Delivery zone and radius: the kilometre that eats your EBITDA

1. What delivery radius is actually profitable for a restaurant?

A profitable radius ends where the contribution margin per order stops covering the cost of serving it, and in most urban kitchens that point lands between 3 and 4 kilometers, not the 8 the app enables by default.

The gap between those two numbers explains much of why sector net margin sits between 3% and 9% (Statista) while well-shaped ghost kitchens hold 10% to 30% (OysterLink, 2025). An order at 7 kilometers does not cost twice what one at 3.5 costs: it costs more, because during peak hours it locks up a courier for twenty minutes, stretches the promised window, drags the rating down and, with it, your position in the marketplace ranking. The cost is COMPOUND, and that compounding shows up on no single line of your income statement. If you bill less than half a million dollars a year, close the radius at 2.5 kilometers and stop negotiating it over the phone when an order comes in from 6.

2. Under USD 500,000 a year: one core, 2.5 kilometers, no exceptions

At that volume you have one kitchen, two or three people on the line and zero slack: every distant order you accept steals capacity from three nearby ones you could have delivered hot. The operating threshold I use with these operators is simple and measurable: if more than 15% of a week's orders exceed 30 minutes door to door, the radius is wrong, not the courier. Keep the scale of the channel in mind before overexposing yourself to it: the delivery app market closed 2024 at USD 110 billion, growing 15.5% (Business of Apps, 2025), and even so the small band wins by concentrating, not by covering map. An operator in this band loses 6 to 9 points of contribution margin not because of food cost, but because of delivering to customers sitting 11 minutes farther out than the kitchen can absorb at peak.

3. USD 500,000 to 1 million: where 6 to 9 points vanish unnoticed

The right move here is splitting the radius into two rings with different pricing: a core from 0 to 3.5 kilometers with the full menu, and an outer ring from 3.5 to 5 with a minimum ticket 35% higher and no dishes that degrade in transit. Diego F. Parra keeps pushing operators at Masterestaurant to check this against the weekly cash, never against the map: measure contribution margin per order in each ring for four weeks and amputate whichever ring fails to cover its own delivery cost. The outer ring almost always adds volume and subtracts profit. Past a million dollars a year, the answer to demand falling outside your radius is not stretching the radius: it is opening a second production point. Initial investment in a ghost kitchen runs USD 75,000 to 200,000 (OysterLink, 2025), a figure you recover faster than the cost of sustaining a year of long deliveries with a deteriorated rating.

4. Above USD 1 million: a second kitchen before a wider radius

The trigger I recommend has two variables: 40 weekly orders rejected or delivered outside window in one zone for eight consecutive weeks, plus density enough for 6 orders per peak hour inside that polygon. Without both conditions together, the second kitchen is a real estate bet dressed up as expansion. With them, it is arithmetic. Above 5 million a different profile shows up: the large-format themed restaurant or the brand signed by a media chef, where the promise of the experience does not travel well on a motorbike. That operator should restrict the delivery radius to 2 kilometers and build a separate delivery menu, because the reputational damage of a signature dish arriving lukewarm costs more than the orders left on the table. The profitable route for monetizing distant demand is the virtual brand: today 32% of restaurant expansion strategies run through virtual brands (Technomic, via Apicbase, 2025), and 86.9% of US virtual brands operate a hybrid model on existing kitchens (Locmatic, 2024).

5. Above USD 5 million: the high-end operator and the format problem

You cover more map under another name, not under the expensive one. Once a group crosses ten million dollars a year, leaving the radius to each local manager is the most expensive silent leak in the portfolio. That decision moves up to the board table and gets made with overlapping polygons: no location should share more than 20% of its service area with another under the same brand, and the radius gets recalculated each quarter against real order density, never against a franchisee's ambition. Delivery Hero billed €12.8 billion in its delivery segment in 2024, up 22% year over year (Delivery Hero, Q4 and FY 2024), and DoorDash moved USD 21.3 billion of marketplace GOV in the fourth quarter alone (DoorDash, 2024): those platforms optimize their network, not yours. Someone on your side has to do it. Widening the radius soothes the anxiety of a slow week and wrecks the economics of the next one.

6. Wide coverage is emotional insurance; density is an asset

According to Alex Canter, founder of Nextbite and operator of Canter's Deli, the advantage of virtual brands lies in squeezing idle capacity out of a kitchen that already exists, and that logic breaks the moment a courier carries twenty minutes of that capacity into a neighborhood where you have three customers. Run the whole scenario: double the radius, you quadruple the area, order density per square kilometer drops to a quarter, average delivery time rises, ratings fall, the algorithm surfaces you less, and you end up with the same revenue spread over twice the logistics cost. It is a trap you keep paying for months. Open the platform report, filter by distance and calculate contribution margin per order across three bands: 0 to 3 kilometers, 3 to 5, and beyond 5. You need each band's average ticket, the commission actually charged, the shipping cost you subsidize and the food cost of the top seller in each zone.

7. What to measure Monday morning

Operators under 500,000 almost always find the beyond-5-kilometer band working below variable cost, while the 500,000 to 1 million band finds it barely at breakeven, which is worse because it does not hurt. China's market projects USD 539.87 billion in delivery revenue for 2026 (Statista, 2026); your problem is not the size of the channel, it is how far your van goes with money on top. The traditional radius is defined by what the platform allows; the MR radius is defined by the point where contribution margin per order crosses the cost of serving it. Two different numbers, and they almost never match. An order at 7 km does not cost twice an order at 3.5 km: it costs more, because at peak it locks the rider, stretches the delivery window, pushes the rating down and, with it, ranking visibility inside the app.

8. What separates a profitable radius from a decorative one?

The real cost compounds. Broad coverage is psychological insurance against weak demand; density is a financial asset.

According to Alex Canter, founder of Nextbite and operator of Canter's Deli, the edge of virtual brands lies in squeezing idle capacity out of a kitchen that already exists, not in covering more map with the same kitchen. Territory is defended with a local digital engine: a complete Google Business Profile, recent 5★ reviews and paid media geotargeted to the serviceable polygon. An ad shown 9 km from the kitchen buys traffic the operation cannot monetise. In dark kitchen vs brick and mortar, the physical site pays for seats and the digital one pays for distance. Confusing those two economies is the error I keep finding in the expansion plans that reach my desk.