Delivery unit economics: the commission myth and the channel-mix reality

How much margin does an aggregator order actually leave versus a direct-channel one?

An aggregator order leaves between 3% and 9% contribution margin, while the same dish sold through your own channel leaves between 18% and 26%, and the gap comes from a stack of deductions nobody adds up, not from price.

Take the roast chicken with fries: 15.90 USD on the app, 4.45 in commission at 28%, 0.62 for the insulated bag, 4.77 in food at a 30% food cost that the National Restaurant Association places inside the optimal 28-35% band, and 0.64 in refunds if 4% of tickets collapse over a forgotten dressing. That leaves 5.42 before payroll or rent. The same dish at 12.90 over WhatsApp with an in-house driver at 1.80 leaves 6.30 clean. The direct channel WINS by 88 cents on a ticket the owner believed was three dollars richer on the app. A 26% commission does not take 26% of your profit: it takes 37% of it, and that accounting detail has closed more kitchens than any ingredient spike.

Commission is charged on the sale, not on what you keep

With a 30% food cost your gross margin is 70%, meaning 10.50 USD on a 15-dollar plate; the app collects 3.90 on the full selling price, and you are left with 6.60 before packaging, backup riders and the 2% of cancelled orders. Push food cost to 32% —the CEILING Masterestaurant considers acceptable and not advisable— and those 6.60 drop to 6.30 while the platform's cut never moves a cent. Your own channel flips the arithmetic: you pay payment processing of 2.9% plus 0.30, roughly 0.74 on that same ticket, and the remaining 96% lands in your account. The direct channel wins, and it wins by construction, not by negotiation. Inflating the digital menu 15% to 25% to absorb commission works exactly until the guest compares both screens, and the guest compares. A 12.90 dish listed at 15.90 climbs 23.3%; at 28% commission you recover 3 gross dollars and hand over 4.45, so you still lose 1.45 against counter sales while burning your price anchor.

Raising app prices buys time, never margin

When that same diner opens your own site and sees 12.90, the app stops being expensive and you become the liar. Your own channel allows the opposite: counter pricing, delivery of 1.80 to 2.50 charged separately and openly, and a ticket the customer understands. Here the comparison is not a fight: the app wins on acquisition, the direct channel wins on clean pricing, and clean pricing is what holds repeat orders together. Paying commission on a new customer's first order is cheap acquisition; paying it on the SAME customer's seventh order is a leak that compounds month after month. Diego F. Parra repeats this in every foodtech audit: commission is not the enemy, commission is the price of a customer you did not have, and that price is only worth paying once. A diner ordering twice a month for a year means 24 tickets of 15.90; through an aggregator at 28% you give away 106.56 USD a year in commission on that single household.

Order number seven is where everything gets decided

Move them to your own channel from order three onward and you recover 80 of those 106 dollars with a 4-dollar coupon printed on the bag. At Masterestaurant we measure the local digital channel dish by dish, because the difference between the order that arrives via Google Maps and the one that arrives via the Rappi carousel never shows up in the monthly P&L. The aggregator sells discovery and your own channel sells relationship, and mixing them up is what wrecks the decision. Discovery scale is real and worth measuring without romance: iFood processed 100 million orders in the single month of August 2024, with more than 380,000 partner establishments across 1,500 Brazilian cities, per its own institutional data compiled by Statista. No neighborhood restaurant builds that traffic alone. But the toll for entering that river is that the platform keeps the diner's name, phone and purchase history while you keep the work of cooking.

Volume against control: what each channel buys with your money

Your own channel reverses the equation: traffic you must earn through Google Maps, reviews and a paper bag, in exchange for a CRM that belongs to you. For a venue with a paid-off kitchen, the direct channel wins in year two and loses in year one. A chicken rotisserie running 47 delivery orders a day shifted its mix from 100% aggregator to 62% aggregator and 38% direct in five months, and its digital contribution margin went from 6.1% to 14.8%. The raw figures: 47 tickets at 15.90 make 747.30 USD in daily sales; with everything on the app at 28% commission the net came to 45.60 of margin. By moving 18 of those orders to direct WhatsApp ordering at the 12.90 counter price plus a 2-dollar delivery fee charged to the customer, sales fell to 692 but margin climbed to 102.40 a day, roughly 3,070 a month.

The rotisserie case: 47 daily orders, two arithmetics

The 20% promotion needed to appear in the carousel, which ate 3.18 per ticket, got cut at the root. The uncomfortable lesson: sales dropped 7.4% and cash rose 124%, and no platform dashboard would ever have shown that. If you shut off the aggregator tomorrow and keep only 35% of those orders through your own channel, do you end up better or worse? With 47 daily tickets at 6.1% margin you earn 45.60; with 16 direct tickets at 26% margin on 12.90 you earn 53.60. You come out ahead, and that stress test turns the conversation from «I can't leave the app» into «how much retention do I need before I can». The critical threshold sits at 33%: below that migration rate, switching off the aggregator costs you cash; above it, you have room to spare. Here is the trade of the trade and its way out: the aggregator draining your margin is the same one building the customer base that later lets you do without it, provided you capture the contact from order one.

The scenario almost nobody runs: switching the app off on a Tuesday

Whoever fails to capture never migrates, and pays commission forever. If your venue is already paid off, your kitchen is built and you run more than 25 orders a day, the winner is the DIRECT CHANNEL, and the aggregator stays as an acquisition channel with a cap: no more than 45% of your digital volume. If you open next month with no customer base and no reviews, live off the aggregator for a full first semester and do not agonize over the 5% margin, because you are buying traffic you have no other way to buy. If you run a dark kitchen with no storefront and no foot traffic, the app is not a channel, it is your entire display window, and the arithmetic changes: your lever is not commission but food cost under 30% and order density per hour. This week, print the 4-dollar coupon, drop it into the next 300 bags and measure how many come back direct.

What to choose for your profile, with no hedging?

That number is your real unit economics. Commission is charged on the SELLING PRICE, never on your margin, and that accounting detail kills more restaurants than any ingredient crisis.

If food cost runs at 30% and gross margin at 70%, a 26% commission takes 37% of everything left to pay payroll, rent and your own salary. On a 15 USD dish with 4.50 of food in it, the app keeps 3.90 and you are left with 6.60 before packaging. Push food cost to 32%, the ceiling Masterestaurant treats as acceptable and never advisable, and the order is technically alive but clinically dead. Inflating app prices solves nothing, it postpones. Most owners raise the digital catalogue 15% to 25% to absorb the commission, and it works until the customer compares. A US Foods consumer study found 65% of diners would rather order direct from the restaurant when the price matches, and platforms also demote restaurants priced well above their category, so the markup buys margin per order and costs search positions.

The four differences that decide your cash

Direct-channel acquisition cost is paid ONCE; aggregator cost is paid every single time. Building a Google Business Profile with weekly photos, review replies and offer posts costs roughly eight hours a month, and once that profile holds the local pack for restaurant searches nearby it keeps producing orders with no toll attached. Here sits the paradox almost nobody resolves: the very customer the app introduced for 4.45 USD in commission can come back thirty times through your site at zero extra cost, provided you bother to ask for his WhatsApp the first time. Refunds and delivery shrink never appear in a platform sales deck, and they consume 2% to 5% of gross channel revenue. An order refunded because the bread arrived cold does not return the bread: you paid the ingredient, the labour, the packaging and then the chargeback on top. Book it separately from food cost or you will keep blaming your kitchen when the expensive thing is your packaging.