The challenge
Many orders, few units each.
E-commerce and retail run two warehouses inside one building: the one that picks a single unit for a person, and the one that builds replenishment pallets for a store. Both lines want the same product in different places.
"The hard part is not picking a thousand orders. It is picking a thousand small orders while loading a replenishment truck from the same stock."
Before & after
Without automation
- Two opposite destinations — single units and pallet replenishment — competing for the same stock and the same aisle.
- Seasonal peaks that multiply volume within weeks and then recede.
- Returns arriving unannounced, taking floor space, with no loop of their own.
- Many slow-moving references living next to a few very fast ones.
- The dispatch cutoff time, which decides how late an order can still be picked, and is not negotiable.
With STOKA
- Physically separate both streams and size each one by its own outbound profile.
- Locate by rotation and order affinity, not by product family or supplier.
- Define the returns loop with its area, its decision criteria, and its time limit.
- Size for both the peak and the valley, then decide how much of the gap is absorbed by process and how much by equipment.
- Bring the product to the operator in the band of references that concentrates most picked lines.
Why automate
Key benefits
- A layout that serves both destinations.
Separate the area feeding single-unit orders from the one building store replenishment, without duplicating stock or forcing a choice about which one waits.
- Shorter travel per order.
Reassign locations by rotation and order profile so the operator walks less per order picked, before buying any equipment.
- A dedicated loop for returns.
Returns stop being an exception piling up in a corner and get their own space, decision criteria, and time limit to go back into sellable stock.
- Capacity added by stations, not by floor.
When the peak arrives, add picking positions instead of square meters: the building stays the same and output capacity does not.
Automation context
Warehouse automation in e-commerce and retail
An e-commerce and retail warehouse is two warehouses inside one building. One picks orders of one or two units going to a person's home. The other builds replenishment loads going to a store. They share the roof, the team, and almost always the stock — and almost nothing else. The first wants the product within arm's reach; the second wants it available by quantity; and both want it at the same time.
That coexistence is what makes the layout hard, and it is what sets this sector apart. In a manufacturing plant the warehouse serves the line. In a third-party logistics operation the warehouse is the product being sold. Here it serves two customers with opposite logics, and most problems blamed on volume are really problems of that coexistence.
The requirement shaping the design is not regulatory: it is operational, and it comes from the sales channel. There are two, and neither is negotiable. The first is the delivery date you promised, which sets a dispatch cutoff: how late an order can still be picked and ship today. Everything that happens before that hour has to fit inside it. The second is the returns window, which decides how much comes back, in what condition, and how urgently it has to re-enter sellable stock. The channel sets both, not the legislator, and missing either has an immediate cost. There is also a regulatory layer — distance selling, shipping documentation — worth knowing about, but it is not what organizes the warehouse.
Keep reading
The process that carries the most weight is order picking. In a warehouse that ships pallets, picking is part of the work; here picking is the work. And there is a second process that is an exception elsewhere and structural here: returns. What comes back is not an incident, it is a permanent stream that needs inspection, a decision on every unit, and a way back.
Returns deserve their own paragraph, because this is where most operations improvise. A returned unit does not enter stock: it enters a decision queue. Someone has to look at it and decide whether it sells again as is, needs refurbishing, goes back to the supplier, or gets scrapped. Each of those four outcomes has a different physical destination, and if they are not defined in advance, returns pile up in the first free spot — usually the one picking needed. The cost does not show up in returns: it shows up as picking getting slower without anyone knowing why.
The typical mistake comes in two shapes and both are expensive. The first is sizing for the average and collapsing at the peak: the warehouse works eleven months and fails in the one that matters. The second is sizing for the peak and paying for idle capacity the rest of the year, which is what happens when the answer to the peak is renting more floor. There is a third path between them: size the structure for normal volume and leave the elasticity in picking, which is where the peak actually lands.
The quieter mistake is locating by product family or by supplier. It is convenient for putaway and terrible for picking: it forces a walk across the whole warehouse to assemble a three-line order. Locating by rotation and by order affinity — what gets ordered together with what — changes travel without moving a single rack.
What improves at each step. Ordering is where the first peso pays off most: review the location criteria, separate the two streams, define the returns loop, and write the picking procedure. Nothing is purchased and travel per order drops. Equipping adds racking and mezzanines where volume justifies them, plus management software that says where everything is and in what order to walk it. Mechanizing appears when movement between zones becomes the bottleneck: conveyors that take the picked order out of the picking area without anyone carrying it. Automating is the step where the product stops waiting for the operator.
When automation makes sense, specifically in this sector. Not when volume grows, but when the limit stops being travel and becomes how many people can work in the same aisle at once. A well-organized e-commerce warehouse hits a physical ceiling: no matter how short the routes get, only so many operators can work in one area without getting in each other's way. Past that point, the only way to add capacity without adding floor is to invert the movement and have the product travel to a fixed station. That is what lets a peak be absorbed by adding work stations instead of square meters, and it is the square-meter argument applied to this sector: the space you free up is not used to store more, it is used to pick more.
One note about space, because the constraint here differs from other sectors: the square meter of an e-commerce warehouse tends to be the most expensive of all, because being close to the customer is part of the service. That changes the arithmetic of any floor-space decision, and it is why densifying pays off more here than in a warehouse on the outskirts.
If the warehouse you are evaluating is not yours but belongs to a third party operating for several clients, the page you want is the one on logistics operators: there, density is not an improvement to the operation, it is the margin of the business itself.
Recommended equipment
Products for this sector
FAQ
Frequently asked questions about automation
Should online orders and store replenishment share one warehouse?
They should share the building, not the loop. Sharing stock avoids duplicating it, but the two streams need different locations, different work units, and different schedules. What gets separated is the path and the timing, not the inventory.
How do you handle the seasonal peak without paying for idle capacity all year?
Size the structure for normal volume and keep the elasticity in picking, which is where the peak actually lands. Adding work stations for a few weeks is a cost you can retire; adding square meters is a cost you pay every month.
Do returns really need their own space?
Yes, and that is the difference between an e-commerce warehouse and one that only ships. Returned goods need inspection, a decision — sellable, refurbish, scrap — and a path back into stock. With no loop of their own, they take the space picking needed and slow down what ships.
When does reorganizing stop being enough?
When travel is already optimized and the limit becomes how many people fit in an aisle at once. From there, the next step is having the product travel to the operator instead of the operator to the product, because that adds capacity without adding floor.
Ready to automate?
Free consultation. Response within 24 business hours.
Our technical team analyzes your operation and proposes the most suitable solution with no commitment.