What is a dark store and when it stops making sense
A dark store is a store closed to the public, operated only to assemble and dispatch e-commerce orders. It sits close to the customer, in a store format, for fast delivery.
It is not the same as a distribution center. A DC supplies stores, in volume, far from the customer. A dark store serves the end customer, in small orders, near them. The geography and order size are opposites.
How the operation measures it
The numbers that matter are orders per hour per picker, cost per assembled order, and profitable delivery radius. A good dark store has predictable flow and a layout designed for the picker’s feet, not the customer’s cart.
The misleading metric is revenue. A dark store can post revenue and still lose money, if the cost per order does not close.
What the number hides
The dark store math is a density math. It only makes sense when there are enough orders to spread the fixed cost of the dedicated space. Below that threshold, the space sits idle and each order carries a rent that does not pay for itself.
The aggregate of “online sales grew” hides this. Sales grow and the dark store stays in the red, because what was missing is density in a radius, not total volume spread thin.
When it stops making sense
At low density, picking orders from inside a live store is usually cheaper. The store is already open, already stocked, already staffed. The online order becomes one more basket assembled in the same aisle.
The dark store wins when the online order gets in the way of the in-store customer, or when density in a neighborhood justifies space for it alone. Before that, it is fixed cost looking for volume that does not exist yet.
Why this changes an AI project
Picking and route optimization help, but they come later. The dark store versus in-store picking versus DC decision is upstream of any model. Getting that choice wrong is expensive in a way no algorithm fixes.
Think about the term your e-commerce operation uses but sizes by guess. “Dark store” is usually that one: it becomes a project because the competitor opened one, not because density calls for it.
Tell me yours. In one hour I will send back a one-page card: the density math, the measurement points, when the dark store wins and when the live store wins. If the card shows the decision has to be rebuilt before any technology, the two-week Diagnóstico is the next step.