How the Distribution Centre Decides What Is on Your Shelf

Why the DC Decides Your Shelf

In one line: the building you never see — the distribution centreA centralised warehouse that receives bulk stock from suppliers and redistributes mixed, store-ready loads to retail outlets. (DC) — is the machine that decides whether your shelves are full or empty on a Monday morning. This playbook explains how it works, what it costs, and where it breaks.

Whether you run a franchise supermarket on the KwaZulu-Natal coast, an independent store in a township, or a cash-and-carry serving spaza shops, your business runs on a logistics model most owners never fully see. The shift over the last twenty years has been total: from dozens of supplier trucks fighting for space at your back door, to a handful of mega-warehouses that run twenty-four hours a day and dispatch a single mixed-load truck to your store.

This document tells that story operationally — with a realistic model of how a large DC is sized, staffed, and funded; a synthetic P&L and payback calculation; the systems that move stock and data; and the pain points (loadshedding, cargo crime, rigid delivery windows) that hit your pocket directly. It is deliberately brand-agnostic: no chain names appear anywhere, and all financial figures are clearly labelled illustrative / synthetic so you can recalibrate them to a real client.

From the Back Door to the Mega-Warehouse

The shift: retailers moved from direct store deliveryDirect Store Delivery (DSD): each supplier drives its own truck to each store. High control, high cost, heavy on the back door. (DSD) — where every supplier drove to every store — to centralised distribution, where suppliers drop bulk at one regional hub and a single mixed-load truck serves each store. The chains that moved early won a logistics lead the late movers are still closing.

The old world: when the back door never stopped

Through the 1970s, 80s and into the 90s, the receiving bay was the busiest part of a supermarket. The bread van, the milk truck, the soft-drink truck, the spice rep, the detergent rep — each arrived separately, each wanted a signature, each needed dock space, and each wanted their invoice checked on the spot. Stores held weeks of inventory in large stockrooms and employed receiving staff whose day was spent arguing about short dates and chasing credits for damages.

For independents and franchisees, the real cost was buying power. A single store negotiated alone, and a supplier could quietly give a bigger chain a better deal down the road. Large retailers knew centralisation was more efficient for decades, but major suppliers pushed back hard — they liked controlling the delivery schedule and the direct relationship with each store manager, and they offered their best discounts to retailers who kept delivering store-by-store1.

The turn: central distribution takes over

By the mid-2000s, the decentralised model of the largest chains was bleeding money at scale — costs rising, availability patchy, back-door chaos unsustainable. One major retailer announced a phase-in of centralised distribution, requiring suppliers to drop goods at regional hubs instead of individual stores4. A single company truck would then carry mixed loads to each store.

The flagship hubs that followed set the template for the whole industry:

Inside a Distribution Centre

ⓘ Modern pallet racking stands six levels high. A picker rarely uses paper — a voice headset directs them to aisle, bay and case count. Note the electric forklift: large floors run almost silently because both the trucks and the picking are hands-free.

The reality: a major DC runs 20–24 hours a day across multiple shifts, with receiving, put-away, cross-dockCross-docking (flow-through): inbound goods are moved straight from the receiving dock to the outbound dock, with little or no storage in between. Used for fast-moving and fresh lines., voice-guided picking, route-planned dispatch, and a returns/waste stream. Every zone has its own equipment, staff and procedures.

The shift pattern

Most major DCs run three shifts, often continuously. A single mega-DC can employ on the order of 3,000–4,000 people directly, plus indirect contractors7. Typical roles: receiving clerks, put-away teams, pickers, dispatch loaders, forklift and MHEMaterial Handling Equipment: forklifts, reach trucks, pallet jacks, conveyors and automated guided vehicles used to move goods inside the DC. drivers, quality controllers, maintenance, and security.

Receiving — where it all starts

Supplier trucks are checked in at the gate and backed into dozens — sometimes over a hundred — receiving doors. One large inland DC has roughly 260 dock levelers across receiving and dispatch12. The receiving clerk checks the delivery note against the purchase order, scans goods into the WMSWarehouse Management System: the software that controls receiving, put-away, inventory, picking and dispatch — the digital brain of the DC., and flags fresh produce, meat and dairy for quality checks. Clean stock then flows one of two ways:

1. Put-away: pallets are stored in racking — in the ambient section, stacked up to six levels high. The system chooses the location to optimise density and FEFOFirst-Expired-First-Out: stock rotation by expiry date (a date-driven version of FIFO) so the oldest-dated goods ship first..
2. Flow-through / cross-dock: fast-moving and fresh lines never touch a shelf — they move straight from inbound to outbound dock.

