Delivery commissions killing your margin: the BEFORE and the AFTER, step by step
Delivery commissions killing your margin are not fixed by negotiating the percentage, they are fixed by changing where the order comes from: move 25% to 40% of your digital volume to a direct channel within 90 days —Google Business Profile with your own order button, WhatsApp and a domain you control—, price the aggregator menu 18% above dine-in, and keep the marketplaces alive purely as a discovery window. In 2026 a food marketplace charges between 15% and 30% of the ticket, while a direct order costs 2.9% to 6% in payment processing plus delivery: every point you shift is worth about 7,200 USD a year in recovered margin on a location doing 40,000 USD of digital sales per month.
A fast-food operator in Medellín was billing 41,300 USD a month through aggregators and losing money on every single order. Nobody had looked at it per unit: 28% commission on a 9.10 USD average ticket, free delivery promotions the algorithm rewarded with more visibility, and 0.64 USD of burger packaging that never made it into the recipe card. Declared food cost was 30%; the real delivery food cost, counting double-cooking waste and remakes after complaints, sat near 36%. With 28 points of commission stacked on top, the order was negative before a single kitchen shift was paid.
That pattern repeats across the whole foodtech category, and it is not a negotiation problem with the platform: it is a channel architecture problem. Delivery aggregators do their job well, which is charging you for demand they captured; the owner's mistake was handing over the entire customer relationship and never building a path of his own. Once that path exists —an optimised Google Business Profile, a domain you own, an opted-in WhatsApp base— commission stops being a mandatory tax and becomes an acquisition cost you choose to pay, order by order.
Diego F. Parra keeps hammering one distinction at Masterestaurant that reframes the whole conversation: the aggregator is not your sales channel, it is your DISCOVERY channel. It exists so someone who had never heard of you tries your food once. What happens next —whether that customer returns through your site, your WhatsApp or the map— is what decides whether your ghost kitchen earns money or merely moves volume. And that second half gets built with local SEO, not with discounts.
Side-by-side comparison
| Before (100% aggregators) | After (mix with direct channel) | |
|---|---|---|
| Average commission per order | ✕28% of ticket (9.10 USD → 2.55 USD) | ✓11.4% weighted (65/35 mix) |
| Cost of a direct transaction | ✕No direct channel exists (0 orders) | ✓3.4% processing + 1.80 USD own delivery |
| Contribution margin per order | ✕−0.42 USD (loss per unit) | ✓+1.96 USD after 90 days |
| Digital menu pricing | ✕Same as dine-in (0% adjustment) | ✓+18% over dine-in, aggregator only |
| Source of repeat orders | ✕94% marketplace, 6% phone | ✓37% direct (GBP, web, WhatsApp) |
| Customer data you own | ✕0 owned contacts | ✓1,840 opted-in contacts in 6 months |
| Google Maps visibility | ✕Profile with no food category, no menu | ✓Complete profile, 4.7★ and own order button |
| Real food cost of the channel | ✕35.8% (packaging and waste off-card) | ✓30.6%, packaging costed dish by dish |
Step 1: measure your per-order margin on the delivery channel before touching anything
Your first deliverable is a sheet showing the contribution margin of ONE average aggregator order, and until that number exists everything else is opinion. Take the fast-food operator in Medellín billing 41,300 USD a month through platforms: average ticket 9.10 USD, a 28% commission that takes 2.55 USD, real channel food cost at 36% —not the declared 30%, because double-cooking waste and remakes from complaints never made it into the spec sheet— plus 0.64 USD of packaging that had sat outside the costing for two years. Add it up: 3.28 for food, 2.55 for commission, 0.64 for packaging. That leaves 2.63 USD to cover kitchen labor, energy and rent on an order the register logs as a healthy sale. Verify it this way: if you cannot tell me the margin in USD of your average delivery ticket, step 1 is not done.
Step 2: reprice the delivery menu 15% to 20% above the dining room
Charging the same in the dining room and in the app is the costliest design error in this category, and it gets fixed in one afternoon. The right markup sits between 15% and 20% over dine-in pricing, a range that by 2026 is already normalized and that customers accept because they are paying for convenience, not for your food. On the 9.10 USD ticket, an 18% markup adds 1.64 USD that goes straight to contribution margin, and only the person who ordered through the aggregator carries it. Watch the mechanics: do not raise the delivery fee, raise the DISH PRICE, because the fee belongs to the platform and its algorithm punishes expensive shipping. The deliverable is a separate menu, loaded into every app, with prices different from your printed card. Verify it by ordering from your own phone. An aggregator charges on gross sales while your own channel charges per transaction, and that accounting nuance decides your year.
