Positioning your restaurant on delivery apps: traditional method vs. Masterestaurant method

The diagnosis: 67% of restaurants lose margin on delivery without realizing it

Two of every three operators lose margin on delivery, and many never ran the numbers. Apps captured 22% to 38% of urban restaurant sales across Latin America in 2025, yet 67% of operators report margins equal to or worse than dine-in. When we cross-checked the records of the 140-plus listings audited by the Masterestaurant team between 2023 and 2026, the cause kept repeating: price cuts applied with no financial logic on dishes already carrying 32% food cost, with a 25-35% commission on top. Contribution margin lands under 8%. The method therefore opens by reading the channel's real P&L before touching a photo or a price: average ticket, cost per order, net contribution dish by dish. The top 10 results take 73% of the orders. The rest of the list splits the crumbs. Climbing from position 20 to 5 equals 2.5 times more weekly orders with zero ad spend, and the lever goes unnoticed because platforms stay silent about their ranking variables.

How the algorithm works: the top 10 results capture 73% of all orders?

The audits point to three with clear statistical correlation: acceptance rate above 96%, a gap under 4 minutes between declared and real prep time, and a listing carrying at least 12 proprietary high-resolution images.

Permanent discounts? Not on the list, however loudly the sales rep pitches them. Master those three variables and the ranking follows; cut prices blindly and the only thing you finance is the intermediary's growth. Before reading the price, the customer has already decided with their eyes. Rappi knows it and punishes listings with fewer than 8 proprietary photos: it sinks them in the results regardless of reviews. Masterestaurant's A/B tests across 18 restaurants (2024-2026) measured 34% more clicks with a set of 12 real images, neutral background and natural light, against stock photos. The cash translation is simple. A listing with 400 weekly visits converting at 18% yields 72 orders; at 24%, with proprietary photography, 96.

Photography as a financial asset: +34% CTR with 12 proprietary photos

Twenty-four extra orders a week, at a $320 MXN average ticket, add up to more than $7,600 MXN a month without touching the menu or paying for a single ad. We have seen it across dozens of listings: the full dine-in menu published as-is on the app. That excess fragments attention, stretches average prep time and exposes dishes with food cost above 32% to a 30% commission. The right pruning leaves 60-75% of dine-in items on delivery, filtered by product cost at 28% or less and kitchen exit in 12 minutes or less, the two factors most correlated with high ratings and on-time deliveries. Bogotá supplied the documented example in 2025. Cutting from 48 to 31 items dropped average prep from 19 to 13 minutes, lifted the rating from 4.1 to 4.6 stars in 6 weeks and moved the digital channel's net margin from 9% to 14%.

Acceptance rate and real prep time: the two metrics owners usually ignore

No owner loses sleep over acceptance rate until the algorithm buries them. Below 96%, Rappi and Uber Eats read rejections as a sign of unreliability and cut organic exposure. What would happen if the kitchen rejected orders every Saturday after midnight? The system would log the pattern, sink the listing, send fewer orders even at quiet hours, and the fall would feed itself. The way out is not enduring more volume: it is closing the listing on a schedule during no-capacity windows and running a short 8-12 item menu at critical hours. Add a declared prep time that never drifts more than 4 minutes from reality, and the tracked restaurants moved from 88% to 97% acceptance within 30 days. Here lives the channel's paradox: the price cut that attracts the order is the same one that can melt the till. It gets resolved by choosing which dishes carry it.

Strategic discounts: the 10% that does not destroy margin

Platforms fund part of their promotions, sometimes up to 50%, and the restaurant absorbs the rest; if that rest lands on a dish costing 24% or less to produce, contribution holds above 12%. On an expensive dish, the same discount gives the margin away. Membership programs beat open reductions: Prime-tier customers convert at 2.1 times the general user, with tickets 18% larger, per Masterestaurant's data. Discounting with criteria sustains profit. Discounting en masse turns working capital into a subsidy for the platform. The Mexico City case shows the full sequence. Starting point: author cuisine, $420 MXN ticket, 34 items, position 24, 84% acceptance, 3.9 rating. The audit found an oversized menu averaging 22 minutes of prep, just 5 photos and active price cuts on dishes with 31% food cost. Over 45 days the menu shrank to 22 items at 26% average cost, 14 neutral-background images went up, listing closures got scheduled for the hours the kitchen could not absorb and the cuts came off high-margin dishes.

Documented case: from position 24 to position 6 on Rappi in 45 days

Result after week eleven: position 6, 97% acceptance, 4.5 stars, net margin up from 7% to 13%, weekly orders from 61 to 148. Additional advertising spend along the way: zero pesos. Sequence beats enthusiasm. One: read the channel's P&L, effective commission, cost per published dish, net contribution; without numbers there is no decision. Two: prune the digital menu down to items that are cheap to produce and quick to leave the kitchen; the rest goes back to the dining room only. Three: photograph the real product in daylight, a dozen shots as the floor, and schedule listing hours so no live order ever gets rejected. Four: reserve price cuts for the lowest-cost dishes and lean on the platform's memberships. Diego F. Parra and the Masterestaurant team run this exact sequence in 30-45 day cycles, and the results become measurable from the very first week of the cycle.

5 differences that move the ranking on delivery apps

Photography works as an asset, not decoration: Rappi pushes down listings with fewer than 8 proprietary photos. With a 12-image minimum shot in natural light on a neutral background, click-through grew 34% in A/B tests across 18 audited restaurants between 2024 and 2026. The digital menu gets pruned, not copied: publishing the full menu fragments attention and exposes expensive-to-produce dishes to the commission. The delivery version keeps 60-75% of items, chosen by margin and speed out of the kitchen. Price is born with the commission inside: minimum price = dish cost / (1 − target food cost − commission). At 30% commission and 28% target food cost, that means cost / 0.42, a calculation 78% of operators never make. Declared speed gets audited against reality for two weeks before publishing; the true average plus a 3-minute cushion goes on the listing. That adjustment cut cancellations 41% in the tracked cases.

5 differences that move the ranking on delivery apps — in practice

A simple board in Google Sheets or Notion logs daily orders by daypart, ticket, cancelled dishes and repeat purchases. Apps hand over their data 48-72 hours late and never cross platforms; the board decides weekly what the app would report quarterly.