The three temperature zones

A modern DC is split into specialist environments. You cannot move equipment freely between them without a temperature lock or equipment switch.

Picking — getting your order ready

When a store orders, the WMS breaks it into picks. A picker receives instructions through a voice headset — no paper, no shouting — telling them the aisle, bay and case count8. The stock goes onto a pallet or into a tote, is scanned, and moves to dispatch. In leading DCs the floor runs in near-total silence because picking is voice-driven and the trucks are electric8.

Picking methodologies & inventory accuracy

How a DC picks decides how fast and how accurately your store gets its order. There is no single “best” method — the choice depends on order profile, volume, and whether the goods are ambient or temperature-controlled. The four methods you will hear about are:

• Discrete (single-order) picking: one picker walks one order end to end. Simple to train and manage, but wasteful of travel time — common in smaller DCs or for unusual orders.
• Batch picking: one picker collects lines for several orders at once into separate totes, then sorts. Cuts walking dramatically for high-volume, small-line orders.
• Zone picking: the DC is divided into zones (often by temperature or category), and a picker owns one zone; an order is assembled as it passes through. Essential for mixed-load grocery because ambient, chilled and frozen must stay in their own environments.
• Wave picking: orders are grouped into waves released against the shift plan, the truck schedule, and the cut-off times. Wave timing is what gets your order onto the right truck, in the right loading sequence.

Underpinning all of this is inventory accuracy — the percentage of locations where the system’s record matches what is physically on the shelf. Best-in-class DCs run at 99% or better, sustained by cycle countingCycle counting: a continuous, rolling count of a small subset of inventory each day, replacing the disruptive full annual stock-take, so accuracy is maintained year-round. rather than a single annual shutdown stock-take. Low accuracy is the hidden cause of “phantom stockouts” — the system thinks a product is available, the picker arrives at an empty location, and your store ships short. For a store owner, every short ship is a lost sale and another credit note to chase.

Dispatch — loading the trucks

This is the logistics brain. Route-planning software decides which truck carries which stock and in what order it is loaded. The last store on the route is loaded first, so when the truck arrives at the first store, that store’s stock is right at the doors8. A single truck carries a mixed loadA single delivery combining ambient, chilled and frozen product, separated by insulated curtains or compartments so one truck serves a store’s full needs. — frozen, chilled and ambient separated by curtains or compartments.

Returns, recycling and waste

DCs also reverse-flow damaged goods, expired product, recalls and unsold promotional stock to a dedicated returns area for sorting, crediting or destruction. At modern sites every used cardboard box is baled on-site into compact “snakes” and sent to recyclers8.

Example document — DC shift KPI dashboard

The DC Economic Model

Sizing the machine: this chapter gives a granular, brand-agnostic operational model — footprint, throughput, headcount, capex and an operating-cost breakdown — for large DCs serving multiple outlets. All figures are illustrative industry-typical ranges; recalibrate to the client’s actuals.

Facility size, throughput and headcount bands

Illustrative operating-cost breakdown

Where a DC’s running cost actually goes. Percentages are typical industry blends for a multi-temperature retail DC; chilled-heavy or automated sites skew differently.

Worked example: unit economics & break-even

Take the DC above: it handles 28 million cases a year with R238m of operating cost. That works out to roughly R8.50 of operating cost per case handled. Add the recovery margin target and the cross-dock premium, and the DC’s blended handling fee lands around R9–10 per case before delivery. Delivery then adds a per-drop and per-kilometre component on top (see the fee menu in Chapter 6).

Break-even: the DC’s fixed costs (facility, equipment, core staff, IT) do not fall much if volume drops, so cost-per-case rises sharply below the design throughput. If this DC ran at 60% of plan (~17m cases), the same fixed cost spreads over far fewer units and per-case cost climbs — which is exactly why under-utilised DCs destroy value. The operational lesson is blunt: a DC is a volume machine; running it part-empty is more expensive than running it full. That single fact is why retailers fight so hard to consolidate volume into fewer, fuller hubs.

Sample P&L & 5-Year Payback

The money: a captive retail DC usually runs as a cost-recovery operation (low single-digit EBITDA) — its real return is captured system-wide as lower delivered cost, fewer stockouts and less spoilage. Below: a synthetic P&L, then a 5-year payback for a medium chilled DC.

Illustrative DC P&L (synthetic)

Illustrative 5-year payback — medium chilled DC

Reading the result: a medium chilled DC pays back in roughly 5–6 years in this synthetic model. Ambient DCs (no refrigeration) typically pay back faster — around 3–4 years — because capex is far lower. Highly automated or fresh-meat-processing centres take longer, often 7–10 years, unless co-funding from suppliers shortens it.

