Shanghai International Logistics Co., Ltd.
Warehouse Distribution Network Design: Facility Selection And Cost Structure
Warehouse distribution is the layer between the arrival of goods and the release of a customer order. It covers where stock is held, how it is received and put away, how it is picked and packed, and how it is handed to linehaul or parcel carriers. It sits upstream of last mile delivery and downstream of international shipping, and it is where most of the physical inventory and most of the fixed operating cost of a logistics network actually live. This guide sets out the network models, facility options, cost structure, and operational choices that decide whether a distribution operation is efficient or merely expensive.
Three decision horizons shape the work. The design horizon fixes how many sites exist, what each site is allowed to do, and where the inventory sits. The operating horizon decides labour rosters, equipment mix, shift patterns, and the wave schedule. The execution horizon covers daily behaviour: receiving discipline, slotting accuracy, scan compliance, and exception handling.
The boundary of the function also differs by company. Some operate their own sites and buy only transport. Others own no property and buy warehousing and fulfilment as a service. A third group runs small owned depots for fast-moving stock and outsources overflow, e-commerce picks, or slow-moving inventory.
Three network shapes cover most situations. A centralised model funnels all inventory into one or two large sites. A regional model replicates capability near each demand cluster. A hybrid or hub-and-spoke model holds fast-moving stock close to the market and consolidates slower-moving stock centrally, moving it down on a planned cadence.
Centralisation buys scale in picking density, labour supervision, and equipment utilisation, and reduces the number of systems to maintain. It pays for that with longer outbound travel, higher transport cost per order, and a replenishment lead time that must be covered with safety stock.
| Network model | Inventory position | Outbound lead time | Cost signature | Fits best when |
|---|---|---|---|---|
| Single central site | All stock in one location | Longest; every order crosses the network | Lowest fixed cost per unit, highest transport cost per order | Low order density, high SKU rationalisation, strong consolidation |
| Multi-site regional | Duplicated stock by market | Shortest; most orders ship locally | Highest total inventory and site overhead | High order density, time-critical service, expensive final mile |
| Hub and spoke | Fast movers local, slow movers central | Mixed; fast lines short, slow lines planned | Balanced, with extra handling at the hub | Broad SKU range with a clear fast and slow split |
| Ship-from-store or vendor stock | Stock never enters a warehouse | Depends on the node | Low building cost, high supplier dependence | Assortment-led retail, partner networks, drop ship |
| Outsourced fulfilment | Flexible contracted capacity | Contract dependent | Variable with volume, no property capital | Demand volatility, seasonal peaks, market entry |
Whichever shape is chosen, the design must still work when volumes shift. A model tuned to a peak quietly requires a second site at the trough; a model tuned to the trough creates a stranded asset. Sensitivity on volume, not a single forecast, is what makes the business case credible.
The physical building is one option among several, and the choice shifts risk, control, and flexibility more than any other single decision. It is not simply a rate comparison: each option carries a different exposure to volume change.
| Facility type | Control of space and labour | Primary advantage | Main limitation |
|---|---|---|---|
| Owned warehouse | Full control, full exposure | Long-run cost stability and customisation freedom | Capital tied up, staffing rigidity, slow footprint change |
| Leased warehouse | Full control, contractual exposure | Flexibility on term and location, moderate entry cost | Break clauses, fit-out cost, index-linked rents |
| Third-party logistics site | Contractual, performance-based | No property capital, variable cost, existing systems | Less control over layout and peak capacity, higher unit rates |
| Bonded or free-trade zone site | Customs-controlled premises | Deferred or suspended duty on qualifying goods | Regulatory complexity, record keeping, restricted activity |
| Cross-dock site | Very short dwell | Removes put-away and storage cost, fast transfer | Little tolerance for arrival errors, needs sorted inbound |
| Micro-fulfilment site | Small footprint, urban | Very short delivery times, low storage depth | High labour cost per order, small range, low unit economics |
Cross-docking is often presented as an efficiency by default rather than a design decision. It only works when inbound freight arrives sorted by outbound destination, when arrival slots are booked rather than open, and when dwell time is measured in hours rather than days.
Warehouse distribution is usually a high fixed-cost, low variable-cost operation, and that shape explains most of the pressure seen in performance reviews. The building, racking, systems, and supervision are largely committed regardless of order volume. Labour on picking and packing flexes with volume but not proportionally, because shift structures are built around coverage assumptions.
| Cost line | Fixed or variable | Primary driver | Most effective lever |
|---|---|---|---|
| Occupancy and rent | Fixed | Contracted area, location index, term length | Better space utilisation before taking or giving up area |
| Direct labour | Semi-variable | Units handled, shift pattern, travel distance | Slotting, wave design, labour management |
| Transport and linehaul | Variable | Order count, weight, cube, consolidation level | Order consolidation, packaging, routing |
| Equipment and maintenance | Mixed | Fleet size, utilisation, service intervals | Right-sizing the fleet, planned over breakdown maintenance |
| Technology and licences | Fixed | Sites, users, modules, integrations | Removing duplicate systems and manual interfaces |
| Shrink and damage | Variable | Handling steps, packaging quality, count discipline | Fewer touches, better packaging, root-cause on claims |
Two observations follow. Most levers that reduce transport cost sit in the warehouse, not in the transport function: order consolidation, packaging reduction, and pick sequence all change the shape of the outbound load. And fixed cost is only controllable if utilisation is measured.
