The gap between clicking order and a parcel arriving involves operations that look nothing like how people imagine a warehouse.

Chaotic storage

The counter-intuitive practice at the centre of modern fulfilment.

Rather than grouping similar products together, items are stored wherever there is space, with the location recorded.

Which sounds like disorder and is more efficient for several reasons.

Space utilisation improves, since gaps are filled as they appear.

Picking distance falls, since a popular item exists in many locations across the building rather than one.

And picking errors fall, because adjacent items are dissimilar rather than nearly identical.

Pick paths

The route a picker takes through the building is calculated rather than chosen.

Which is a version of a well-studied routing problem, and small improvements matter enormously when multiplied across millions of picks.

Batch picking collects items for several orders in one trip, then sorts them afterwards, which reduces travel at the cost of a sorting step.

Goods to person

The alternative approach that removes travel entirely.

Mobile robots bring storage units to a stationary worker, who picks from whatever arrives.

Which eliminates walking, which is the largest component of picking time in conventional operations.

It requires substantial capital and works best at high volume with predictable demand.

Slotting

Deciding where items live, based on demand patterns.

Fast-moving items go closer to dispatch, heavy items at accessible heights, and items frequently ordered together are placed to reduce combined travel.

Demand changes constantly, so slotting is recalculated continuously rather than set once.

The packing decision

Selecting box size affects shipping cost, since carriers charge on dimensional weight as well as actual weight.

Which means a large box containing a small item is charged as though it were heavy, and choosing correctly across millions of parcels is worth a great deal.

Systems calculate the optimal box from item dimensions, and machines that cut boxes to size exist and are being deployed.

Returns

The reverse operation, and considerably harder than the forward one.

Each returned item must be received, inspected, graded, and either restocked, refurbished, liquidated or disposed of.

That process costs more than the original fulfilment in many cases, which is why some retailers instruct customers to keep low-value items rather than returning them.

Return rates in clothing are high enough that the reverse operation is a substantial part of the business.

The labour question

Working conditions in fulfilment have been the subject of sustained scrutiny.

Productivity monitoring at individual level, targets derived from measured rates, and injury rates have all been examined by regulators and researchers in several countries.

Injury rates in the sector have been reported above comparable warehousing averages in several analyses, with disputes about methodology.

Regulation limiting algorithmic management and requiring transparency about performance targets has been introduced in some jurisdictions.

The last mile

Where the cost actually concentrates.

Moving a container across an ocean is cheap per item. Delivering one parcel to one door is not.

Failed deliveries requiring a second attempt are the single largest inefficiency, which is why lockers, pickup points and delivery windows have expanded.

Density is everything — a driver delivering many parcels on one street is efficient, and one delivering to scattered rural addresses is not, which is why rural surcharges exist.

Inventory accuracy

The foundation everything else depends on.

If the system believes an item is present and it is not, the order fails at the pick face after all the upstream work has been done.

Cycle counting — continuously counting portions of the inventory rather than shutting down for an annual count — is the standard approach.

Radio frequency identification tagging allows near-continuous visibility and has been adopted heavily in clothing retail where item-level tagging is economic.

Forecasting and placement

Where inventory is held determines delivery speed and cost.

Holding stock close to customers reduces delivery time and requires more total inventory, since each location needs its own buffer.

Which is a direct trade between service level and working capital, and it is why network design is a continuous exercise rather than a fixed decision.

Seasonal peaks

Volumes in retail fulfilment concentrate enormously into short periods.

Which requires temporary labour at scale, trained quickly, working alongside permanent staff.

Peak planning starts months ahead and is where most operational failures become visible to customers.

Cold chain

A specialised category with different constraints entirely.

Temperature-controlled storage and transport for food and pharmaceuticals requires continuous monitoring and documented temperature history.

A break in the chain can render a shipment unusable regardless of appearance, which is why monitoring devices travel with sensitive consignments.

Energy costs are substantial, and door openings are a significant source of loss, which shapes how these facilities are laid out.

Fire and safety

High-bay storage of packaged goods presents substantial fire load, which drives sprinkler design and racking layout.

Lithium battery storage has become a specific concern, with separate requirements in several jurisdictions following warehouse fires.

Which affects what can be stored where, and it is a growing constraint as battery-containing products proliferate.