Own Delivery vs Delivery Apps: Before vs After Statistics 2026

The commission nobody calculates right: 28% that destroys real margin

A 29% food cost looks healthy until the order leaves through an app. Add packaging (3%), platform commission (28%) and internal logistics (4%), and real margin lands under 2% or in the red. Arithmetic, not bad management. Masterestaurant audited more than 200 independent restaurants between 2022 and 2026 and that was the most frequent scenario in the sample. Diego F. Parra also documented the figure that explains the ending: before 2023, 78% of those businesses relied on platforms for over 60% of their off-premise sales. Every order drained 8 to 12 points of contribution margin. Volume never paid that leak back; it only made it wider. The dish was never the problem.

The real threshold: 35 daily orders before migrating

Thirty-five daily orders. That is the threshold separating a profitable in-house channel from an expensive hobby, per the costing Masterestaurant ran across 40 Latin American kitchens between 2024 and 2026. Below it, the fixed cost of a fleet (courier, motorcycle, insurance, fuel) eats the commission savings and ends up pricier than paying the app its 28%. Cross the line and the equation flips: variable cost per delivery falls from USD 8.40 to USD 4.20 on average, and the 15 to 18 recovered margin points show up on the monthly income statement rather than in a promise. Zone matters as much as volume. Keep the radius at 3 km or less; beyond that, fuel and dead time devour the differential.

Repurchase data: what platforms never hand over

The most valuable asset in a delivery order is not the ticket, it is knowing who placed it. Platforms keep the name, phone and history, and turn them into their own property. Restaurants Masterestaurant supported in 2024 and 2025 moved to WhatsApp Business with a basic CRM and watched repurchase jump from 22% to 41% in six months. The lever? Reactivation campaigns with an acquisition cost 5 times lower than paying for ads inside the app. A personalized message to a customer with 12 orders on file converts at 18%; a paid ad inside the platform, at 3.4% on average, per those same operations in 2025. Against that gap, every phone number captured beats any ranking in the app's search box.

Delivery times: the operational advantage of a proprietary fleet

Thirty-eight minutes door to door at peak with platform couriers; 27 with an in-house fleet on fixed routes. Eleven minutes less, a 29% cut. The stopwatch contradicts the perception that apps deliver faster. The cause is structural. A platform courier carries orders from several restaurants at once and optimizes the route for himself; your own courier serves one kitchen and rides a 2.5 km zone he knows by heart. The difference shows where it hurts: in 2025, 64% of the sample kitchens that completed the switch closed their first 60 days without a single late-delivery complaint. No assignment algorithm matches that consistency on a rainy Friday night, when every order in the city competes for the same courier.

Dependency and risk: from 92% on one app to a hybrid model

Putting 92% of delivery volume on one platform equals buying 90% of your ingredients from a single supplier. When we cross-checked the sector's visibility records for this observatory, the pattern repeated: the app changed its algorithm in 2024, the restaurant slid down the internal search results, and sales fell 35% in three weeks with nothing altered in the operation itself. It hurts because the owner controls none of it. The hybrid model Masterestaurant documented spreads the risk: a direct channel for loyal volume, apps for acquiring new customers. Dependency dropped from 92% to 31% of total volume in restaurants that implemented it between 2024 and 2026, with steadier monthly revenue and a better margin per order.

Average ticket: the effect of real pricing without platform inflation

Inflating app prices 12% to 20% above the dine-in menu is the usual fix for the commission. The remedy poisons: customers read the restaurant as expensive, conversion inside the platform drops, and the algorithm punishes visibility. Once sample restaurants published real prices on the direct channel, the average ticket rose 18%. The explanation sits in the cart. An order stuck at USD 22 inflated on the app reached USD 26 at fair pricing, because the diner added the dessert or drink previously skipped over perceived cost. That increase is clean gain; no commission touches it. And it holds as long as the dine-in menu and the direct channel tell the same price story.

The 2024–2026 turning point: what sector statistics show

Restaurante Cerca de Mí documents 2024-2026 as the period of deepest reconfiguration in Latin America's independent delivery. Sixty-four percent of kitchens that moved to a hybrid model reported cash improvements in under 90 days. Diego F. Parra frames the shift as mathematical, never ideological: when the commission exceeds the dish's contribution margin, selling through that platform subsidizes the intermediary with the owner's capital. The 2025 figures widen the gap. At similar volumes, restaurants with an active direct channel generated 1.8 to 2.4 times more contribution margin per order than platform-dependent competitors. Eighteen months of that difference decide who reinvests in quality and who posts the closing notice. The sector data leaves little room for neutrality.

How to execute the migration: the first 90 days with real numbers?

Week 1: WhatsApp Business live, catalog and payments included. Weeks 2 and 3: capture the history of the last 200 app customers, with name, order and frequency.

Week 4: first in-house reactivation campaign. The protocol Masterestaurant applied between 2025 and 2026 starts under USD 300 in tools and switches off no platform. Results from those restaurants set the expected pace: the direct channel carried 40% of total volume by day 60 and 58% by day 90, while app sales held steady. Dependency falls on its own when the alternative grows faster. Nobody has to slam the apps' door; building a more profitable channel right next to it, order by order, is enough.