Third-party delivery costs you $7–10 per order, every order, forever. An employed courier costs roughly $20–25 an hour once you load in wage, fuel or vehicle allowance, and insurance. Whether your own fleet beats the hired one comes down to one variable: how many deliveries that courier completes per hour. The rest is detail.
The math on one courier-hour
Take a courier costing $22 an hour all-in. At 1 delivery per hour, each delivery costs you $22, which is hopeless. At 2 per hour it's $11, still worse than most third-party fleets. At 3 per hour it's $7.33, and you've matched them. Push to 4 per hour and it's $5.50: now you're 30–40% cheaper than the fleet, and the gap widens with every order.
Which means the question you actually have to answer is whether you can keep a courier at 3+ deliveries per hour. That's a density problem. It rides on order volume per hour, the size of your delivery zone, and how well orders get stacked onto one run. A compact zone doing 25+ orders on a dinner shift clears the bar without much effort. A sprawling zone doing 8 won't, and no amount of willpower fixes that.
What the per-order fee doesn't show
The fee comparison understates the real difference, and it does so in both directions. On the in-house side, count the hidden fixed costs honestly: recruiting and churn (riders turn over fast), a manager's time on scheduling, thermal bags, and the dead hours on a rainy Tuesday when you're paying wages against four orders. The dead hours sink an in-house fleet long before the busy nights do.
On the third-party side, count what the fee buys you out of. The courier who shows up doesn't wear your logo, might be carrying three other restaurants' orders in the same bag, and is the only human your delivery customer meets in the whole transaction. When he's late, the review names your restaurant. You also give up control of timing: his courier arrives when their algorithm says so, not when your kitchen is ready. If delivery quality is part of why people order from you, that gap is worth real money. It just never shows up on an invoice.
Bring your weekly order count and zone map. We'll do the courier math with you in 30 minutes.
The retention math tilts the scale further
There's a second-order effect that almost never makes the fee comparison, and it favors your own fleet: your couriers bring people back. A courier in your uniform, carrying your food in a real thermal bag, ringing the bell when the app said he would, is marketing you were going to pay for anyway. Restaurants that move their direct-channel orders onto an own fleet tend to see better reviews and higher reorder rates on exactly those orders. The food arrives hotter because it didn't share a trunk with two other kitchens' deliveries, and a complaint reaches your team instead of a marketplace bot. Put $7.33 next to $8.50 and none of that is visible. It shows up later, in how often the same customer orders again next month.
The hybrid most restaurants actually land on
The choice was never really binary, and the best operators don't run it that way. The pattern that keeps recurring: a small core of own couriers sized for your reliable baseline volume, the orders that land almost every shift, plus a third-party fleet on standby for peaks, bad weather, and the far edge of the map. Your own couriers stay busy because they only cover demand you already know is coming, and the fleet soaks up the spikes you'd otherwise hire for and then waste.
The catch is coordination. Hybrid falls apart the moment a human has to decide, order by order, who delivers what. It holds together when the software makes that call: our logistics module sends each order to your own fleet first and hands it to a connected third-party service the moment your couriers are saturated, with the same tracking either way. The customer sees one experience; you see one screen. Let the dispatch rules do the arguing, and here's how that auto-assignment works.
Run your own break-even this week
Three numbers, one evening with your delivery reports: your fully loaded courier cost per hour; your realistic deliveries per courier-hour at dinner peak, counting the wait at hand-off and not just the driving; and your current third-party cost per order, including the fees you eat rather than pass to the customer. If your peak-hour density clears 3 drops per courier-hour, put your own people on the peaks and keep the fleet for overflow. If it doesn't, fix density before you hire: tighter zones, smarter stacking. Hiring couriers into low density only turns a fee problem into a payroll problem.
Whatever mix you land on, keep it reversible: run the same three numbers again every quarter. Fleet fees drift, your volume shifts, and the break-even point moves right along with them.
Own fleet, third-party overflow, auto-assignment, and live tracking — one platform, launched in about two weeks.