How to increase restaurant sales on Rappi: the mistakes that drain cash and the method that fixes the margin

Step 1: calculate real per-dish profit in the delivery channel before touching anything

Before uploading a single photo you need the contribution margin of each dish AFTER commission, and that number almost never exists on the owner's paper. Take the selling price in the app, subtract the commission you are charged —in Latin America it runs between 18% and 30% depending on the plan—, subtract the dish food cost and subtract packaging, which on a hot dish with dome lid and bag costs you between 900 and 1,400 pesos. What is left is your real margin. The deliverable is a table with the 20 fastest-moving items, sorted by absolute margin in pesos, not by percentage. You verify it by cross-checking the total against the platform's biweekly settlement: if your table projects four million and the settlement pays one, the table was built wrong and has to be redone. Preparation time is the lever that moves visibility most and the one almost nobody measures with their own clock.

Step 2: bring real preparation time below 15 minutes

The app shows you an average, but that average starts when you accept and ends when you mark ready, so a kitchen that confirms in 40 seconds and hands over the packed order the moment the courier arrives frees the rider faster — and a free rider is capacity for the whole platform. That is their business, and that is why they reward you with impressions. Put a stopwatch on thirty real orders over a week, separate the ones past fifteen minutes and look at which dish is causing them: usually two or three long-cooking items you should pre-cook, move to another time slot or pull from the digital menu. The deliverable is the list of bottleneck dishes with measured times, verified when next week's average drops. Rejecting orders is the most expensive way to save money that exists in delivery.

Step 3: hold acceptance above 95% and kitchen cancellations near zero

The algorithm does not hand out visibility out of sympathy: it hands it out by probability that the order ends well, and every rejection or kitchen cancellation lowers the score that decides whether your listing shows up in row three of the first screen or on page two, where according to public industry data less than 8% of traffic ever goes. The real cause of rejections is almost always unreported stockouts. Configure the app inventory so an item switches off automatically when it runs out, and make one single person per shift responsible for turning dishes on and off. The deliverable is a weekly acceptance rate above 95%; you verify it in the merchant panel, performance tab, and if it stays below the problem is internal process, not the platform. A delivery menu with the same dine-in prices is a subsidy you are paying the application.

Step 4: redesign the digital menu with channel pricing, not the dining-room card

If commission is 28% and your food cost is 34%, the dish was born dead, and that is why I saw a March settlement with 1,940 orders, 71 million billed and an operating loss of 3.1 million: it was not a sales problem, it was an arithmetic problem. The fix is a channel differential between 12% and 18% over the dine-in price, applied with judgment: hard on high food-cost dishes, gentle on the fast movers that bring you traffic. Cut every item that lands under 25% margin post-commission and build two or three combos with a drink, where margin rises because liquid costs little and packaging does not change. The deliverable is a published digital menu with a minimum 30% margin on every item. The customer never sees your façade or smells your kitchen: they see a 400-pixel photo, an estimated time and a price compared against twenty other kitchens in the same grid.

Step 5: photos and listings that win the four-second comparison

Shoot the ten highest-margin dishes with side window light, dark background and the plate filling 80% of the frame, always horizontal because that is how the app crops it. No stock images, no photos from your sauce supplier. Descriptions stay short and concrete, with weight and side —«180 g of short rib, criolla potato and house chili»— because that cuts complaints, and a complaint about the product itself damages your rating more than a delay does. The deliverable is ten new listings published; you verify it by measuring view-to-order conversion on those items across the following two weeks against the previous two. A promotion is justified only if the post-commission margin absorbs the discount without going red, and that calculation is yours to make, not your account executive's. A permanent 2-for-1 on your fastest-moving dish is the quickest way to multiply orders and multiply losses at the same time.

Step 6: promotions and in-app advertising, with a calculated spending ceiling

Use discounts of 10% to 15% on items with margin above 40%, restricted to low-demand windows —Monday through Wednesday between 3 and 6 in the afternoon— and with an end date. In-app advertising charges per click or per order and competes against virtual brands and dark kitchens that have multiplied in every delivery zone; in the United States roughly 7,606 ghost kitchens are already operating according to OysterLink 2025. Set a monthly ceiling of 3% of channel sales. The deliverable is a sheet with incremental cost per order for every active promotion. The error that destroys the most cash is accepting the highest commission plan believing it buys visibility: it buys a little, but visibility is won mostly through time and acceptance, and those two are free. Second comes bad packaging, because a soggy fry or a spilled soup drops your rating below 4.7 and from there the listing starts falling in the ranking.

The four mistakes that sink the operation and how to avoid them

Third is keeping the full menu in delivery when the kitchen cannot handle it: every slow dish raises your average and punishes the entire storefront. Fourth, and the quietest, is not reading the settlement line by line; that is where charges for cancellations, customer compensations and promotions you never authorized show up. In the Masterestaurant method, Diego F. Parra insists on reviewing that settlement with the same discipline used for the liquor inventory. You know the work is done when you can answer six things with numbers in front of the merchant screen. Average preparation time over the last thirty days below fifteen minutes. Acceptance rate above 95%. Rating held above 4.7 with at least fifty recent reviews. Post-commission margin on every published item equal to or above 30%, verified against the latest settlement and not against your spreadsheet. Spending on promotions and in-app advertising below 3% of channel sales.

Closing checklist: how to know everything landed right

And a positive delivery operating profit, calculated separately from the dining room, because mixing both channels into one P&L is exactly what lets a losing channel hide behind a profitable one for years. If all six come out green, push the volume; if one fails, fix it before asking for more orders. An aggregator's algorithm does not hand out visibility on charm: it hands it out on the probability that the order ends well. Every signal you emit — real prep time, acceptance rate, rating, kitchen-side cancellation share — feeds a score that decides whether your listing shows in row three of the first screen or on page two, where public industry data puts less than 8% of traffic. The gap between the operator who grows and the one who stalls is rarely price. It is latency: a kitchen that confirms in 40 seconds and hands over a packed order the moment the courier arrives frees that rider faster, and a free rider is capacity for the whole platform.

Why the app rewards what it rewards?

That kitchen earns impressions because it makes the app money. Keep a courier waiting nine minutes at the door and you are costing the system cash, and the system notices.

Here sits the uncomfortable tension of the channel: you need Rappi for volume and Rappi charges precisely what you need to survive. The way out is not fighting commission, which rarely moves more than two or three points, but changing the mix you sell inside it. A delivery menu with 26% average food cost and a 48,000-peso ticket absorbs a 28% commission; that same menu at 34% food cost and a 29,000-peso ticket does not. The lever is the menu, not the negotiation. And there is a second-order effect almost nobody calculates: every rating point you gain pushes free impressions. Moving from 4.4 to 4.8 stars can double listing views without a peso of in-app advertising, while buying that same visibility through funded promotions costs between 10% and 18% of ticket.

Why the app rewards what it rewards — in practice

Reviews are the cheapest acquisition channel in delivery, and you manage them with packaging, temperature and written replies rather than budget.