How DCs Pay for Themselves

The funding model: a DC “pays for itself” through five reinforcing levers — distribution fees, supplier contributions, scale economies, cross-dock savings, and centralised procurement power. Rarely does one lever carry the whole load; the business case is the stack.

The five commercial levers

Example fee structures (illustrative)

Stores, Systems & Integration

The nervous system: a store order is rarely typed by hand anymore. It is generated from POSPoint of Sale: the in-store till system that records every sale; its sell-through data feeds replenishment and forecasting. sell-through and demand forecasts, pushed to the WMS over EDI/FTP or an API, fulfilled by the DC, and confirmed back to the store on receipt.

How a store order is created

1. POS signals: every scan at the till updates store inventory. When stock hits a reorder point, a replenishment need is raised.
2. Demand forecast feed: a forecasting engine adjusts the order for day-of-week, promotions, seasonality and weather.
3. Order creation: the replenishment engine builds the order and transmits it to the DC via EDIElectronic Data Interchange: standardised B2B messages (e.g. orders, ASN advance-ship notices, invoices) exchanged between systems., SFTPSecure File Transfer Protocol: an encrypted file drop used to exchange batched order/inventory files between trading partners. file drop, or a real-time APIApplication Programming Interface: a way for two systems to talk in real time over the web, exchanging small structured requests and responses. call.
4. DC fulfilment: the WMS converts the order into picks; the WCS orchestrates conveyors/equipment; the TMS plans the route.
5. Store receiving: an ASNAdvance Ship Notice: an electronic message telling the receiver exactly what is on the truck, before it arrives — enabling fast, accurate check-in. tells the store what is coming; the store scans it in and reconciles.

How forecasting & replenishment actually work

Behind every auto-generated store order is a forecast. Modern demand-planning engines do not simply repeat last week’s sales — they combine historical sell-through with a stack of adjustment factors: day-of-week and day-of-month patterns, public holidays and pay-week effects, planned promotions and price changes, seasonality, and increasingly external signals such as weather and local events. The output is a predicted demand curve per SKU per store, usually looking a few days to a few weeks ahead.

The replenishment engineReplenishment engine: software that turns demand forecasts plus live POS/stock data into specific order quantities, automatically creating purchase orders for the DC. sits on top of that forecast. For each SKU it compares predicted demand against current on-hand stock, stock already on order, supplier lead times, and a safety-stock buffer. When projected stock is due to fall below the safety level inside the replenishment window, the engine raises an order line — sized to the case-pack or pallet multiple — and bundles many lines into a single store order. The result is “just-in-time” ordering rather than “just-in-case” ordering: stores hold less buffer stock because the system is trusted to top them up before they run dry.

Where this goes wrong is usually a data problem, not a maths one. If the forecast is fed stale POS data, wrong promotional flags, or a broken master-data record (a wrong case-pack, a discontinued code still listed), the engine will confidently order the wrong quantity. That is why master data managementMaster Data Management (MDM): the discipline of keeping shared core data — product dimensions, case-packs, barcodes, supplier records — accurate and consistent across every system that uses it. and clean promotional data are prerequisites, not nice-to-haves. A store that overrides auto-replenishment constantly is usually signalling a data-quality issue upstream, and the fix is rarely more manual ordering — it is fixing the forecast inputs.

Order & planogram flow (text version)

Supplier  →  DC (inbound)  →  WMS / WCS  →  Cross-dock / Replenishment  →  Store order  →  Store receiving

Supplier Engagement

Who pays, who delivers: suppliers fund DCs through slotting fees, delivery-window premiums, co-funding and co-packing. Goods arrive either direct-to-store (DSD) or via the DC. The split is a commercial decision per category, not a technical one.

How suppliers support the DC

DSD vs DC delivery — the trade-off

Common supplier SLAs (illustrative)

Distribution Centre Types

ⓘ Two temperature regimes side by side. Chilled holds 0–5 °C; frozen sits at −18 °C or colder. Energy, capex and handling discipline all rise sharply as temperature falls — a frozen chamber can cost two to three times as much per square metre to build.

Not all DCs are equal: a dry-goods shed and a fresh-meat processing centre share a name but almost nothing else. Each type has distinct infrastructure needs (HACCPHazard Analysis and Critical Control Points: a food-safety management system identifying and controlling biological, chemical and physical hazards across the chain., GMPGood Manufacturing Practice: minimum standards for hygiene, process and documentation in food/production facilities., plant access, specialised racking), cost drivers and throughput profiles.

