Rappi delivery strategy: the mistakes burning your margin and the right method

Step 1: settle the channel math before you price anything

Before touching the digital menu, pull the last three months of actual Rappi settlement statements and divide them by gross billed volume: that ratio is your effective channel rate, and it almost never matches the commission you were quoted at signing. In Colombia the contractual range runs between 18% and 30% plus VAT, yet once you add live promotions, refunds for incomplete orders and the days of paid visibility, the effective rate at the Cali grill house mentioned earlier reached 41% of gross. The DELIVERABLE here is a single cell in your sheet: effective channel rate, expressed as a percentage with two decimals. You verify it by adding three consecutive fortnights and checking that the weighted average does not drift more than two points between them; if it drifts further, you have promotions running that you do not even remember approving. Packaging cost belongs inside the food cost of the delivery dish, and that single accounting decision flips the verdict on half your menu.

Step 2: cost each dish with packaging inside, never on the side

A pulp container with lid, the bag, the tamper sticker and cutlery add between 900 and 2,100 pesos per order depending on format, and you pay that money, not the app. Take your eight best sellers on Rappi, write down the recipe cost, add full packaging, then compare against the current app price minus the effective rate from the previous step. What remains is real contribution per dish. At the grill house, the 38,900-peso platter left 1,200 pesos of contribution, roughly 3.1%, while the dining room ran at 28%. Your deliverable is an eight-row table showing contribution in pesos and in percentage terms, sorted from highest to lowest. Mirror pricing means publishing the same dish on the app with a markup that absorbs the effective rate, and it is the only clean way to run the channel without donating margin. The formula is blunt: app price equals dining-room price divided by one minus the effective rate in decimal form.

Step 3: build the mirror price with an explicit channel markup

At a 28% rate, a 30,000-peso dish in the dining room goes live on Rappi at 41,700, not at 30,000. Large regional chains have applied this differential since 2021 and nobody has punished them for it; independents keep avoiding it out of fear that customers will compare. Customers compare less than you fear, and what they truly register is food arriving cold. The deliverable is the republished digital menu with the new prices plus a dated screenshot of every category, because Rappi takes between 24 and 72 hours to propagate bulk changes. A profitable delivery menu carries between 40% and 60% fewer items than the dining-room version, because every dish that travels badly charges you twice: the refund and the rating. Fried food that steams itself, tempura, thin fries, plated desserts with cream and anything depending on texture at minute eight leave the digital menu without debate.

Step 4: cut the digital menu down to what travels well

Cross the table from step 2 against the complaint rate per item that the app shows in your operations panel, and remove everything under 15% contribution or above 3% complaints. At the grill house we pulled fourteen of thirty-two items, and the average ticket climbed from 34,200 to 41,800 pesos because customers migrated toward the platters that survived the trip. Your deliverable is the digital menu with a before-and-after item count, plus the signed list of withdrawn references. A permanent 20% discount switched on to «rank higher» is not marketing, it is an extra commission you charge yourself. The operational difference is that a discount applies to 100% of your volume, while a visibility campaign with a closed budget has a ceiling and an end date. If you plan to invest in placement inside the app, set a monthly cap in pesos that stays under 4% of your channel revenue, ask your account executive for the incremental orders attributed to that campaign, and switch it off when the incremental fails to cover at least three times the spend.

Step 5: buy visibility with a closed budget, not open-ended discounts

Consumer behaviour backs the bet: Deliveroo reported a record frequency of 3.5 orders per month per consumer across the UK and Ireland during 2024, which confirms that the channel repurchases only when the experience holds up. The deliverable is a signed monthly budget with its shutdown rule. How long your kitchen takes to mark an order ready is the operational variable that moves your algorithmic placement most, and it gets measured in minutes, never in impressions. Put a visible timer on the line and log, across two weeks, the gap between acceptance and ready-marking, keeping lunch separate from dinner. A sensible target for a volume kitchen sits between eight and fourteen minutes depending on format; past eighteen, the courier waits, the food cools and the rating slides. Economics on the other side of the counter are unforgiving too: according to Gridwise, DoorDash couriers averaged US$12.23 per hour in 2024, down 3% year over year, and a courier who loses ten minutes at your door learns not to come back.

Step 6: treat dispatch time as a cash figure

Your deliverable is the median and 90th percentile dispatch time per daypart, with the three slowest dishes listed. Four mistakes explain nearly every failure of this playbook, and none of them is technical. First: reading the app's gross report and believing that is your revenue, when the number that matters sits in the fortnightly settlement. Second: raising app prices without trimming the menu, which makes the badly travelling dishes more expensive and accelerates the ratings drop. Third: replicating dining-room timings in the channel while expecting a different outcome. And the fourth, the costliest one: switching on promotions from the panel without running them through costing, a habit that turned 41 million pesos of monthly revenue at the Cali grill house into an operating loss. Diego F. Parra insists at Masterestaurant on reading the settlement before the report, because the channel only turns profitable once you stop watching sales and start watching what actually lands in the bank account.

Closing checklist: how to know it landed

You know the strategy is properly built when you can answer six questions with an exact number and without opening the app. One: what is your effective channel rate for the latest fortnight. Two: how many pesos of contribution each of your eight main dishes leaves after commission and packaging. Three: which markup factor you applied to mirror pricing and whether the digital menu reflects it across every category. Four: how many items you withdrew and how far the average ticket moved since then. Five: what your monthly visibility cap is and which rule shuts it down. Six: what your median dispatch time is at the lunch peak. If any of the six gets answered with «roughly» or «I'd have to check», that is the step still missing. The minimum defensible target is 22% contribution after commission and packaging across 80% of live references. The first difference is accounting, and almost nobody sees it.

Three differences that decide whether the channel funds you or drains you

In the dining room you charge the menu price and you receive the menu price; on Rappi you charge the menu price and receive between 70% and 82% of it, because commission comes off before settlement. A dish engineered at 30% food cost lands at 38% or 41% effective food cost without a single recipe change. Mirror pricing is the clean fix — the same dish carrying a channel markup that absorbs commission, a practice large chains have run since 2021 while independents still avoid it, fearing customer comparison. Customers compare far less than owners imagine; what they do notice is cold food. The second is operational. A seated guest waits and forgives two minutes; a delivery customer watches a countdown while a courier penalises your storefront for every minute of delay. One Italian kitchen in Bogotá listed 47 items and averaged 31 minutes to dispatch; cutting to 12 dishes with dedicated mise en place brought that to 19 minutes and lifted the rating from 4.2 to 4.8 over seven weeks.

Three differences that decide whether the channel funds you or drains you — in practice

The food never changed. What changed was how many decisions the line had to make per ticket. The third is strategic, and it answers the question every owner eventually asks about dark kitchen vs physical restaurant. A ghost kitchen strips out prime rent, servers and furniture, but it also strips out brand, foot traffic and any chance to sell an experience; break-even drops while algorithm dependence hits 100%. My position is firm: if you already hold a lease with an underused kitchen in off-peak hours, the correct play is a virtual brand on that same kitchen — its own menu, its own costing, measured as a separate business unit for three months before a single peso goes into construction.