The density only partially helps, because there is still an overhead in even a single delivery that has to be paid. I'll give you an example, 饿了吗 will deliver food for 5-10 RMBdelivery fee in Shanghai. I would often order say, 生煎包 for 8 RMB and pay 5RMB delivery. A guy on a delivery bike would take like 15-30 minutes to deliver, and I often noticed they only delivered one order in my condo (his bike had nothing else in the basket), and just getting from the street to my condo's door, ringing the phone, and having me buzz him up the elevator would consume 5-10 minutes of his time getting in/out of the place and back to his next delivery.
Shanghai minimum wage is 2500 RMB/mo. So in order to make minimum wage, this guy would need to make 500 orders in 30 days, or 16 deliveries per day. In an 8 hour day, could he pick up and make 16 deliveries? Sure, but it's tight, and obviously at some point, they have to pay the delivery guys more.
And they have, but in order for O2O players to pay more to the delivery people, they've ended up charging more to the restaurants, and so the restaurants have then ended up raising the prices on the menus displayed.
So this indicates that the current O2O system in China wasn't stable. Despite massive revenues, the heavy competition, discounts, and artificially low introductory fees, and ease of switching, doesn't give me confidence their current structure will work.
And 饿了吗‘s logistics seem pretty chaotic to me and no different than Postmates or Doordash in the US. I rarely saw delivery bikes loaded down with multiple orders, because 15-30 minute delivery windows pretty much dictate a point-to-point delivery, rather than a hub and spokes model.