Delivery app data and reviews: what actually moves your ranking

What works better when delivery is 20% to 45% of your sales?

Work your delivery app data and reviews before pulling any other lever, because that is where the money you already paid for sits uncollected.

A neighborhood grill in Medellín climbed from 4.1 to 4.6 stars on Rappi in eleven weeks without touching the menu or prices: it corrected the declared prep time, which said 15 minutes while the kitchen took 26, and changed the packaging on the fries, which arrived soggy and cost one star every third order. Sales rose 31% while the city grew 4%. At that channel weight you are already paying 30% to 40% in effective cost per order according to ActiveMenus, so every conversion point you miss by ranking low is margin handed away. Fix the listing first, then buy ads.

A review is not a thermometer: it drives the listing

The aggregator hands out impressions based on the probability that an order ends well, and it calculates that probability from your cancellation history, your time compliance and the score customers leave. Almost every owner who reaches Masterestaurant arrives convinced of the opposite: the rating mirrors service, end of story. It does mirror it, yes, but it also FEEDS the order in which you appear when someone opens the app at 8:40 in the evening. Diego F. Parra, restaurant consultant and founder of Masterestaurant, starts every delivery diagnosis with the restaurant's cancellation rate —around 6% in poorly calibrated operations, and it should live below 2%— before looking at a single dollar of advertising. Below 4.3 stars you compete for shelf space you never see.

Best for operations under 15 tables: fix the listing, don't move into a dark kitchen

If your place has fewer than 15 tables and delivery carries a third of your sales, a dark kitchen solves a problem you do not have yet and creates two you certainly will. The 18% to 30% commission does not vanish by changing addresses; it concentrates, because with no dining room 100% of sales cross a channel that charges a toll. Add that labor costs eat 25% to 35% of revenue according to the U.S. Bureau of Labor Statistics, a figure no industrial container reduces on its own. And a brand-new listing starts with zero reviews against the 300 your corner location already banked. A hidden kitchen suits you when the dining room contributes under 20% of sales and your delivery unit economics already stand alone.

When NOT to choose the popular option?

Three scenarios turn in-app advertising —the fix everyone buys first— into burned cash that moves nothing. One: a rating below 4.2. Ads buy impressions, but the customer reads the star before tapping;

paying for traffic into a punished listing fills a leaking bucket. Two: a cancellation rate above 4%, because the algorithm penalizes broken promises harder than your budget can offset. Three: food cost above 32% with packaging booked separately, a setup where every incremental delivery order subtracts margin; at 30% to 40% effective cost per order (ActiveMenus), selling more there accelerates you toward the cliff. In all three cases the right sequence is kitchen, timing and costing. Advertising comes after.

Red flags when comparing platforms, agencies or delivery consultants

Four concrete trade signals that should change your face in the meeting. First: they promise to raise your stars without asking for your prep-time report or your cancellation report, which means they will dress up reviews instead of repairing operations. Second: they quote the nominal commission rather than the effective cost per order, which stretches to 30-40% once you count packaging, co-funded promotions and refunds, according to ActiveMenus. Third: they suggest launching five virtual brands from one kitchen, a reliable way to dilute the signal —five listings with 40 reviews rank worse than a single one with 300—. Fourth: they discuss the market without separating territories; DoorDash closed 2024 holding 60.7% share in the United States (Earnest Analytics), a number worth nothing for a decision in Bogotá or Mexico City.

What if the aggregator charges you five more commission points tomorrow?

That exercise changes everything, and it deserves a pencil before it happens. If you pay 22% commission today and your food cost runs at 30% with packaging included, five extra points leave you at a 27% toll:

the dish that used to carry operating margin now funds the platform. The answer is not to walk away —Meituan and Ele.me concentrate more than 90% of orders in China according to Mordor Intelligence, and that concentration is where the sector is heading, not a fluke—, the answer is a listing strong enough that the same order arrives with less promotional spend. A restaurant with 4.7 stars and 2% cancellation absorbs that increase; one at 4.1 and 6% does not. Resilience against commission gets built with measurable reputation, not nostalgia for the landline.

Best for those selling in markets still growing: ride the wave, forget the commission fight

If you operate in an expanding territory, your advantage lies in stacking reviews while the market fills, not in negotiating points. Vietnam grew 26% in delivery GMV during 2024 according to Momentum Works, Asia-Pacific already holds more than 41% of online food delivery (Grand View Research), and DoorDash's Marketplace grew 20% year over year that same period. In a market that widens, the listing gathering 300 verified reviews today becomes untouchable once thirty new competitors arrive next year. Real review volume first, price later. The recurring mistake is waiting for a perfect operation before asking for ratings; start with the packaging insert this week, measure the response rate after fifteen days and adjust the script based on what people actually answer.

Myth against reality: where the money leaks

MYTH: reviews are a consequence of service. REALITY: they are an input to the algorithm. Store rating feeds directly into listing order, so a restaurant under 4.3 stars competes for impressions it never sees. A review is not the mirror of your business, it is the key to the window. MYTH: a dark kitchen removes the commission problem. REALITY: it concentrates it. With no dining room, 100% of sales run through a channel charging 18-30%, so delivery unit economics must stand alone, with food cost under 32% per dish and packaging inside the costing, never as a loose expense. MYTH: more virtual brands mean more sales. REALITY: more brands dilute the signal. A ghost kitchen with five storefronts of 40 reviews ranks below a single one with 300 recent reviews, because platforms weigh volume and freshness over catalog variety.

Myth against reality: where the money leaks — in practice

MYTH: you must be on every app. REALITY: each aggregator carries real admin cost —tablet, sold-out syncing, payment reconciliation— and below 20% delivery the third app subtracts more than it adds. MYTH: the QR replaces the printed menu. REALITY: at Masterestaurant the recommendation is BOTH, each with its role. The physical menu governs service pace, menu narrative and suggestive selling at the table; the QR handles delivery, accessibility, price changes and analytics. Dropping the printed menu to save on printing trades average-ticket points for pennies of paper. MYTH: cutting prices in the app accelerates growth. REALITY: in a channel charging 22%, each dollar of discount costs roughly 1.28 dollars of lost contribution, and the bargain order brings the customer who scores hardest.