Cut delivery commissions: the numbers an owner decides on in 2026

The real aggregator cost is not the commission on your invoice

Add 34% to 41% of the ticket, not 27%: that is the true cost of the aggregator channel on an average Latin American order, and until it sits in a spreadsheet cell you are negotiating blind. The contracted commission is only the visible layer; on top of it rides the promotional discount the platform demands to hold your ranking, somewhere between 5 and 12 points depending on the city, plus packaging, which at the Bogotá steakhouse was still priced at 2023 levels, 2,400 pesos per order. With channel food cost at 31% once portions are adjusted, the arithmetic leaves no contribution margin. Arguing five commission points with an account executive means fighting half the problem while the other half drains away in promotions and cardboard.

Why shift 20% to 35% of orders to your own channel instead of negotiating?

Because nobody at the platform has authority to move you out of the commission band, and that percentage lives in the business model, not in the goodwill of whoever answers your email.

DoorDash closed 2024 with roughly 2,583 million orders and 685 million in the fourth quarter alone, up 19% year over year (DoorDash, 2024 results); a company at that scale does not adjust its take rate for a 180-seat steakhouse. Leverage switches sides the day you prove you can live without them. Moving 20% to 35% of volume into your own channel, with the same kitchen and the same third-party courier, turns those 34 to 41 points of cost into something near 12 to 18, and it hands back the customer data that the platform will never release to you.

The demand already next door that you are not billing

Your Google Business Profile pulls searches from people standing 800 meters away and costs you nothing in commission, yet the Bogotá steakhouse had gone months without an order button, with stale photos and 41 unanswered reviews. Meal delivery user penetration reached 27.5% in 2024 and Statista projects 29.2% by 2026, with 2.6 billion users by 2031; in Colombia the segment already touches 19.8% penetration (Statista Market Insights, 2024). That market does not live inside the app: it lives on the phone of someone deciding where to eat in under a minute. As Diego F. Parra, restaurant consultant and founder of Masterestaurant, argues, the owned channel does not start with an expensive app but with a properly built profile, WhatsApp answered in under three minutes, and a delivery menu carrying its own prices.

Delivery is a different business, with a different menu and a different price

Running delivery as an extension of the dining room burns more money than any other habit in the region, and you pay for it with channel food cost at 31% while the dining room sits near 28%. Portions travel differently, packaging weighs, transport shrinkage is real, and the plate that looks sharp on the table arrives washed out twenty minutes later. A serious delivery menu trims down to the 18 or 22 dishes that survive the trip, raises price between 12% and 18% to absorb the commission band without punishing dine-in guests, and cuts every item with contribution margin below 60%. The share data backs this up: ghost kitchens already accounted for roughly 15% of U.S. foodservice delivery sales in 2023 (Statista), and not one of them cooks the dining room menu.

A virtual brand doubles sales, though it does not necessarily build an asset

Launching a virtual restaurant on the kitchen you already run changes the arithmetic and will not rescue you on its own. If the new brand lives exclusively inside the aggregator, you doubled sales paying a high commission and gained no customer of your own, no email address, no repeat-purchase history. The dark kitchen that actually works is born with two channels on day one, and the owned one carries at least 25% of its orders from month one. Capital chasing this sector understands the difference: agrifoodtech fell to 5.5% of global venture capital dollars (AgFunder News, 2024), and the money still moving goes to models with a direct consumer relationship. In China, Meituan and Ele.me clear more than 60 million orders a day (Mordor Intelligence, 2025), where the fight was never about discounts but about who owns the customer.

How to read these numbers in YOUR operation?

The three scenarios share almost nothing, so copying the tactic of the restaurant next door works about as well as copying its menu. Small venue, up to 60 seats and under 15 million pesos a month through aggregators:

skip the app, build the Google profile with an order button and WhatsApp Business, and set the target at 20% owned volume within six months. Group of three venues or more: negotiate third-party couriers by volume, centralize customer data in a single CRM, and aim for 35%, because your cost per owned order drops with every location you add.

Where these benchmarks come from and how far they stretch?

These figures come from verifiable public sources: Statista for user penetration and ghost kitchen share, DoorDash quarterly filings for order volume and growth, AgFunder News for agrifoodtech investment, Mordor Intelligence for the Asian market, and CloudKitchens for published commission tiers.

None of it is primary research by Masterestaurant, and none of it comes from a proprietary sample. They also carry limits worth saying out loud: the commission data describes specific markets, not every Latin American city; commission bands shift by city, category and contract age; channel food cost belongs to a steakhouse with expensive protein and does not transfer to a pizzeria. Treat them as orders of magnitude and measure your own numbers this week.

Where the comparison breaks?

Commission is not the aggregator's real cost. On top of 27% you carry the promotional discount the platform demands so your ranking does not sink, another 5 to 12 points depending on the city, plus packaging almost nobody recosts.

True channel cost on an average Latin American order lands between 34% and 41% of ticket. With that number on the table, arguing five commission points argues half the problem. A virtual restaurant or virtual brand built on the existing kitchen changes the arithmetic without rescuing it alone: if the new brand lives only inside the aggregator, you doubled sales paying a high commission and gained no asset. A dark kitchen from scratch that actually works is born with two channels on day one, and the owned one carries at least 25% of orders by the end of the quarter. Local SEO does not fight the aggregator for the same customer. It fights for the moment before: the person typing «sushi near me» or «pizza delivery open now» into Google, 900 meters from your door, who has not opened any app yet.

Where the comparison breaks — in practice?

According to John Mueller, Search Advocate at Google, a local business listing is only as good as its consistency with the site itself, so identical hours, category and NAP are the floor before any advanced tactic.

Rappi, Uber Eats and DiDi Food are not one channel with three names. Their algorithms weight different things —acceptance time, cancellation rate, promotional density— and the same kitchen can score 4.7★ on one and 4.2★ on another. Reporting aggregate «delivery» contribution hides the likelihood that one of those three platforms is costing you money. The physical menu does NOT disappear when the QR arrives. In the room, print controls service pace, menu narrative and suggestive selling; the QR complements it for delivery, accessibility, price changes and analytics. Keep both, each in its role: operators who scrap print to «save on printing» lose 6% to 11% of average table ticket, several times what the printing cost.