In-house delivery is a restaurant running its own drivers and its own dispatch, instead of handing every order to a third-party app. For most independent restaurants, the honest answer is a mix: own drivers for the orders you can predict, on-demand for the rest, and the marketplace apps as a billboard, not as the home of your regulars.

The hard part is deciding which order belongs to which lane.

What does in-house delivery actually mean?

In-house delivery means three jobs stay with the restaurant: taking the order on a channel you own, assigning a driver you employ, and running the delivery under your brand. The order enters your system, your staff dispatches a driver on payroll, and the bag at the door has your logo on it.

Third-party delivery hands all three jobs to someone else: the customer, the driver, and the data about the order. Most restaurants we work with use both. The question is which orders should go through your own drivers, and which should go through someone else.

How much do the big delivery apps charge today?

On the marketplace side, the two biggest apps publish tiered commissions tied to how much visibility a restaurant gets.

DoorDash lists three plans for Marketplace orders, with 15% commission on the Basic plan, 25% on Plus and 30% on Premier. Pickup is 6% on all three plans. DoorDash Drive, where the customer orders on your own channel and a Dasher delivers it, is a flat $6.99 to $10.99 per order (RestoLabs guide, updated July 27, 2026, citing DoorDash merchant pricing).

Uber Eats restructured its fees on March 11, 2026. The Lite tier moved from 15% to 20%, Plus stayed at 25% (and 30% for orders from Uber One members), and Premium remained at 30%. Pickup rose from 6% to 7% with validated in-store pricing. Uber Direct, the on-demand version where the customer orders on your site, starts at $7.99 per order (QSR Pro coverage of the fee restructure, April 2026).

These rates sit on top of the subtotal. Taxes and the delivery fee the customer pays are outside the commission. The number that lands in your bank account is lower still, after the processing fees come out.

What does a $40 order look like on each model?

A single order makes the differences concrete. The table assumes a $40 food subtotal, before tax, tip and delivery fee.

Delivery modelFee on the subtotalWhat you keep from $40Who owns the customer
DoorDash Marketplace, Basic15%$34.00DoorDash
DoorDash Marketplace, Plus25%$30.00DoorDash
DoorDash Marketplace, Premier30%$28.00DoorDash
Uber Eats Marketplace, Lite20%$32.00Uber Eats
Uber Eats Marketplace, Premium30%$28.00Uber Eats
DoorDash Drive (your site, their driver)$6.99 to $10.99 flatabout $29 to $33You
Uber Direct (your site, their driver)from $7.99 flatabout $32You
Your own driverfixed hourly cost plus mileagedepends on volumeYou

Three things stand out. The marketplace plans leave you with the smallest piece of the subtotal, and someone else keeps the customer. The on-demand options (DoorDash Drive and Uber Direct) land close in cash to a Marketplace Basic plan, but the order sits in your system and the next one does too. Your own driver only pays off when the shift stays busy.

When do your own drivers pay off?

A driver on payroll turns delivery from a per-order cut into a fixed cost. The math is simple: hourly wage, plus mileage reimbursement, plus insurance, divided by the number of deliveries that fit into the shift.

The real question for the hourly driver is whether the shift stays full. Three or more deliveries per hour, and the per-order cost of your own driver undercuts almost every third-party model. One or two per hour, and the fixed cost eats the margin fast.

Signals that in-house delivery earns the operation:

  1. A core delivery radius inside about three miles, with recurring neighborhoods and repeat customers.
  2. Lunch and dinner peaks that run two hours or more, several nights a week.
  3. A catering or large-order flow where the delivery is part of the brand, not a stranger in a jacket from an app.

Where own drivers struggle: late night, long-distance one-offs, and the hour between the lunch rush and the dinner build. That is where the on-demand options earn their keep.

Where on-demand delivery fits

On-demand delivery is the middle lane. The customer orders on your site, pays you, and gets a driver from a third-party fleet. DoorDash Drive and Uber Direct are the two biggest. The cost is a flat fee per order, so you can decide which orders are worth it.

Three practical uses:

  1. Covering the hours your own driver is off the clock, without turning down the order.
  2. Reaching outside your in-house radius for a known good customer.
  3. Catering runs that cannot fit on the shift schedule you already have, without exposing the account to a marketplace listing.

The commission math changes too. A $40 order with a $7.99 Uber Direct fee leaves about $32 on the table, close to a Marketplace Lite plan, with the difference that the customer and the order history stay with you.

What a hybrid setup looks like

Most restaurants we work with run all three channels through one screen. The deliveries queue shows every order waiting for a driver, your own drivers logged in for the shift, and the ones routed through an on-demand service.

Deliveries queue in a CGRabbit dashboard, with own drivers and on-demand assignments side by side, sample data
The deliveries queue a manager watches during the shift. Sample data.

Three details matter inside that screen:

  1. One place to see which order is late, no matter which driver has it.
  2. An auto-rule that pushes orders to DoorDash Drive or Uber Direct when no in-house driver is free, with a per-order ceiling the owner sets.
  3. A customer tracking link, the same for every delivery, so the calls asking where the order is stop coming in.

That is the version of in-house delivery that pays off. The older version, where every delivery has to go through the restaurant, breaks the first time two drivers call out on a Friday.

If this looks like how your delivery should run, our driver app and live map service page walks through the pieces, and our commission-free online ordering guide covers the site that feeds the queue.

Our take

Pick the model per order, not per restaurant. The marketplace apps are a billboard: good for finding a new customer once, poor as the home of your regulars. Your own drivers are a payroll decision tied to the shape of your week. On-demand delivery is a safety net, not a plan.

The restaurants that keep delivery profitable almost never run a pure in-house operation, and almost never give the whole channel to an app. They keep the queue in one screen, pay their own drivers for the busy windows, and let the on-demand services clean up the edges. The customer never knows the difference, and the margin shows up where it belongs.