OperationsAdvanced8 min read

Distribution Network Design

Distribution Network Design is the strategic decision of where to place warehouses, distribution centers (DCs), and fulfillment nodes — and how product flows between them — to minimize total landed cost while meeting service-level promises (e.g., 2-day delivery to 95% of customers). It is one of the highest-leverage decisions in operations because network footprint locks in 60-70% of future logistics costs and 80% of delivery speed for the next 5-10 years. The math optimizes across four cost pools: inbound freight (factory → DC), DC fixed costs (rent, labor, equipment), outbound freight (DC → customer), and inventory carrying cost (more nodes = more safety stock). Add one node, you cut outbound freight ~30% but raise inventory ~15%. The best networks aren't the cheapest — they're the ones that put inventory where it can ship fastest at the lowest marginal cost.

Also known asDC Network DesignLogistics Network OptimizationWarehouse Network StrategyFulfillment Network Design

Challenge a friend Browse library

The Trap

The classic trap: optimizing the network for today's demand pattern instead of where you'll be in 3-5 years. Companies build a 5-DC network for $400M revenue, then scale to $1.2B and discover they need 12 nodes — but they've signed 10-year DC leases. The other trap is over-indexing on outbound freight savings without modeling inventory cost: every additional DC raises safety stock by √n (square root of nodes), so going from 4 to 16 DCs doubles inventory carrying cost. Most companies also forget the 'service tail' — 95% coverage feels like a marginal step from 90%, but the last 5% requires DCs in markets that will never generate the volume to pay for themselves. Optimize for 90-92% same-day-shippable, not 99%.

What to Do

Run network optimization every 24-36 months — sooner if revenue doubles or channel mix shifts (e.g., wholesale to DTC). Build a model with: (1) demand by 3-digit ZIP, (2) all candidate node locations with rent/labor/utilities cost, (3) inbound freight from origin ports/factories, (4) outbound parcel/LTL rates by lane, (5) service-level constraint (e.g., 90% of demand within 2-day ground). Use tools like Llamasoft (now Coupa Supply Chain), Optilogic, or Python with PuLP/Gurobi. Test scenarios: +1 DC, +2 DCs, regional vs hub-and-spoke, 3PL vs owned. Pick the network that's optimal for your demand profile 24 months out, not today's, and prefer 3PL nodes for the 'long tail' DCs you may close later.

Formula

Total Network Cost = Inbound Freight + Fixed DC Cost + Variable DC Cost (per unit) + Outbound Freight + Inventory Carrying Cost (where Inventory ≈ base inventory × √(# nodes))

In Practice

Amazon's FBA distribution network grew from 7 fulfillment centers in 2005 to 175+ by 2020 and 1,000+ nodes (including delivery stations and sortation centers) by 2023. The strategy: place inventory within 50 miles of 80% of US urban population, enabling same-day and next-day delivery without paying premium parcel rates. By owning the middle-mile and last-mile, Amazon converted what was a $40-60B annual outbound freight bill into a strategic moat — competitors using FedEx/UPS couldn't match the speed at the same cost. The lesson: network density above a threshold becomes a competitive weapon, not just a cost line.

Pro Tips

01
The √n inventory rule: doubling distribution nodes doesn't double safety stock — it multiplies it by √2 ≈ 1.41x. Use this to model the inventory penalty of network expansion before committing to leases.
02
Always include 'phantom inventory' cost: extra inventory needed to balance across nodes due to forecast errors. Real-world inventory growth from network expansion runs 1.5-2x the theoretical √n.
03
Lease length is a strategic decision, not a real estate decision. Sign 3-5 year leases (with options) on 'experimental' nodes; only sign 10-year leases on the 4-5 anchor DCs you're certain you'll keep.

Myth vs Reality

Myth

“More distribution centers = better service”

Reality

Past a certain density, additional DCs create more complexity than they save. The 5th DC in a 4-DC network might cut outbound freight by 8% but raise total cost by 3% from inventory and operating overhead. Network design is an inverted-U curve — find the peak, don't keep climbing.

Myth

“Outsourcing to 3PL eliminates network design decisions”

Reality

Even with a 3PL, you choose which of their facilities to use, how to allocate SKUs across nodes, and what service tier to pay for. The optimization math is the same — you just don't own the buildings. Companies that 'outsource the problem' to 3PLs typically pay 15-25% more than they would with deliberate network design.

Try it

Run the numbers.

Pressure-test the concept against your own knowledge — answer the challenge or try the live scenario.

🧪

Knowledge Check

Your DTC company currently ships from 1 East Coast warehouse. 60% of customers are West of Mississippi and complain about 5-day delivery. You're considering adding a Las Vegas DC. Which factor will MOST surprise you in the cost model?

