Delivery commissions killing your margin: myth vs reality — 2026 white paper

Chapter 1 — Why the commission isn't what's making you lose money on delivery

Marketplace commission destroys margin only when the menu was costed on the assumption that guests would order wine, dessert and leave a tip, and none of those three lines exist in a delivery order.

A three-unit operator in Bogotá was selling 41% of his volume through apps at an average commission of 26% on annual revenue of 780 thousand dollars, and his blended food cost reached 34,8%; packaging, free in the dining room, took another 4,1 points out of the digital ticket. The channel wasn't breaking him: it was showing him arithmetic the dining room had been hiding for years.

Walking away solves nothing either, because Colombia's digital channel already carries too much weight in the operation to give up on it. You don't switch off a channel growing at that pace; you price it properly.

Chapter 2 — Channel pricing: the 12 to 18 points given away out of commercial shyness

Keeping TWO price lists, one for the table and one for the app, is the decision that recovers the most margin in under a week, and the restaurant that earns money digitally states it inside the app without apologising. That gap pays the commission, the packaging and the thermal bag. I got this wrong for years, recommending price parity for the sake of brand coherence; coherence doesn't cover Saturday's payroll.

Chapter 3 — A trimmed digital menu: ABSOLUTE margin in money, not percentage

Put on the digital channel the dishes with the highest absolute contribution margin, measured in money per unit sold, and forget the percentage, which is the favourite trap of badly applied menu engineering. A dish with 68% margin and a 6 dollar ticket leaves 4,08 dollars of contribution; another at 58% with a 14 dollar ticket leaves 8,12, nearly double the cash for the same minute of kitchen time.

The operator who publishes 80 SKUs to capture every niche multiplies inventory, waste and dispatch times, and ends up subsidising items that don't even cover their own commission. A digital menu of 18 to 24 items, with packaging standardised into three formats, holds up better in a market as large as Brazil's, the biggest in Latin America. Fewer items, more cash per order.

Chapter 4 — Revenue bands: the same commission hits differently by size

That 26% behaves in radically different ways depending on the size of the operation, and mixing up the bands is why so much delivery advice helps nobody. Below 500 thousand dollars a year there is no acquisition budget capable of replacing the discovery traffic of a marketplace: pay the commission, raise digital prices 15% and don't invent your own ecommerce.

Between 500 thousand and 1 million, a defensive direct channel makes sense to capture 15% to 20% of recurring demand. Above 1 million, every badly negotiated commission point on 40% of volume takes roughly 4 thousand dollars of EBITDA a year, straight out. And above 5 million you already negotiate volume rates and run your own satellite kitchen, in a Mexican cloud kitchen market that keeps expanding.

Chapter 5 — Above 10 million: the celebrity restaurant and its invisible cost

Above 5 million and especially above 10 million dollars, the celebrity-chef house or the large-format themed venue plays a different game, and its enemy isn't commission but the brand promise. These operations cannot dispatch in generic packaging or tolerate a plate arriving lukewarm, because the guest paid for a 60 dollar expectation and grades it by dining-room standards.

Their own cost shows up in three lines: branded packaging that triples the standard, between 1,80 and 3,20 dollars per order; a dedicated expediting line with its own payroll; and active temperature control. Reputation before coverage.

Chapter 6 — The five hidden costs the app report never shows you

Your real commission is not the one in the contract, and across the operations I review the gap runs 6 to 11 points above the number the owner repeats from memory.

Add packaging, which in the Bogotá case weighed 4,1 points of the ticket; co-funded promotions of the 2-for-1 kind, usually another 3 to 5 points; in-app advertising, between 1 and 3; refunds for incomplete orders, close to 1,2; and the waste of a mise en place sized for a peak the app decides, not you. With a 26% contract, effective commission lands between 32% and 37%.

That is the figure that belongs in the recipe costing before you set a digital price. Measure it for a month, with invoices, and stop arguing with the nominal rate.

Chapter 7 — The direct channel: when it pays and what it really costs

Direct ordering only wins when the cost of acquiring and retaining that customer sits below the commission it saves, and that calculation fails more often than its advocates admit. A direct order costs between 8% and 14% of the ticket once you add payment gateway (2,9% to 3,9%), ordering platform, outsourced courier per trip and the marketing needed for anyone to remember your domain.

Against an effective commission of 34%, the saving is real; against a volume of 200 monthly orders, the platform eats the benefit. The break-even I use with clients sits near 600 digital orders a month. Below that, the marketplace is your rented sales force; above it, start buying your own database one order at a time.

Chapter 8 — The Masterestaurant method to rebuild digital margin in 90 days

Rebuilding digital margin takes one quarter and follows an order that Diego F. Parra applies with Masterestaurant clients without exception, because reversing it produces cuts that destroy sales. Month one: measure the real effective commission and re-cost the 20 best-selling dishes on the channel, with per-dish food cost below 32% as a ceiling, never as a target. Month two: differentiated pricing by channel and pruning the digital menu to 18-24 items.

Month three: volume rate negotiation and activation of the direct channel if it clears 600 monthly orders. In a one million dollar operation with 40% digital volume, that sequence recovers between 3 and 6 points of EBITDA, in a global cloud kitchen market that MarkNtel Advisors (2024) sizes at 45.650 million dollars. Start this week with your effective commission.

Chapter 9 — Five differences that separate a profitable delivery operation from one that bleeds

PRICE BY CHANNEL, not a single price. The profitable digital restaurant keeps two price lists and states it openly in the app. The bleeding one fears the comparing customer and gives away 12 to 18 margin points out of commercial shyness. A TRIMMED DIGITAL MENU. Menu engineering applied to the channel: you publish the dishes with the highest ABSOLUTE contribution margin in currency, not the best percentage.

A dish at 68% margin on a 6-dollar ticket contributes less cash than one at 58% on a 14-dollar ticket. The operator listing 80 SKU multiplies inventory, waste and dispatch times to capture a long tail of demand that does not pay. PACKAGING COSTED AS AN INGREDIENT. In the digital dish recipe cost, the container, bag, tamper seal and cutlery enter as a cost line, not as an administrative expense.

Chapter 10 — Five differences that separate a profitable delivery operation from one that bleeds — in practice

That reclassification alone flips the profitability verdict on 20% to 30% of the digital menu in an average operation. AN OWNED DIGITAL ENGINE RUNNING IN PARALLEL. Optimised Google Business Profile, reviews managed toward 4,6 stars or better, and a direct-order link inside the Maps listing. The marketplace keeps delivering discovery; the owned channel captures repeat purchase, and repeat purchase is where margin lives. READING THE ALGORITHM instead of complaining about it.

Rappi, iFood and Uber Eats rank on acceptance rate, actual prep time against promised time, store rating and menu availability. An operation at 96% acceptance and 18 minutes of prep pays the same commission as one at 82% and 31 minutes, yet earns two to four times the impressions. That is where commission cheapens itself, diluted across more volume.