Rappi delivery strategy: before vs after with Masterestaurant

When does Rappi as your only channel stop working?

Rappi stops working at order number two, the moment your repeat rate by channel shows you paying acquisition fees for customers who were already yours.

While the app keeps handing you people who had never tried the food, the math holds: your own digital ads in Lima, Bogotá or Mexico City rarely deliver a new customer under USD 4 to USD 7, and a 22% commission on a first USD 15 ticket costs USD 3.30. Cheap. Then that same person orders nine more times, always through the aggregator, and you paid nine tolls for somebody who had been yours since the first package. Pull the monthly settlement and split orders into two columns, new and returning. Past a certain share of app revenue coming from returning customers, what you run is not an acquisition channel: it is a LEASE on your own customer base.

Commission is not the cost: add packaging and co-funded discounts

Add commission, packaging, co-funded discounts and a food cost calculated for a dish that was never built to travel: that is the cost of the channel, not the percentage printed in the contract. Round numbers make it obvious: bill USD 9,000 a month on Rappi at 24% commission, with 9 points of co-funded promotion, and you hand the channel USD 2,970 instead of the USD 2,160 quoted in hallway conversation. Those USD 810 are the part-time kitchen wage you chose not to pay without noticing the choice. Packaging has not even entered: box, seal, cutlery, bag and label weigh USD 0.45 to USD 0.80 per order across most Latin American operations. Diego F. Parra opens the settlement line by line before anyone says the word strategy, because nobody negotiates a percentage whose weight on break-even they cannot name.

Option 1: your own channel via WhatsApp and QR, for whoever already has repeaters

Moving REPEAT purchase to a channel you own, with WhatsApp Business for the order and a QR menu for the catalog, suits the owner who already spotted returning customers in the settlement: little money, plenty of discipline. The WhatsApp Business line is free, the digital menu runs USD 15 to USD 40 monthly, and each drop, in-house or contracted, sits between USD 1.50 and USD 2.80 depending on the city. What costs real money is human: somebody answering messages at peak without letting dispatch times collapse. In exchange you keep the phone number, the consumption history and 100% of the ticket minus the delivery fee. Against it, and this deserves saying plainly: nobody new shows up here, so a small customer base stays exactly as small as it was.

Option 2: rebuild the delivery menu to absorb the commission

A portion that travels, gets reheated and lands twenty minutes later is not the portion that leaves for the dining room, and that kitchen truism is the second alternative: leave the channel alone, rewrite the recipe. It serves the operator who needs Rappi for volume. The delivery menu gets redesigned SKU by SKU: dishes that collapse in transit come out, expensive garnishes nobody misses inside a box get reformulated, real portion weights get corrected. Masterestaurant S.A.S. documents a regional case where rebuilding 14 SKUs for transport brought food cost from 34% to 29%, and the selling price never moved a cent. Mind the house ceiling: 32% per dish is the MAXIMUM, never the target. Five recovered points absorb much of a 24% commission, and the guest never notices.

Option 3: a dark kitchen or virtual brand inside your own kitchen

If your kitchen spends three to six in the afternoon staring at the ceiling, a virtual brand fits there: same production line, same crew, a separate identity that lives only inside the app. Market context backs it, since cloud kitchens moved USD 45.65 billion in 2024 according to MarkNtel Advisors, and Mexico alone accounted for USD 1.1 billion that same year with a 10.74% CAGR toward 2033, per IMARC Group. No passing fad. Entry stays cheap on equipment and staff you already pay for: photography, health registration for the brand, roughly USD 300 to USD 900 of setup. Risk lives at the pass window, where two brands that go out through the same opening multiply dispatch errors, and a prep time that stretches from 18 to 27 minutes costs you ranking inside the platform.

What Rappi genuinely solves and you cannot build alone?

Real incremental demand, a delivery fleet no small operator builds alone, a payment system that pulls cash out of the till and a documented formalization effect:

that is what the platform contributes, and the trade slides into caricature whenever it denies the list. Scale is not marginal either. DoorDash's marketplace moved USD 21.3 billion in GOV during the fourth quarter of 2024 alone, up 21% year over year, per its financial results. Business of Apps puts the delivery app market at USD 110 billion in 2024, up 15.5%, while Statista figures Brazil, the largest in Latin America, at roughly USD 18.8 billion. Fighting infrastructure of that size is losing by definition. Using it for what it does well, and nothing else, is the only position numbers support.

Data as an asset: why this is development policy, not an industry complaint

Here sits the tension almost nobody resolves: the same platform that digitizes a restaurant can leave it weaker than it was. According to Ángel Melguizo, economist and former head of the Latin America Unit at the OECD Development Centre, digitalization of small firms generates productivity only when the company keeps the transaction data; if the intermediary holds it, the company digitizes and loses bargaining power on the way. The bridge is operational, not ideological. SATE Institute treats it as productive-development policy: the Inter-American Development Bank puts MSMEs at the majority of formal regional employment. Masterestaurant S.A.S., as technology partner, instruments food cost per SKU, dispatch minutes and repeat rate by channel, and an operator who knows margin by channel walks into a bank as a different borrower.

When you should NOT change a thing?

Keeping Rappi as your main channel is the correct call in three situations, and saying so belongs to the job. You opened under six months ago and have no base to bring back;

without repeat customers, an owned channel is a silent WhatsApp line. Your average ticket falls below USD 8 while your own delivery costs USD 2.20, so taking over the drop eats 27 points of margin and leaves you worse off than the commission did. Or the kitchen already runs at 85% of peak capacity, where one more channel stretches dispatch and sinks your rating. Measure for a month before moving anything: new orders against repeat orders, food cost per delivery SKU, dispatch minutes. With those three numbers on the table, the alternative picks itself.

The differences that actually move cash

Nobody goes under because of a 24% commission: they go under because of the sum nobody totals, commission plus packaging plus co-funded discounts plus a food cost designed for a dish that never travels. On USD 9,000 of monthly billing, with 9 points of co-funded promotion on top, the channel takes USD 2,970 rather than the USD 2,160 the headline percentage promises. Delivery food cost is not dining-room food cost, since the portion travels, gets reheated and arrives under different conditions, so the recipe gets redesigned before anyone argues about prices. And the house ceiling remains 32% per dish: a ceiling, never a target. The printed menu does not disappear when the QR arrives.

The differences that actually move cash — in practice

They do different jobs: at the table, paper governs service pace, narrative and suggestive selling; on the screen, the QR handles delivery, accessibility, price changes and analytics. The verdict is BOTH, each in its role. Dark kitchen against physical restaurant is a cost-structure debate, not a trend debate. A hidden kitchen erases premium-zone rent and erases the dining room, true, yet it also erases walk-in demand, and whoever operates without a dining room leaves 100% of demand in third-party hands. Owned data is the asset. Eighteen hundred and fifty customers with name, frequency and ticket are worth more than any ranking position inside the app, because that series is what commercial banks use to build alternative scoring for an MSME without collateral.