Virtual restaurant business model: the numbers nobody shows you before you sign: with AI

A virtual brand's real margin starts where the aggregator's screen ends

A virtual brand delivers healthy margin when total channel commission stays below 26% of the ticket, and that threshold is no whim: it is what survives after subtracting food cost, packaging and promotional discounts from an operation whose kitchen was already paid for. The Bogotá case that opens this piece shows it without decoration: 1,180 orders, a ticket of 42,000 pesos, nearly 50 million billed, and a contribution margin of 3.1% after a 28% commission, packaging, the 20% discount the platform suggested to enter the featured carousel, and cancelled orders. Same oven, and the dining-room menu paid better. When you look at your dashboard, cover the gross sales box with your thumb and hunt first for the four lines the aggregator never puts on top: commission, promotion, packaging, cancellations.

Virtual brand, dark kitchen and kitchen operator are not the same arithmetic

Three models share the name "virtual restaurant" and their cost structures look nothing alike. The virtual brand lives only inside the apps and cooks where you already produce: it adds food cost, packaging and commission, nothing else. A dark kitchen is a production facility with no dining room, so rent and payroll enter the equation and get paid even when it rains and nobody orders. The kitchen operator rents you equipped square meters for a fixed fee plus a percentage, meaning you pay twice for the same order. That gap decides the outcome: at 400 orders a month, a virtual brand riding an existing kitchen can earn 1,900 dollars while the standalone dark kitchen loses 2,300, with identical menu and identical ticket. The global dark kitchen market is projected at 171.3 billion dollars by 2033 according to Global Growth Insights, though category size has never paid anyone's payroll.

Commission is not negotiated by store size, it is negotiated by account

The mistake I see repeated in this model is asking for a better rate while showing one location's volume. Aggregators set terms by exclusivity and by the consolidated volume of the ACCOUNT, not by the branch that calls. A group running four brands under a single tax ID earns points a standalone operator never sees, and those two or three points on a 42,000-peso ticket are worth more than any weekend campaign. The scale is real: DiDi Food Mexico reported close to 74,000 restaurants on its app, 70% of them local small businesses, and Swiggy declared 196,000 partner restaurants across 653 cities in its FY 2023-24 annual report. You are one among hundreds of thousands. So the honest argument is not launching a fifth brand; it is consolidating the four you already run under one account before sitting down to negotiate.

Packaging is a product cost, and saving there gets expensive fast

Packaging belongs to product cost, not logistics, and miscounting it turns a "30%" food cost into 35% without anyone noticing. Between 3 and 6 points of the ticket for well-solved packaging is normal for hot food travelling twenty minutes; below 3 points the dish arrives cold or tipped over, and then you pay the full refund plus a one-star review that drags your ranking for weeks. Run it on a 42,000-peso ticket: two points saved on packaging is 840 pesos, and a single refund eats the savings of fifty orders. My audit rule is blunt: if the container cannot survive twenty minutes inside a closed courier bag, it is not cheap, it is unfinished. Book packaging inside the dish's recipe card, right next to the ingredients, where it belongs.

How to read these numbers in YOUR operation: three scenarios?

Benchmarks do not apply the same way to everyone, so here are the three cuts I use.

If you are SMALL —one location, under 400 delivery orders a month— the only defensible version is a virtual brand on your current kitchen, with no added rent, and total commission under 26%; if the channel charges 30% and asks for a 20% discount, the math lost before the oven was lit. If you are MID-SIZED, between 400 and 1,200 monthly orders, a dark kitchen begins to make sense only when rent plus fixed payroll get split across two or three brands sharing mise en place. And if you are a GROUP with four locations or more, your lever is not opening kitchens: it is consolidating the tax ID, negotiating commission points on the account and using the square meters you already pay for during dead hours. That order matters: consolidate first, expand later.

An uncomfortable counterfactual: the platform raises you two points

Assume your virtual brand bills 50 million pesos a month at an 8% contribution margin, already better than the average I find on real dashboards, and the platform lifts commission two points through a policy change. That adjustment does not cost you two points of margin: it takes a quarter of the entire margin, because the two points come off the full ticket while your margin is calculated on what remains. Now add that commission is charged on the discounted price while the discount comes out of your pocket, and the real hit lands near three points. A business with 8% margin and no contract in between sits one outside decision away from losing money. That is why at Masterestaurant, when Diego F. Parra audits a virtual brand, the first indicator is never revenue but the share of income you control directly: own ordering, WhatsApp, in-store pickup.

Direct channel is not ideology, it is the only point you decide

Nobody is going to build a virtual restaurant on owned channels alone, and promising that would be a lie: volume lives in the apps and will stay there. Uber Eats gross bookings reached roughly 74.6 billion dollars in 2024 according to its SEC filing, and worldwide online food delivery is estimated at 1.51 trillion dollars for 2026 with a 6.24% CAGR through 2031 according to Statista. Either you surf that tide or your competitor does. Yet the 20% or 25% of orders you manage to move onto your own channel is the only stretch with no commission, where you keep the customer record, and where somebody else's policy change cannot touch you. The tension resolves cleanly: use the aggregator to acquire, use your own channel to retain, and measure every month how many sales points migrated. If nothing migrated in six months, the aggregator is not your channel, it is your landlord.

Methodology: where these numbers come from and how far they reach

The market figures in this piece come from public sources cited one by one: Global Growth Insights for the dark kitchen projection to 2033, Statista Market Insights for global delivery size in 2026 and its CAGR, Uber Technologies' Form 8-K filed with the SEC for Uber Eats gross bookings in 2024, Swiggy's FY 2023-24 annual report, and DiDi Food Mexico's 2024 press release. The limits deserve saying out loud: these are projections from research firms with proprietary methodologies, not audits, and global chain figures describe nothing about what your aggregator charges in your city. The operating ranges —26% total commission, 3 to 6 points of packaging, the 400 and 1,200 order cutoffs— are consulting criteria, not statistics. Replace them with your own numbers the moment you hold three months of profit and loss per brand, which is the only benchmark that decides anything.

The five differences that decide the outcome

Marginal cost beats total cost. A virtual brand adds only ingredients, packaging and commission; a dark kitchen also adds rent and payroll that get paid whether it rains and nobody orders. At 400 orders a month that difference decides between making 1,900 dollars and losing 2,300. Aggregator commission does not negotiate on size, it negotiates on exclusivity and on the volume of the ACCOUNT, not the location. Four brands under one tax ID pull points a standalone site never sees, and that is the real argument for consolidating before expanding. Packaging is a product cost, not a logistics cost, and almost nobody books it that way.

The five differences that decide the outcome — in practice

Three to six points of ticket in properly solved packaging is normal; below three points the food arrives cold or tipped over and you pay the full refund plus the review. App visibility is bought with discounts, and the discount comes out of your margin, not the platform's. A 20% promotion on a 28% contribution margin leaves eight points: you are working for the algorithm. Customer data stays with the aggregator. Anyone who does not build a direct channel in parallel is renting demand for life, and the day the platform changes its ranking the brand vanishes with no notice.