Site selection for a distribution facility is a multi-year commitment, and the criteria extend well beyond freight distance. Transport access matters, but so do the labour market, power capacity, structural limits of the building, and the regulatory environment.
Proximity to a dense concentration of carriers and forwarders also carries value that a freight rate comparison misses. It widens the supplier base, improves backhaul availability, and speeds exception handling, which matters most for sites handling a high share of urgent orders.
Racking choice follows SKU profile and throughput rather than space saving alone. Selective pallet racking gives the best access and lowest capital cost per position, but consumes aisle space and suits only a consistent pallet footprint. Narrow aisle systems raise density considerably at the cost of slower access and stricter vehicle discipline.
| Storage method | Typical use | Space efficiency | Handling equipment | Main constraint |
|---|---|---|---|---|
| Selective pallet racking | Full pallets, medium velocity | Moderate | Counterbalance or reach truck | Aisle space for truck access |
| Narrow aisle racking | Full pallets, higher density | High | Reach or turret truck | Operator training and guide rails |
| Drive-in racking | High density, low variety | High | Counterbalance truck | Sequential access, FIFO discipline |
| Shelving and mezzanine | Small parts, picked items | High vertically | Manual or pick module | Weight limits on upper levels |
| Automated storage and retrieval | High throughput, stable volume | Very high | Integrated machines | High fixed cost, inflexible layout |
Pallet footprint constrains everything else. European pallets near 1200 by 800 millimetres, standard pallets near 1200 by 1000, and North American 48 by 40 inch pallets do not fit the same bay or trailer in the same way. Mixing footprints produces awkward residuals that quietly consume capacity.
Positioning is the decision of how much of each item to hold where. The basic arithmetic is a reorder point equal to demand over the replenishment lead time plus a safety buffer, and the buffer grows with variability in demand and in supply.
Segmentation makes the position manageable. ABC classification by value shows where accuracy and control matter most, because a small share of SKUs carries a large share of value. XYZ classification by demand variability shows where extra buffer is genuinely needed.
Picking is where most of the labour content sits, and where most of the available improvement sits as well. The method should follow the order profile: many small lines with tight windows, or fewer large orders with predictable volumes.
| Picking method | Best suited to | Throughput profile | Main risk |
|---|---|---|---|
| Single-order picking | Low to medium volume, high order value | Low; one trip per order | Travel time dominates cost |
| Batch picking | Medium volume, overlapping order lines | Moderate and predictable | Batch release delay affects service |
| Wave picking | Large volume, scheduled dispatch | High and smooth | Planning and resourcing complexity |
| Zone or cluster picking | Large sites, split travel | High | Handover points between zones |
| Pick-to-light or voice | High line accuracy requirement | High with fewer errors | Fit-out cost and process rigidity |
Two details carry more weight than they appear to. The first is the position of the pack bench relative to the pick face, since travel between stock and dispatch is pure overhead and small layout changes can remove a meaningful share of it.
A warehouse management system coordinates location, inventory, and task assignment, but it can only be as good as the data entering it.
| Symptom | Likely cause | Consequence | Corrective action |
|---|---|---|---|
| Cost per order rising while volume is flat | Site underutilised, fixed cost spread thinly | Margin erosion | Measure utilisation, consolidate volume, renegotiate footprint |
| Pickers travelling further each year | Slotting not maintained after range change | Direct labour cost increase | Automated slotting review on a fixed cycle |
| Stockouts on fast-moving lines | Reorder point based on average rather than variability | Lost sales and expediting cost | Recalculate safety stock, shorten the replenishment cycle |
| Same-day and next-day failures | Network too centralised for the promise | Service credits, customer churn | Split the range by velocity across sites |
| Rising damage and shrink | Extra handling steps, weak packaging specification | Claims cost and lost margin | Map touch points, cut redundant handling, standardise packaging |
| Peak failures every year | Capacity and labour assumed at average volume | Expediting, overtime, service failure | Plan peak capacity and labour explicitly, pre-agree overflow space |
Most corrective actions here are process changes rather than capital projects, which is usually good news: process changes can be tested in weeks, while a facility move or automation project takes months to specify and longer still to amortise.
A warehouse distribution operation earns its keep through a few well-understood levers: where inventory is positioned, how many sites carry it, how well the network is filled, and how many touches each order needs. Each lever can be measured, and each carries a known trade-off against service, cost, or flexibility.
Start by measuring what exists: real cost per order including allocated overhead, utilisation by site, travel time per pick, and the true order profile by size and destination. Those figures usually identify two or three changes worth more than a new facility.