Industry benchmarks

Is your number good?

Calibrate against real-world tiers. Use these ranges as targets — not absolutes.

Logistics Cost as % of Revenue

Consumer goods / e-commerce DTC

Best-in-class

< 6%

Good

6-9%

Average

9-12%

High

12-16%

Out of control

> 16%

Source: Council of Supply Chain Management Professionals (CSCMP) Annual State of Logistics Report

Real-world cases

Companies that lived this.

Verified narratives with the numbers that prove (or break) the concept.

📦

Amazon

2005-2023

mixed

Amazon scaled from 7 fulfillment centers in 2005 to 175+ by 2020, then exploded to 1,000+ nodes (including delivery stations) by 2023. The strategy: place inventory within 50 miles of 80% of US urban population. This converted what could have been a $50B+ annual parcel bill (paying FedEx/UPS) into an owned middle-mile/last-mile network. By 2022, Amazon Logistics delivered more US packages than UPS or FedEx individually.

Fulfillment Centers (2005)

Total Nodes (2023)

1,000+

% US Population within 1-day

~80%

Annual Logistics Capex

$60B+ peak

Network density above a threshold becomes a strategic moat, not just a cost line. But Amazon over-built in 2020-2021 (pandemic surge) and announced closures of 14+ facilities in 2022. Even Amazon learned the hard way that building for peak demand burns cash when demand normalizes.

Source ↗

🏪

Walmart

1962-Present

success

Walmart's network strategy was radical: place stores in rural towns competitors ignored, then build distribution centers in concentric circles to serve them. Each DC serves 100-150 stores within a 250-mile radius — a 'hub and spoke' that minimized outbound freight. By 1990, Walmart had 1,400 stores served by 19 DCs — a ratio competitors couldn't match. The cross-docking innovation (90%+ of products bypassed warehouse storage) eliminated inventory carrying cost while maintaining 24-48 hour replenishment.

Stores served per DC

100-150

Avg distance store to DC

<250 miles

Cross-dock %

85-90%

Logistics cost vs Sears (1990)

~3% lower

Network design is a multi-decade commitment. Walmart's geographic strategy (rural-first hub-and-spoke) created cost advantages that took competitors 30 years to neutralize — and many never did. Choose your geometry deliberately.

Source ↗

Decision scenario

The DTC Network Expansion Decision

You're VP Operations at a $80M DTC apparel brand. Currently 1 East Coast DC ships everything. 55% of customers are West of Mississippi and getting 5-day delivery (industry standard is now 2-day). NPS is dropping. You can: add 1 West Coast DC, add 2 (West + Central), or sign with a 3PL like ShipBob (15+ nodes nationally).

Annual Revenue

$80M

Current DCs

1 (East Coast)

% West Coast customers

55%

Current outbound freight

$6.4M (8% of rev)

Current inventory

$12M

Decision 1

Each option has trade-offs. Owned DCs require $1.5M+ in fixed cost each plus $1.2M one-time setup; you control the experience but are locked in. ShipBob charges per-pick + per-shipment but gives you 15-node coverage. Your 24-month forecast: revenue grows to $130M.

Build 2 owned DCs (West + Central) — control the customer experience, optimize for long-term cost at scaleReveal

You commit $2.4M one-time + $3M annual fixed cost. Outbound freight drops 30% ($1.9M savings) but inventory grows to ~$20M ($2M added carrying cost). Net Year 1: roughly break-even on the $3M added overhead. By Year 3 at $130M revenue, the network is optimized and you're saving $4-5M/year vs single-DC. The catch: you're locked into 7-year leases. If revenue stalls or channel mix shifts (e.g., wholesale wins), you're stuck with overhead.

Year 1 Net Impact: ≈ break-evenYear 3 Annual Savings: +$4-5MRisk Exposure: 7-year leases

Sign with 3PL (ShipBob) for 3 nodes initially, expand as you grow — variable cost, no lock-in, scales with youReveal

You pay 8-12% more per shipment than owned DCs would at scale, but eliminate $2.4M one-time capex and have zero lease risk. Outbound freight + 3PL fees come to ~$5.8M (vs $6.4M baseline). Customer delivery times drop to 2-3 days nationwide within 60 days. By Year 3 at $130M, you're paying $2-3M/year more than owned would, but you have flexibility to shift channels, add 5 more nodes, or pivot if the business changes. Smart for sub-$200M brands; revisit at $250M+.

Time to deploy: 60 days vs 9 monthsCapex: $0 vs $2.4MYear 3 Cost vs Owned: +$2-3M (insurance premium)

Related concepts