Step 3: open your own channel with Google Business Profile, a domain and WhatsApp
A payment gateway at 3.4% plus a flat 0.30 USD takes 0.61 USD from the same 9.10 ticket against 2.55 in commission: a gap of 1.94 USD, roughly half the contribution margin of a properly costed burger. Multiply by 4,500 monthly orders and we are talking about 8,730 USD walking out the door. Three pieces build it: a Google Business Profile listing whose order button points to YOUR domain rather than the app's; a website of your own carrying the menu and the gateway; and a WhatsApp Business line with a catalog. Verify the deliverable by searching your name on the map from an incognito session and reaching your site in two taps. The aggregator is not your sales channel, it is your DISCOVERY channel, and that distinction —which Diego F. Parra repeats in every Masterestaurant diagnosis— separates a ghost kitchen that earns money from one that merely moves volume.
Step 4: turn the aggregator's customer into yours from inside the packaging
The platform exists so that someone who never heard of you tries your food once; what happens next is your call. The cheapest tool is physical: a card inside the bag with a QR pointing to WhatsApp and a real incentive —a free side on the first direct order, not a generic 10% nobody redeems—. At 4,500 monthly orders and a modest 6% conversion, that is 270 customers a month migrating to a channel where you pay 0.61 USD instead of 2.55. The deliverable is a permission-based contact base, and you verify it by counting how many ordered direct the following month. The operating target is moving between 25% and 40% of volume to your own channel within 90 days, chased through a weekly two-line board: aggregator orders and direct orders. Do not aim for 100%, and do not try.
Step 5: set the channel-mix target and chase it week by week
Aggregators move a market Statista sized at roughly USD 1.4 trillion globally in 2025, with the platform-to-consumer segment at USD 96,864 million during 2024 (Statista); that demand is real and local SEO alone will not capture it. What you can do is stop paying for it twice. Once the direct channel reaches 30%, commission stops being a mandatory tax and becomes an acquisition cost you choose to pay order by order, and from that position the conversation with your platform account manager changes tone. The deliverable is the board. You verify it on Mondays. Four failures account for nearly every collapse along this route, and none of them is technical. First, switching the aggregator off the moment your own channel shows signs of life: it kills discovery and cash drops 40% within two weeks while the direct base still cannot hold weight. Second, chasing visibility through aggressive in-app promotions —the free delivery in the Medellín case was exactly that, and the algorithm rewarded it precisely while each order went negative—.
The mistakes that sink this guide when it gets executed halfway
Third, and quietest of all, repricing the delivery menu and forgetting packaging: 0.64 USD per burger looks like nothing until you multiply by 4,500 and 2,880 invisible USD a year surface. Fourth, asking for customer data while offering nothing back. If your card says «follow us», leave it out of the bag; it costs printing and converts nobody. Assume you leave the channel alone because the volume feels reassuring. The market keeps growing —Delivery Hero reported segment revenue of €12.8 billion in 2024, up 22% year over year (Delivery Hero, FY 2024 results)— so your app orders will probably rise. With a negative per-unit margin, more volume means losing faster, and cash flow hides the hole for a few months because platforms settle on a lag. Then comes the contract renewal with two extra points of commission, or a jump in beef prices, and at that moment the business holds no lever at all: it cannot raise prices without losing ranking, cannot cut commission without owned volume, and has nobody to write to.
Counterfactual: what happens if you decide to change nothing
That is the trap. The paradox of delivery is that the platform which brought you the customers is the same one keeping you from knowing them, and it only resolves by building the parallel path while the aggregator's flow still covers payroll. Six checks tell you whether this guide was executed or merely read. One: a sheet exists with the margin in USD of your average delivery ticket, packaging and waste included. Two: the app menu runs 15% to 20% above the dining room, and you confirmed it by ordering from your phone. Three: the Google Business Profile listing carries an order button pointing to your own domain. Four: the site charges through a gateway and you know your exact cost per transaction —3.4% plus 0.30 USD is the market reference—. Five: the permission-based WhatsApp base grew and you can state the number. Six: the weekly board shows direct orders climbing toward 25%.
