A delivery promise looks like a marketing decision made on a product page. It is really a commitment about geography, and the warehouse network has to be built to keep it.
Distance is the constraint that cannot be automated
Goods travel at the speed of the transport network. Once an order leaves the building, the delivery time is fixed by how far the parcel has to go.
Faster promises therefore require stock to be closer to customers before the order exists. The only way to shorten transit is to shorten distance.
This is why next-day and same-day promises multiply warehouse locations. The buildings are positioned around delivery times rather than around property cost.
Cut-off times set the working day
A promise of dispatch today requires the order to be picked, packed and handed to a carrier before that carrier's collection. That collection time sets the cut-off.
Everything upstream compresses into the hours before it. Peak picking load concentrates in the final part of the afternoon rather than spreading across the day.
Extending the cut-off by an hour is a competitive move that costs labour, because staffing must cover the extended peak whether or not orders arrive.
More locations means more inventory
Holding a product in several warehouses requires safety stock in each, because demand cannot be predicted precisely at any single site.
Total inventory therefore rises as locations multiply, even though sales are unchanged. The capital tied up in stock grows with the promise, not with revenue.
Splitting inventory also raises the chance that an order contains items held at different sites, which produces multiple parcels and multiplies the shipping cost.
Layout follows the order profile
A warehouse serving many small orders is designed differently from one serving few large ones. Picking dominates the first, and handling dominates the second.
Fast-moving items are placed near packing stations, and slow ones further away, because walking distance is the largest component of picking time.
As the order profile changes, layouts that were efficient become inefficient. Reslotting is routine maintenance rather than a one-off design exercise.
Returns need their own capacity
Returned goods arrive individually, in unpredictable condition, and require inspection, grading, repackaging and restocking before they can be sold again.
That work has almost nothing in common with outbound picking, so returns are usually handled in a separate area with separate staff.
Categories with high return rates need this capacity sized properly. Where it is underprovided, returned stock queues and is unavailable for sale at exactly the time demand exists.