Rappi delivery strategy: the channel that looks like revenue and behaves like debt

1. What is the real decision before uploading a dish to Rappi?

Only dishes whose contribution margin absorbs the aggregator's commission and still leave cash belong on the channel; the rest stay in the dining room.

That filter applies dish by dish, never to the full menu, and it is calculated as channel selling price minus food cost, minus packaging, minus commission. A dish carrying 68% contribution margin in-house, loaded with 27% commission and 4% packaging, drops to 37% and remains a business; run the same math on a 52% dish and it lands at 21%, which no longer pays for the kitchen time it consumes. Food delivery apps moved USD 110 billion in 2024, up 15.5% according to Business of Apps (2025), and that market size never rescues a dish that was weak to begin with. Open a dedicated accounting column for the aggregator before you touch the menu, because a channel measured inside the general P&L is not managed: it gets subsidized by dining-room profit.

2. A channel-level P&L is the first deliverable, not a pretty report

That column needs four lines of its own — channel gross sales, commission withheld, packaging cost and assigned kitchen hours — plus a monthly close that produces its own operating margin. One operator in the 500,000 to 1 million USD band arrived with 31% of total sales on Rappi and, once the column was separated, that 31% contributed 4% of operating profit. Online food delivery in Colombia moved USD 1.18 billion in 2024 with a 7.32% CAGR through 2029, according to Statista Market Insights (2024): the channel grows with you or without you, so the accounting has to arrive before the commercial anxiety. Below 500,000 USD a year, enter Rappi with a maximum of 12 references and a ceiling of 20% of total sales; past that ceiling the kitchen loses the dining room, which is where the ticket pays with no commission attached. The numeric rule is plain: no channel dish under 55% contribution margin before commission, reviewed every 30 days.

3. Under 500,000 and 500,000 to 1 million: concrete thresholds

In the 500,000 to 1 million band the ceiling rises to 28% of sales with 18 to 24 references, and one hard requirement appears — the channel must post 8% operating profit of its own, or you cut the digital menu instead of negotiating commission. Mexico moved USD 9.22 billion in online delivery during 2024 with a 14.66% CAGR, according to Statista (2024); in a market running that fast, the expensive mistake is growing in sales while shrinking in cash. Past 1 million USD a year, the decision stops being how many dishes and becomes how many stations, with an expedition line dedicated to the channel once delivery clears 250 orders a day. The threshold there is channel operating profit above 11%, and packaging gets negotiated by volume, where a container at 1,150 pesos falls to 820 on a 40,000-unit order and hands back nearly 3 margin points.

4. Above 1 million and above 5 million: when the channel becomes infrastructure

Above 5 million you meet the large-format profile or the media-chef profile, which brings its own demand and therefore negotiates tiered commission and in-app visibility, never dish price. The cloud kitchen market is projected at USD 83.5 billion in 2026 with a 9.7% CAGR through 2034, according to Fortune Business Insights (2026): at that scale the channel is an asset, not an experiment. A group billing more than 10 million USD a year does not ask for a discount, it presents a case: consolidated volume, measured preparation times and a cancellation rate below 1.8% are the three arguments that move commission from 27% to 19%. Without clean data per location and per dish that conversation simply does not happen, because the aggregator negotiates against numbers and not against intentions. Diego F. Parra and the Masterestaurant team treat Rappi delivery strategy as a portfolio decision in this band: every virtual brand is approved with its own break-even and switched off if it misses that mark within 90 days.

5. Groups and chains above 10 million: commission is bought with data

Brazil moved roughly USD 18.8 billion in online delivery in 2024, the largest in Latin America according to Statista (2024), and regional scale is precisely what creates bargaining power. Cutting the digital menu to 18-24 dishes that travel well raises station turnover and lowers preparation time, and that protects your rating, which is the variable deciding your visibility inside the app. Uploading 80 references does not deliver variety, it delivers slowness: every extra SKU adds mise en place, occupies fridge space and multiplies the stockouts that end in cancellation. Across 100 daily orders, moving from 14 minutes of preparation to 9 recovers roughly 500 kitchen minutes a month, which amounts to almost three full shifts without hiring anyone. The Asia-Pacific ghost kitchen market reached USD 21.73 billion in 2024 and is projected at USD 60.59 billion by 2032, a 12.8% CAGR according to Coherent Market Insights (2024); operators growing at that pace run short menus and fast stations, not endless catalogs.

6. Packaging is an ingredient and has to be costed like one

Charge packaging to the dish cost, line by line, because it is the cost almost nobody books and the one that quietly eats between 3 and 6 margin points. A container with an airtight lid, the bag, the tamper seal and the sauce sachet easily add 1,400 pesos per order; on a 38,000 ticket that is 3.7% which never showed up in the standard recipe. I got this wrong for years, treating it as operating expense rather than raw material, and the outcome was a reported food cost sitting below the real one. Spain's ghost kitchen market hit USD 928.22 million in 2023 with a 4.5% CAGR through 2032, according to Expert Market Research (2024): in mature markets, packaging is already bought on annual contract with a technical spec, not at the hardware store around the corner. If delivery climbs from 20% to 45% of your sales while the digital menu stays long and packaging stays uncosted, the outcome is not one bad quarter: it is an operation that bills more and distributes less.

7. What happens if the channel grows and you change nothing?

The sequence is predictable — the kitchen saturates, times stretch, the rating drops, the algorithm shows you less, and you compensate with promotions funded by the same margin that was already thin.

A business growing 25% in sales while losing 6 points of operating margin ends up with less absolute cash than before it grew. The global virtual restaurant and ghost kitchen market was valued at USD 65.3 billion in 2023, according to Next Move Strategy Consulting (2023), and that size coexists with a high mortality rate among digital brands for exactly this reason. This week: open the channel column in your P&L and measure the margin of your five best sellers on Rappi. The first difference sounds like accounting housekeeping until you see the number: the traditional method buries delivery inside the general P&L, where commission dissolves among all the other costs until it turns invisible, while the Masterestaurant method opens a separate column and forces the channel to pay for itself.

8. Four differences that move EBITDA

A channel nobody measures separately is not managed. It is subsidized. Second comes portfolio. Uploading 80 SKUs to an aggregator does not buy variety, it buys a slow kitchen and low ratings for delivery times. Cutting down to 18-24 dishes that travel well raises station throughput, drops prep time and protects the contribution margin of every order. Smaller menu, bigger cash. Third is packaging cost, which almost nobody charges to the dish. A container at 900 USD per thousand units, sitting on a 12 USD plate, moves food cost close to a full point, and that point across thousands of orders is the entire profit of the channel. The 32% food cost ceiling remains the absolute maximum, never the target. Fourth is AI applied to the operation, which is where the thought leadership of Diego F. Parra departs from generic foodtech advice: AI is not there to write pretty dish descriptions, it is there to watch margin by daypart, shut the channel down when the kitchen saturates, and answer the two-star review before the aggregator algorithm punishes it in the ranking.