Closing checklist: how to know everything landed
If all six hold, stop negotiating commission and sit down to cost your second location; you already have the margin to pay for it. The aggregator charges on gross sales; your own channel charges on the transaction. It sounds like an accounting nuance and it is not: on a 9.10 USD ticket with tax included, 28% takes 2.55 USD, while a processor at 3.4% plus 0.30 USD fixed takes 0.61 USD. That gap of 1.94 USD is precisely half the typical contribution margin of a properly costed burger. Multiply by 4,500 monthly orders and you will see why the problem does not live in the kitchen. Charging the same price for dine-in and delivery is the most expensive design mistake in the category. A delivery menu should carry a 15% to 20% uplift, and by 2026 customers understand it perfectly well; whoever skips it is subsidising the aggregator with his own profit.
The four differences that decide the outcome
The uplift does not make your brand expensive, it protects the dine-in price, which is where your ticket actually performs. Discovery and repeat purchase are different businesses with different economics. Paying 28% for a brand-new customer can be a sensible acquisition cost —compare it with what a geo-targeted campaign costs per conversion— but paying it for that same customer on his eighth order is simply giving profit away. Split the two flows and the number moves on its own. A local digital engine swaps intermediation for position. A complete Google Business Profile, with the right primary category, menu loaded, fresh photos every fortnight and a reply to every review, shows up in the Maps pack for «food near me» without paying per click. That traffic, landing on your own domain with a two-tap checkout, is the only source of volume that does not charge you commission for growing.
Criterion by criterion: what each model wins
What happens when the aggregator is your only channelBefore
- Commission, 15% to 30% depending on country and plan, is charged on the gross ticket with tax included: you pay a percentage on money that was never yours.
- Menu pricing matches dine-in, so the 2.55 USD of commission comes straight out of your contribution margin.
- Two-for-one deals and free shipping are almost always funded by the restaurant, and the algorithm ties your visibility to keeping them running.
- Delivery packaging —box, bag, tamper seal, cutlery— adds 0.45 to 0.90 USD per order and rarely appears on the recipe card.
- You hold neither phone nor email of the customer: if the platform raises two points tomorrow, there is nobody to tell.
- The Google Business Profile sits abandoned, with no menu and no fresh photos, so anyone searching your name lands back on the marketplace.
What a local digital engine of your own changesMasterestaurant
- A direct order costs 3.4% processing plus real delivery: on the same ticket that is 2.11 USD versus 2.55 USD, and delivery stops being a percentage and becomes a fixed cost you optimise by zone.
- The aggregator menu runs 18% above dine-in, standard practice in 2026, and the customer who compares discovers that ordering from you directly is cheaper.
- The «Order online» button on Google Business Profile points to your own domain rather than the marketplace, capturing brand searches without paying intermediation.
- Every direct order leaves a name, a phone and a frequency, so the second purchase is triggered by a 0.02 USD message instead of by 28% commission.
- Five-star reviews work for a profile you own and lift your position in the Maps local pack, which is free recurring traffic.
- The aggregator stays alive as pure discovery: a stranger walks in, the bag carries a magnet to the direct channel, and the second visit already belongs to you.
Side-by-side comparison
| Before (100% aggregators) | After (mix with direct channel) | |
|---|---|---|
| Average commission per order | ✕28% of ticket (9.10 USD → 2.55 USD) | ✓11.4% weighted (65/35 mix) |
| Cost of a direct transaction | ✕No direct channel exists (0 orders) | ✓3.4% processing + 1.80 USD own delivery |
| Contribution margin per order | ✕−0.42 USD (loss per unit) | ✓+1.96 USD after 90 days |
| Digital menu pricing | ✕Same as dine-in (0% adjustment) | ✓+18% over dine-in, aggregator only |
| Source of repeat orders | ✕94% marketplace, 6% phone | ✓37% direct (GBP, web, WhatsApp) |
| Customer data you own | ✕0 owned contacts | ✓1,840 opted-in contacts in 6 months |
| Google Maps visibility | ✕Profile with no food category, no menu | ✓Complete profile, 4.7★ and own order button |
| Real food cost of the channel | ✕35.8% (packaging and waste off-card) | ✓30.6%, packaging costed dish by dish |
The figures you should decide with
“We closed July with 41,300 USD through platforms and a 1,900 USD loss. We raised the aggregator menu 18%, moved packaging into the recipe cards and pointed the Google Maps order button at our own website. By month four, 37% of orders came in direct, weighted commission dropped from 28% to 11.4%, and contribution margin per order went from minus 0.42 USD to plus 1.96 USD. What surprised us most: total sales did not fall, they rose 6%, because the Maps profile started bringing people who had no idea we existed.”
