Scale Masking Per Market Losses
Ninja
The core risk is that more orders and more cities can make a quick commerce business less healthy, not more healthy, when each new zone adds fixed cost before demand is dense enough to carry it. GoPuff showed this clearly, growing to roughly $2B in 2021 revenue while carrying about a $500M EBITDA loss, after scaling a network of hundreds of micro fulfillment sites that each needed labor, inventory, and local delivery capacity. In this model, revenue scale can hide weak per market economics until the store base is already too large.
-
Dark store economics only work when enough orders are packed into a tight radius. Prior research modeled a mature ultrafast store at about 500 orders per day and $25 AOV for positive contribution margin, while also noting online grocery often needs baskets closer to $50 to be fully sustainable after broader costs.
-
The trap gets worse when convenience pushes basket sizes down. Expert interviews found quick commerce users often order only what they need right now, which lowers AOV even as frequency rises, and adding fresh grocery to lift basket size can raise spoilage and picking complexity at the same time.
-
Density is the release valve. In hyperdense markets, labor efficiency improves sharply because one courier can stack more orders per trip. That is why campus models like Duffl reported 10 to 12 orders per hour versus a more typical 3 to 6, and why sparse expansion is so dangerous for dark store operators.
Going forward, the winners in quick commerce will be the operators that treat each micro market like a store level P and L, not a land grab. For Ninja, that means earning the right to open the next city only after the current one shows dense repeat demand, disciplined SKU selection, and enough basket quality to cover riders, shrink, and store labor.