Own drivers vs delivery apps: the real cost per order

Short answer: delivery strategy is one arithmetic problem. Apps charge 15–30% commission, and with processing and promotional fees the true take can pass 40% of an order, on a $40 order many venues net about $25 before food and labour. Your own driver costs wage ÷ drops per hour plus vehicle, roughly $9–11 per drop at three drops an hour. So in-house wins where orders are dense and loses where they are scattered, and the answer for most independents is a hybrid: apps for discovery and overflow, your own channel and drivers for regulars, run inside a tight radius. Here is the maths and the setup.

The same $40 order, three ways

Read it twice. Even paying your own driver, the direct order keeps $6–9 more than the app order, because the delivery fee is now revenue to you instead of a platform. And pickup, the forgotten channel, keeps nearly everything, which is why every delivery strategy should quietly push pickup for nearby customers.

The driver economics: density decides

Everything that improves this table is a radius or batching decision: deliver 10–15 minutes’ drive, not a heroic circle on a map; group nearby orders into runs; and set delivery windows on quiet nights so drops cluster instead of trickling. Where demand exists beyond the tight zone, charge that zone honestly rather than subsidising cold food.

When each model wins

• Apps only: low order volume, scattered customers, or delivery as a sideline. Rent reach, keep zero fixed cost, accept the margin.
• In-house only: dense neighbourhood demand, strong direct brand, food that suffers in a courier’s backlog (pizza built this model for a reason).
• Hybrid (most independents): stay listed for discovery, cap app dependence, and route every repeat customer to your own channel, apps are a billboard you pay per order, not a home.

The conversion playbook: rented traffic → owned regulars

The hybrid only pays if app customers actually migrate. The moves: a flyer in every app bag with a first-direct-order code; your Google profile’s order button pointed at your own page (free traffic should never pay commission, see GBP); direct-only sweeteners funded by the saved commission, lower delivery fee, loyalty points, a free-item threshold; and your list reminding lapsed customers where the better deal lives. The deep maths on why this matters is in taking orders without the app fees.

Running dispatch without enterprise software

Own-delivery logistics used to be the blocker; now it is a feature checklist. You need zones and fees drawn around your kitchen (with a minimum order per zone), quoted times that reflect real kitchen + drive time, and dispatch that fits your volume: batched runs (several orders, one loop) for steady nights, or drop-by-drop assignment with a simple route view for surges. Add driver accountability, who has what order, when it left, and delivery stops being chaos held together by a group chat.

Mistakes that burn delivery margins

• Pricing the menu identically on apps while eating 25–30% commission, most operators justifiably price app menus higher.
• A heroic radius: cold food, slow drivers, and per-drop costs that double.
• Treating the delivery fee as untouchable, it is a pricing tool per zone, not a constant.
• Never testing pickup incentives on nearby customers, the highest-margin fulfilment there is.
• Building the whole channel on one app’s traffic and never capturing a single customer relationship.

Frequently asked questions

Should my restaurant do its own delivery or use delivery apps?
Do the per-order maths. Apps charge 15–30% commission and the true cost with processing, marketing placements and fees can pass 40% of the order. In-house delivery costs a driver’s wage plus vehicle costs, which beats app commissions once you have enough orders per hour in a tight radius. Most independents land on a hybrid: apps for discovery and overflow, their own channel for regulars and peak volume.

What does third-party delivery really cost per order?
On a $40 app order at 25% commission you lose $10 immediately, and after the platform’s processing and promotional fees many restaurants net around $25 before food and labour. The same $40 order through your own ordering page costs roughly $1.20–1.60 in payment processing (from about 2.5–2.9%) plus your delivery cost. The gap per order is commonly $6–9, which across a few hundred orders a month is a manager’s salary.

How much does in-house delivery cost per drop?
Wage divided by drops per hour, plus vehicle cost. A driver at $25/hour completing three deliveries an hour costs about $8.30 per drop plus fuel and insurance, call it $9–11 all-in. At two drops an hour it is $13–14; at four, near $7. That is the whole game: in-house wins where orders are dense (tight radius, batched dispatch) and loses where a driver idles between far-flung singles.

How do I move customers from delivery apps to ordering direct?
Use the apps as advertising you convert. A flyer in every app bag with a first-order code for your own site, your Google profile’s order button pointed at your page, direct-only perks (free item threshold, loyalty points), and a lower or free delivery fee direct, funded by the commission you are not paying. Regulars switch quickly once ordering direct is one tap and visibly cheaper.

What radius should my restaurant deliver to?
Whatever keeps food hot and drivers dense, for most kitchens that is a 10–15 minute drive, not a distance on a map. A tight radius means more drops per driver-hour (better economics), hotter food (better reviews), and honest quoted times. Expanding the circle grows orders and quietly degrades all three; if demand exists further out, price that zone higher rather than subsidising it.

Do I need special software to run my own drivers?
You need three abilities: take direct orders, define zones and fees, and dispatch drivers sensibly (grouping nearby orders into a run, or assigning drops one by one with an easy route view). That used to require enterprise logistics software; it is now a built-in feature of good ordering platforms, so the practical requirement is choosing a system where ordering and dispatch already live together.