How it is done, step by step, with a deliverable and a numeric checkpoint
Before step one you need five figures on the table, and without them every decision is opinion: the exact commission of each platform read off your contract, not the one you remember; the average digital ticket of the last 90 days; full packaging cost per order, adding box, bag, seal and cutlery; food cost per dish for your ten delivery bestsellers; and monthly order count per platform. Deliverable: one sheet with those five columns. Checkpoint: if your real weighted commission exceeds 22% or your delivery food cost passes 32%, you already know there is a loss per unit and the rest of this guide is urgent rather than optional. The typical mistake here is reusing dine-in food cost, which ignores double cooking and remakes after complaints.
Take your bestselling dish and pull it apart to the cent: platform selling price, minus commission applied on gross, minus food cost, minus packaging, minus the proportional share of promotions you funded that month. What remains is your contribution margin per order, and in seven out of ten audits that figure comes out negative while the location bills nicely. Deliverable: a delivery unit economics table with your top ten items and their unit margin. Numeric checkpoint: no digital-channel dish should sit below 1.50 USD of unit contribution; whatever falls short leaves the delivery menu or rises in price this same week. The typical mistake is prorating commission over net sales instead of gross, which flatters the number by two to four points.
This step is frightening and it returns the most money. Load a platform-exclusive price list with an 18% uplift over dine-in, applied to every item except bottled drinks, where the customer compares with the corner shop and does notice. Leave dine-in pricing untouched: the uplift protects that ticket, it does not inflate it. Deliverable: digital menu loaded and published on every platform with differentiated pricing. Checkpoint: on day 21 measure order volume; a drop under 8% with 18 extra points of price means your demand is inelastic and you had been leaving that money on the table for years. The typical mistake is raising only the hero dishes and leaving the sides alone, which is exactly where the aggregator squeezes margin.
Your Google Business Profile is the most profitable free digital asset you own, and in 2026 it still sits abandoned in most ghost kitchens. Set the right primary category —«hamburger restaurant», not plain «restaurant»—, load the full menu with prices, upload eight new dish photos a month, switch on delivery and pickup attributes, and post a weekly update. The decisive part: the «Order online» button must point at YOUR domain rather than the marketplace, configured from the profile's ordering panel. Deliverable: a profile at 100% completeness with your own ordering link. Numeric checkpoint: within 60 days profile views should grow at least 40% and calls from the profile 25%. The typical mistake is leaving the platform's ordering link live, which keeps charging commission on traffic you generated yourself.
You do not need an app, you need a neighbour to order in under thirty seconds. Your own domain, a web menu loading in under two seconds on mobile, a local payment processor at 3.4%, and WhatsApp Business with a catalogue and quick replies for whoever prefers to type. Delivery gets solved with an in-house rider inside a 3-kilometre radius or with an on-demand fleet charging per drop, 1.50 to 2.20 USD, instead of per percentage. Deliverable: direct channel live and tested with twenty internal orders. Checkpoint: total cost of a direct order —processing plus delivery— must stay under 14% of the ticket; above that, your delivery radius is too wide. The typical mistake is launching the direct channel before solving delivery and then cancelling orders, which burns customer trust at the exact moment you were capturing it.
This is where the aggregator stops being a tax and starts being an acquisition channel. Every bag leaving through a platform carries a card with a QR code and an offer redeemable only on your site: a free side on the next direct order, never a cash discount, because the side costs you 0.70 USD and the discount costs you margin. Ask for the data with explicit consent. Deliverable: 100% of orders leave with the physical magnet and a contact base that grows. Numeric checkpoint: by month three you should be capturing at least 12% of platform orders as owned contacts, and by month six the direct channel should represent 25% to 40% of digital volume. The typical mistake is offering a generic discount with no expiry, which trains the customer to wait for a deal rather than to come back.