Handling notes for fresh & perishable DCs

The further a DC sits toward fresh and perishable, the more its economics are governed by shrink (lost stock to spoilage) rather than by throughput alone. Fresh produce centres add ripening rooms (bananas, avocados), grading and quality lines, and wash-down hygiene zones; their waste percentage runs far higher than ambient, and a single delayed truck can turn a day’s perishable intake into write-off. Meat and poultry processing centres are part warehouse, part factory: they receive carcasses or primal cuts, break them down, portion, tray-pack, gas-flush or vacuum-seal, apply date codes, and dispatch under continuous chilling — all under GMP and HACCP controls with sanitation cycles built into the shift. Fish and seafood cold chains are the tightest of all, often relying on ice or brine and super-chilling to stretch a shelf life measured in hours, not days.

The commercial implication: perishable DCs cannot be run as pure cost centres the way ambient sheds can. Their value is measured in freshness delivered and spoilage avoided, and they are typically located as close to the demand cluster as possible to shrink transit time. For a store owner, this is why your fresh delivery’s arrival temperature and remaining shelf life matter more than the invoice price — a cheaper pallet that loses a day of life costs more in the end.

Technology Stack & Integrations

The stack: a modern DC runs on a family of specialised systems — WMS, WCS, TMS, ERP, demand forecasting, planogram tools, replenishment engines, and EDI/API gateways — joined by an integration layer. Below are capabilities only (no vendor names); evaluate tools against these, not against logos.

Pain Points & Risk Mitigation

What breaks: loadshedding and the cold chain, cargo crime, supplier consolidation, and rigid cut-off times are the live SA pain points. Below is a risk-and-mitigation matrix plus the SA-specific numbers that make them urgent.

SA-specific risk signals (illustrative, to be verified)

• Loadshedding cost: one major retailer reported a ~R200m spoilage loss in a single year14; another’s monthly generator diesel spend climbed from ~R16m to ~R150m at peak15. Cold-chain operators added 15–30% to operating cost and apply 5–15% energy surcharges16.
• Cargo crime: SAPS recorded 420 truck hijackings in Q2 2025 alone17; over 18 months to mid-2022, ~2,670 incidents and ~R577m in losses were logged across all provinces17. The logistics “security tax” rose from ~0.3% to ~2.3% of operating cost (up to ~4.1% for high-value goods)19.
• Cold-chain physics: roughly +1 °C per ~10 minutes outside the chain13; chilled > 5 °C or frozen > −18 °C = compromised.

Risk & mitigation matrix

Dark Stores & the Next Decade

The next layer: on-demand grocery is the fastest-growing part of the logistics picture. A dark storeA retail-format facility closed to walk-in shoppers, optimised purely for picking and packing online orders for rapid delivery. removes the conflict between online pickers and in-store shoppers fighting for the same shelf. Expect more micro-fulfilment, automation and township reach over the next decade.

Where the numbers are going (generic, SA)

• SA online retail reached ~R71bn in 2023 and is on track to exceed ~R130bn by end-2025 — approaching ~10% of total retail25. On-demand grocery is the fastest segment within that.
• One large discounter’s sub-hour service now runs multiple dedicated dark stores and fulfils from hundreds of regular stores turned into micro-fulfilment nodes; that one service alone grew revenue by ~35% in a recent half-year and now represents roughly a tenth of group sales21.
• A premium grocer’s on-demand app grew ~70% year-on-year and is adding more dark stores2223; a value retailer’s online business grew nearly 50% and is now profitable on a fully-costed basis24.
• Estimates put an untapped township economy on the order of ~R900bn25 — both a threat (more competition) and an opportunity (better supply access than ever).

What this means inside the DC and on your floor

• Automation creeps in: conveyor sortation and automated guided vehicles move totes; IoT sensors flag the moment a truck’s temperature drifts26.
• Labour still dominates: because local wages are lower than in Europe/US, full automation is hard to justify — but voice picking, route optimisation and WMS are already standard.
• Your store may become a node: staff could be asked to pick online orders from your shelves while serving walk-in customers — plan dedicated pickers and clear processes.
• Resilience goes green: solar + battery backup at DCs to cut diesel dependence is the coming default.

Practical Playbooks by Role

Expand the panel for your role. Each is a short, actionable checklist you can act on this week.

Glossary, References & Developer Notes

Glossary

References

Every citation in this playbook is a live link to its source. Hover a number in the text to see the source; click to open it.