

Chapter 6 — Low CapEx, high variable OpEx: where the cash is lost

In dozens of operations I have seen it: the owner celebrates opening three virtual brands with little capital and six months later the register is in the red because no one managed the variable costs. Diego F. Parra sums it up simply: in the ghost kitchen the investment does not kill you, the operation bleeds you. Daily control of commission, packaging and ad spend is the difference between renting and giving away. The 90-day plan for a profitable ghost kitchen chases one measurable goal: cut blended commission from 30% to 18-20% and take the owned channel to 25-30% of sales. Days 1-30: optimize the Google Business Profile, activate your own order link and audit real food cost per dish against the 32% ceiling. Days 31-60: run a 5★ review campaign to win «restaurant near me», renegotiate commission by volume with the main aggregator and test cheaper packaging without hurting the experience.

Chapter 7 — 90-day roadmap with KPIs for the board

Days 61-90: report three hard KPIs to the board —contribution margin per order, share of sales through the owned channel and effective blended commission— with the goal of recovering 6-8 points. If by day 90 the owned channel does not reach 25% and blended commission stays above 28%, the virtual brand is not viable and it is closed before it drains more cash. The mistake treats aggregator commission as a fixed business cost; the method treats it as a variable cost to manage channel by channel. One commission point on $40,000 monthly sales is $400 a month straight off the margin: moving 20% of orders to the own channel recovers 6-8 points. The cash-burning approach measures orders; the method measures contribution margin per order. An $18 order with 38% food cost, 30% commission and $1.10 packaging leaves near-zero contribution margin. The same order with 30% food cost on the own channel leaves $4-$5.

Chapter 8 — The differences that decide the margin

In the mistake, demand is hostage to the delivery algorithm: if the aggregator changes ranking or raises commission, the cash collapses. In the method, local SEO and Google Business Profile generate an order base that doesn't depend on the aggregator's ad auction. The mistake scales into virtual brands to 'dilute fixed costs'; the method scales only when the base brand shows sustained positive contribution margin. Multiplying negative unit economics multiplies the loss, it doesn't hide it.