The own-fleet delivery math most restaurants never do

A restaurant doing 200 delivery orders a week has three ways to get food to doors: the marketplace (DoorDash/Uber Eats), gig fulfillment on their own site (DoorDash Drive and equivalents), or their own drivers. Most restaurants pick one by gut feel and never run the numbers. The numbers are the whole story.

The three models, honestly priced

Marketplace. You list on the app, they bring the customers and the drivers, and they take roughly 15–30% per order depending on the plan. On a $30 order, that's $4.50–9.00 gone before food cost. What you buy: discovery and zero logistics. What you lose: margin, the customer relationship, and the data.

Gig fulfillment (Drive-style). Orders come through your own site; a gig driver fulfills each delivery for a flat per-delivery fee (DoorDash's self-delivery page lists a $6 flat fee when engaging Dashers; Drive On-Demand is a flat fee per order that varies). No commission on the order itself. What you buy: logistics without employees. What you lose: $6-ish per order, every order, forever — and control over the handoff.

Own fleet. Your drivers, your insurance, your scheduling headache. The per-order cost is driver time + distance + packaging, and it's the only model where the marginal cost per order can drop below $3 once volume is steady.

The math that decides it

The honest comparison is cost per order at your volume:

• Under ~50 delivery orders/week: gig fulfillment usually wins. A part-time driver's fixed cost spread over 40 orders is brutal; $6 flat fees are cheaper than a scheduled human standing around.
• ~50–150/week: the crossover zone. This is where you must actually compute it — driver hourly cost × hours needed ÷ orders, versus the flat fee. Most restaurants in this band are surprised in one direction or the other.
• 150+/week: own fleet usually wins, and it's not close. At 200 orders/week, the difference between $6/order gig fees ($1,200/week) and a driver crew costing $700/week is $26,000 a year. That's a kitchen renovation.

The marketplace model is a separate question — it's a customer-acquisition channel wearing a delivery uniform. Compare its commission against what you'd pay to acquire those customers any other way.

What the spreadsheet misses

Three costs don't show up in per-order math:

1. The customer. Marketplace orders belong to the marketplace. Your-site orders — gig or own-fleet — belong to you. The lifetime value of a direct customer is the quiet fortune in this business.
2. The handoff. Your driver hands over your bag with your standards. A gig driver is a stranger to your food. Quality complaints track this difference.
3. The bad night. Gig networks surge-price and deprioritize when it's busy — exactly when you need them most. Your driver is yours on your worst Friday.

The honest conclusion

There is no universally right answer — there's a right answer at your volume, this quarter. Run the per-order math with your real numbers, include the customer-ownership value, and re-run it every six months. The restaurants that do this tend to land on the same arc: marketplace for discovery, own site for regulars, gig fulfillment while small, own fleet when the volume justifies it.

Delivery isn't a single decision. It's a ladder. Climb it when the math says so.