Delivery unit economics in figures: what changes before and after you measure the actual order

Why delivery margin is calculated per order, not per month?

Delivery margin gets calculated per order because the channel total hides exactly the losses you need to see. A steakhouse in Medellín was billing 41 million pesos a month through the app and celebrating that number;

once we opened the income statement of a single 38,000-peso order, 1,900 pesos were left for the owner, roughly 5% net that no monthly sheet ever showed. The arithmetic is simple, which is why it stings: the aggregator commission takes between 15% and 30% of order value, packaging adds another 3% or 4%, and the promotion the algorithm rewards eats whatever survived. That channel does move real money — DoorDash generated nearly US$ 60 billion for local merchants in 2024, according to its annual results — but the money passes through the register instead of staying. These figures together trigger one decision: build a per-order P&L before you upload one more dish to the app.

The channel's size justifies running it as a separate location

Off-premise stopped being an appendix of the dining room: the National Restaurant Association measured in 2025 that 29% of US industry sales already happen outside the venue, with a 35% projection for 2026. Six points in one year is not noise, it is a full kitchen shift. Over the same period DoorDash reported marketplace volume near US$ 80.2 billion (2024) and Just Eat Takeaway closed with EUR 26.3 billion in GTV, also in 2024. Once a channel reaches that size, running it as an extension of the dining room becomes an accounting error before it is a strategic one. Diego F. Parra keeps pressing the same practical consequence at Masterestaurant: if the digital channel carries more than 20% of your sales, it deserves its own income statement, its own menu and its own owner with a name. Decision triggered: split the channel P&L this month, even on an ugly spreadsheet.

Channel pricing is not a trick, it is accounting

Charging the same dining-room price inside the app means handing over 15 to 30 points of margin, which is precisely the commission range aggregators charge. A dish closing at 32% food cost on the table —the ceiling the MASTERESTAURANT method sets and never negotiates— jumps above an effective 40% the moment commission lands, and at that point the dish stops paying for the kitchen. A 12% to 15% markup on menu price pulls channel food cost back under the ceiling without touching the recipe. Customers accept it, because their reference point is the app next door, never your table: ordering from a phone, nobody remembers what the same dish cost on Saturday with the family. I got this wrong for years, recommending price parity out of fear of the review. Decision: reprice the digital menu this week, dish by dish, not with a flat percentage. A vented carton costs about 600 pesos more than the generic one and prevents the soggy fries that sink your rating, so that expense belongs to the channel P&L rather than to the purchasing budget.

Packaging is an ingredient, not a purchase

On a 38,000-peso order, 750 pesos of packaging is under 2% of the ticket and buys half a point of rating on a platform where placement follows the score. Flip it around: if cheap packaging costs you two tenths of a star and those two tenths drop you three spots in the listing, saving 600 pesos per order will cost you dozens of orders a month that never show up in any report, because orders that never arrive are never counted. That is the trap. With 37% of adults ordering delivery at least once a week (UpMenu, 2024), the review is your storefront. Decision: test premium packaging on your three best sellers for 30 days and measure the rating, not the spend. Raising the average ticket leaves more absolute margin than clawing three commission points out of the aggregator, and that is the conversation almost nobody has with their team.

Average ticket beats commission negotiation

Run the numbers: on a 38,000-peso order, three commission points are worth 1,140 pesos; moving that same order to 51,000 with a combo and a drink adds 13,000 in sales that, even after 25% commission and 30% food cost, leaves roughly 5,800 pesos clean. Five times more, without calling anyone or signing anything. With more than 40% of adults ordering delivery or takeout three to five times a month (UpMenu, 2024), frequency is already there; what is missing is the size of each visit. The decision these figures trigger: build two combos with a drink and dessert, place them at the top of the digital menu and review average ticket after 14 days. Promotions are the line item that breaks the channel, and the owner rarely decides them. A two-for-one switched on some Thursday by the community manager on a dish with 32% food cost pushes it to an instant 64%, and combined with 25% commission the order is born at a loss before the cook fires the griddle.

Discounts you never approved land in your P&L too

Discounts the app applies and then bills back to you behave the same way, with the added insult of surfacing in the biweekly settlement, once thousands of orders have already been cooked. The hard rule is uncomfortable but it saves kitchens: no promotion goes live without calculating the promoted order's margin first, and no dish above 28% food cost enters a two-for-one. Decision: revoke promotion access for everyone except whoever signs the P&L, and set a biweekly alarm to check the discount line against gross sales. An order arriving through your own search presence —your Google listing, your site, your WhatsApp— is worth 15 to 30 points more than the identical order arriving through the app, because commission there is zero and the customer data belongs to you. The useful analogy is drive-thru: more than 50% of US QSR revenue comes from that channel (Business Research Insights, 2024), and independent drive-thru-only units sold a median of US$ 9.227 million in 2024 according to the QSR Drive-Thru Report.

Owned traffic is the only asset that charges no commission

Nobody built that by renting someone else's traffic. The aggregator is an expensive window and a decent acquirer of new customers; the mistake is leaving it as the channel for customers who already know you. Decision triggered: slip a QR flyer with your own code into every bag that leaves through the app and measure what share of those orders returns through your direct channel within 60 days. Three numbers and their action, no decoration. First: 32% food cost as the per-dish ceiling, measured AFTER the channel markup — if an app dish misses it, raise the price or pull it off the digital menu this Friday. Second: absolute margin in pesos per order, the one that in the Medellín steakhouse came to 1,900 out of 38,000; calculate it for your ten best sellers and sort the digital menu by absolute margin instead of popularity.

The three figures worth tattooing

Third: 35% of sales projected off-premise by 2026 (National Restaurant Association, 2025), the figure that turns everything above into something urgent — if a third of your business is going to live outside the dining room, that third needs an owner with a name, its own P&L and a biweekly review. Start today with the second one: without per-order margin, the other two have nothing to stand on. Channel pricing is not a trick, it is accounting. Charging the dining-room price inside the app means giving away 15 to 30 points of margin; a 12-15% uplift keeps food cost under the 32% ceiling the MASTERESTAURANT method sets, and the delivery customer accepts it because they compare across apps, not against your table. Packaging is an ingredient. A vented board costs 600 pesos more than a generic one and prevents the cold fries that sink your rating, so that call belongs in the channel P&L rather than in purchasing, where 750 pesos of packaging buys half a star.

Five differences that actually move margin

Average ticket does what commission talks never will. Lifting 38,000 to 51,000 pesos with a bundle and a drink yields more absolute margin than squeezing three points out of an aggregator with a thousand restaurants waiting in line. Own traffic is the only commission that falls by itself. A worked Google Business Profile pushes orders to your site, and every percentage point migrating from marketplace to direct is worth roughly 10,600 pesos on every 38,000 of sales at a 28% commission. Reviews are distribution, not reputation. Rappi, Uber Eats and DiDi rank by rating, prep time and cancellation rate, so climbing from 4.1 to 4.7 stars lifts your virtual restaurant several screens and multiplies orders without a peso of paid media.