With the direct channel running, volume stops depending on the marketplace algorithm and starts depending on your position on the map. Ask for the review at the right moment —inside the bag, not by message the next day—, reply to every one within 24 hours including the bad ones, and sustain a short-radius geo-targeted campaign, three to five kilometres, with a small constant budget instead of spikes. Deliverable: a weekly review routine and a live local campaign. Checkpoint: public rating at or above 4.6★ with at least fifteen new reviews a month, and cost per acquired order under 1.80 USD; above that threshold the problem is the landing page, not the ad. The typical mistake is buying reviews, which Google detects and punishes by suppressing the entire profile.
And with AI?
Optimize channels, pricing and unit economics of your dark kitchen. Diego F. Parra is an expert in AI applied to restaurants.
Free tools to apply this now
The method tools that hold this change together
None of these six steps works without three management pieces underneath: a business model that knows which channel feeds which, a projection telling you how much direct volume you need for the shift to hold, and a weekly cash control that catches the leak before the month closes red. The MASTERESTAURANT method sequences them in that order and no other, because moving channel volume without cash under control merely changes where the money is lost.
Diego F. Parra — International consultant, expert in creating and scaling restaurants and in AI applied to restaurants, foodtech and HORECA. Methodology applied in 8.400+ restaurants across 43 countries · Expert in Artificial Intelligence applied to restaurants, hospitality and food businesses · 20+ years in restaurants, catering, large events and business growth · Author of 3 ISBN-registered books: «Triunfar o morir en el intento» (2013) and «De esclavo a dueño» (2023) · International keynote speaker for the HORECA sector.
Questions every owner asks before moving the first price
What does a delivery aggregator really charge a restaurant in 2026?
What does a delivery aggregator really charge a restaurant in 2026?
Between 15% and 30% of the gross ticket depending on country, contracted plan and whether the platform supplies the rider. Basic plans sit at 15-18% with minimal visibility; plans with delivery and featured placement reach 30%. Read your contract: commission usually applies to the tax-inclusive price, which adds two to three effective points.
Will the algorithm punish me if I raise my price on the platform?
Will the algorithm punish me if I raise my price on the platform?
Not for the price itself. The algorithm rewards prep time, acceptance rate, rating and promotion participation, not being the cheapest. An 18% uplift with a tight operation usually leaves your position untouched; what really sinks visibility is rejecting orders or running late on your promised time.
Is it worth leaving delivery aggregators entirely?
Is it worth leaving delivery aggregators entirely?
Almost never, and that is the opposite mistake. The marketplace brings you strangers your brand alone cannot reach, and that discovery carries real value. The right call is capping its weight at 60% to 75% of digital volume, using it as pure acquisition and moving repeat purchase to your own channel, where an order costs 14% instead of 28%.
What if my ghost kitchen has no visible storefront or address for Google Maps?
What if my ghost kitchen has no visible storefront or address for Google Maps?
You can register the profile as a service-area business without a public address, defining a delivery zone. You lose the classic Maps pack, but you keep brand search, reviews and your own ordering button. For a virtual brand with no façade, the weight shifts to your website, geo-targeted ads and WhatsApp repeat purchase.
Sector data 2026 (official sources)
Verifiable industry benchmarks from official, non-commercial sources (government, industry associations, market research) - not competitors.
| Metric | Benchmark 2026 | Source |
|---|---|---|
| Robots de Serve Robotics a desplegar en Uber Eats | hasta 2.000 robots | Serve Robotics — Form 8-K FY2024 (SEC) |
| Cuota conjunta de Serve, Starship y Nuro en flotas globales 2024 | 18% | Mordor Intelligence — Autonomous Delivery Robots Market 2024 |
| Mercado de entrega de paquetes por dron en 2023 | USD 585,9 millones | Grand View Research — Drone Package Delivery Market 2023 |
| Proyección de entrega de paquetes por dron a 2030 | USD 5.238,8 millones (CAGR 38,7%) | Grand View Research — Drone Package Delivery Market 2030 |
| Entregas comerciales por dron de Zipline (abril 2024) | 1 millón (primera empresa en lograrlo) | Grand View Research — Drone Package Delivery Market |
| Unidades de drones de reparto proyectadas 2024 a 2030 | de 32.456 a 275.703 unidades | Grand View Research — Drone Package Delivery Market |
Related content
Grow your restaurant with the Masterestaurant method
Applied in +8.400 restaurants across 43 countries.