[Retric]

Delivery apps have a vastly higher overhead because the drivers are not based out of a single high volume restaurant.

With traditional deliveries operating over a tiny radius back to base. Drivers can do multiple deliveries at the same time. Pick up a new stack of orders and quickly be out the door again, this means they need fewer people during rush and thus much lower overhead.

Some people are willing to pay 10+$ an order to have a much wider selection of restaurants from a huge area. But, that doesn’t scale to the kind of volume these companies expect.

[c22]

For what it's worth, these ride-share delivery drivers also grab multiple orders at a time if they can. It's one of the reasons delivery takes so long. I've observed drivers putting in additional orders that just came in (and waiting for them) while they are picking up. I also think they'll swing by (and often wait at) two or three places to get orders before heading out to deliver.

[FartyMcFarter]

There's a limit to this though. If the food gets cold or otherwise bad due to waiting too long, people will ask for money back or orders will go down.

[munk-a]

Compare this to a restaurant running their own service that can chose not to prepare meals while waiting for more orders to roll in - so the delivery might not be as quick as it potentially could be, but the food is fresher.

[rightbyte]

The main edge the restaurants with their own delivery has. UberEats just black boxes the restaurant excepting it to treat the delivery guy as a ordinary customer. That process has to be so wasteful compares to bulk preparation and delivery.

[munk-a]

The high volume portion is an interesting component - uber eats and whatever else cast very wide nets of the restaurants they'll deliver for while, prior to the SV injection, only some restaurants could reasonably afford to sustain delivery drivers. Sure, being centralized allows you to allocate only a portion of a driver to a given restaurant but that comes at the cost of complexity and likely a reduction in either parallelized deliveries or quality of service.

[aeyes]

That was actually the first Uber Eats business model, sadly I can't find a source. They'd have a restaurant cook a couple hundred meals in bulk and then the drivers would bulk deliver them.

Letting customers order from restaurants they already know is obviously a bit easier to scale.