📦 Warehouses and Fulfillment: Ordered Today, Delivered Tomorrow

What you’ll learn

1. The Building That Is Not a ShopGrasp the course's central claim — that a fulfilment centre is organised for a database rather than for human eyes, which is why nothing in it is arranged the way you would expect.A supermarket groups by category because it serves a browsing human who has not decided yet. A fulfilment centre serves requests that arrive already decided, as data with a unique code attached. With no eye to please, the building is free to abandon human legibility entirely — and it does.
2. Chaos Beats TidinessUnderstand chaotic (random) stow — why shelving items wherever there is a gap beats organised storage on space, walking distance and error rate, and what its single point of failure is.Items are shelved in any bin with room; the worker scans item and bin, and the database owns the location. This wins on space (no category sized for peak), on walking (popular items spread evenly, so mixed orders have short paths), on mis-grabs, and on eliminating re-filing entirely. Its one failure mode is a skipped scan — which loses the item while you are looking at it.
3. Pick, Pack, ShipFollow an order through receive, stow, pick, pack and ship — and understand why orders are deliberately dismantled, and why the picker's real problem is a routing problem.Your items are three unrelated retrieval problems that share a destination. The system batches them with dozens of strangers' nearby items onto single short loops, then reassembles orders at a consolidation wall. The software's job is not finding — that is free — but choosing short routes, a cousin of the travelling salesman problem that is approximated continuously.
4. Bring the Shelf to the PersonUnderstand goods-to-person robotics as the deletion of one specific motion — the walk — and see how a floor with no fixed homes reorganises itself around demand.Freestanding pods of bins are carried to stationary human workstations by drive units — the Kiva technology Amazon acquired in 2012 and now calls Amazon Robotics, with units named Hercules and Titan; Amazon says it has deployed over a million robots since 2012. Because pods have no home, hot stock drifts toward the workstations as a side effect of putting pods down wherever is convenient.
5. Sorting by Where, Not WhatUnderstand sortation as a second great re-sort — by destination rather than by contents — and why parcel networks converge on a hub-and-spoke tree.Once labelled, a parcel is a destination with a mass; nothing downstream cares what is inside. Sorting proceeds from continent to region to city to street. Direct sorting between a thousand origins and a thousand destinations would need a million relationships; a hub needs two thousand — which is why a 40 km parcel may travel via a distant hub.
6. The Last Mile Costs the MostUnderstand why the last mile costs the most — a division problem rather than a technology problem — and derive lockers, missed-delivery costs and urban/rural cost differences from it.Every other leg consolidates: divide a crossing's cost by 100,000 parcels and it vanishes. A van divides a driver-day by perhaps 150 stops, because the whole point of the last mile is that everyone is going somewhere different. Industry estimates commonly put the last mile above half of shipping cost (Statista's 2023 figure is 53%, up from 41% in 2018) — a figure best treated as an order of magnitude.
7. Why the Building Is Where It IsUnderstand fulfilment-centre siting as the intersection of four conflicting constraints, and see next-day delivery as a real-estate achievement rather than a logistics one.Vans leave at dawn, so the parcel must be sorted overnight and already close: count the road hours backwards and that radius is the promise. But the building also needs cheap flat land, motorway access, and a commutable labour pool. Those pull opposite ways, which is why these sheds cluster on the outer edge of metro areas beside motorway junctions.
8. The People in the BuildingName the live dispute over warehouse working conditions factually — what is established, and what the critics and the operators each actually argue — without adjudicating it.Established: the work is physical and paced, and because every action is a scan, individual productivity is measured continuously. A 2025 American Journal of Industrial Medicine study of Californian warehouses found injury rates peaking at 5.87 per 100 workers in 2021 and job-transfer/restriction rates still elevated in 2023 (3.21 vs 1.84 per 100 in 2018), against a US private-industry average of about 2.4 per 100 (BLS, 2023). Critics argue the measurement sets the pace; operators argue measurement is what makes safety improvable and that comparisons to whole-economy averages are not like-for-like.

Questions this course answers

Grounded in trusted sources

• Quartz — 'Amazon built one of the world's most efficient warehouses by embracing chaos' (chaotic storage; Amazon's 2013 statement to 60 Minutes)
• AllAboutLean.com — Storage Strategies: Random, Chaotic and ABC; The Inner Workings of Amazon Fulfillment Centers
• Amazon (aboutamazon.com) — 'Amazon uses robots that sort, lift, and carry packages': Hercules, Titan, Sparrow, Proteus, Sequoia; over 1 million robots deployed since 2012
• Wikipedia — Amazon Robotics (Kiva Systems acquisition, 2012)
• Beckman et al. — 'California Warehouse Industry Worker Injury Rates in OSHA's Injury Tracking Application, 2018–2023', American Journal of Industrial Medicine (2025)
• US Bureau of Labor Statistics — Employer-Reported Workplace Injuries and Illnesses, 2023 (private-industry incidence rate)
• Statista — Last-mile share of total shipping costs, 2023 (widely cited industry estimate)
• Jean-Paul Rodrigue, The Geography of Transport Systems (transportgeography.org)

Every Wunder lesson is built from real, reputable sources — never invented.

Related Math courses