Supply Chain Optimization

Most e-commerce supply chains are not designed; they accumulate. A brand signs a warehouse contract because it needed one fast, adds a second carrier after a peak-season failure, and layers on marketplace requirements one channel at a time. Two years later, freight spend is up, delivery promises are inconsistent, and nobody can say what it actually costs to move one order from factory to doorstep.

This guide is a practical framework for turning that accumulation back into a design. It covers the four levers that drive most of the cost and service improvement available to mid-size brands: network design, inventory placement, freight consolidation, and a disciplined KPI baseline that makes continuous improvement possible. The full PDF version includes worksheets, a KPI glossary, and a 90-day optimization plan; this web edition gives you the core method.

Who This Guide Is For

• Operations and supply chain leads at brands selling across Amazon, Walmart, Target, Shopify, or eBay
• International brands entering the U.S. market that need a distribution footprint without building one from scratch
• Founders and COOs who suspect their logistics costs are too high but lack a baseline to prove it
• Teams preparing to renegotiate 3PL, carrier, or freight contracts and wanting leverage grounded in data

What You Will Learn

1. How to map your current network and quantify its true cost per order
2. Network design principles: how many nodes you need and where they belong
3. Inventory placement rules that cut split shipments and zone-skipping costs
4. Freight consolidation tactics for inbound, middle-mile, and outbound moves
5. The KPI baseline: OTIF, fill rate, cost per order, and inventory turns
6. Running a continuous improvement loop that compounds quarter over quarter

Start With a Baseline, Not a Project

Optimization without measurement is guesswork. Before changing anything, spend two weeks building a baseline across four metrics:

• OTIF (on-time, in-full): the percentage of orders delivered complete and by the promised date. Marketplaces effectively enforce this through seller metrics; retail programs like Target and Walmart enforce it through chargebacks.
• Fill rate: the percentage of demand you could actually fulfill from stock. Anything below 95 percent means you are paying for demand generation and then failing to convert it.
• Fully loaded cost per order: inbound freight, storage, pick and pack, packaging, outbound parcel, returns processing, and chargebacks, divided by orders shipped. Most brands only track two or three of these components.
• Inventory turns and weeks of cover: by SKU, not just in aggregate, so slow movers cannot hide behind bestsellers.

If you can only track one number, track fully loaded cost per order. It is the single metric that exposes bad network design, poor inventory placement, and wasteful freight all at once.

Network Design: Fewer Decisions, Made Deliberately

The core question is how many fulfillment nodes you need and where. The pattern for most brands:

• One node works when you ship under roughly 1,000 orders per month or your delivery promise is 4-7 days. Prioritize a location near your inbound port and your densest customer region.
• Two nodes (typically one West, one East) covers about 85 percent of the U.S. population within two-day ground transit and usually cuts average parcel zones by 1.5 to 2 zones.
• Three or more nodes only pays off with high volume, heavy or bulky products, or strict retail delivery commitments. Every added node increases safety stock and adds a stock-out surface.

The common mistake is treating warehouse selection as a real estate decision. It is a portfolio decision: you want the flexibility to add, drop, or shift nodes as your channel mix changes. This is why Pi-Commerce operates through a vetted multi-warehouse network rather than owned buildings; brands get placement flexibility without long leases. See how that model works under warehousing services.

Inventory Placement: Put Stock Where Demand Is

A two-node network with wrong placement performs worse than a well-run single node. Three working rules:

1. Split by demand, not evenly. Allocate each SKU across nodes in proportion to regional demand history. A 50/50 split of a SKU that sells 80 percent in the East guarantees costly cross-country transfers.
2. Classify SKUs before you place them. Fast movers belong in every node. B and C movers usually belong in one node only; the occasional slower shipment is cheaper than duplicated safety stock.
3. Rebalance on a cadence, not in a crisis. Monthly review of placement versus actual regional sales prevents the slow drift that produces split shipments, where one order ships from two buildings and you pay double parcel cost.

Demand forecasting is what makes placement proactive instead of reactive. Forecasting at the SKU-by-region level, the approach behind Pi-Commerce's demand forecasting AI, lets you position inventory ahead of demand shifts rather than chasing them.

Freight Consolidation: The Overlooked Middle

Parcel gets the attention, but inbound and middle-mile freight often hide 10-20 percent savings:

• Consolidate inbound. Multiple small ocean or air shipments cost far more per unit than fewer, fuller containers. Coordinate purchase orders so factories ship on a consolidated cadence.
• Use zone skipping. Line-haul palletized orders to a regional hub, then inject them into the parcel network close to the customer. This converts high-zone parcel rates into cheap LTL plus low-zone parcel.
• Audit accessorial charges. Detention, reweighs, address corrections, and residential surcharges routinely add 5-10 percent to freight bills. Audit monthly; carriers rarely volunteer refunds.
• Match mode to margin. Air freight on replenishment of a healthy-margin bestseller can be rational; air freight to cover a forecasting miss on a C-mover never is.

For international brands, consolidation starts at origin. Pi-Commerce's global freight team manages origin consolidation, customs, and drayage as one flow rather than three handoffs.

The Continuous Improvement Loop

Optimization is a loop, not a project. Run it quarterly:

1. Re-measure the four baseline KPIs and compare against last quarter
2. Identify the single largest gap between target and actual
3. Diagnose the root cause with order-level data, not averages
4. Change one lever: placement, node count, carrier mix, or packaging
5. Hold the change for a full cycle before judging it, then lock in the win

Brands that run this loop typically find 3-5 percent cost-per-order improvement per cycle for the first year. The prerequisite is unified data: when orders, inventory, and freight live in separate systems, the diagnosis step fails. A single operating view, such as the Pi Data Center commerce data platform, is what makes the loop repeatable.

Key Takeaways

• Build the KPI baseline first: OTIF, fill rate, fully loaded cost per order, and inventory turns by SKU
• Two well-placed nodes cover most U.S. two-day promises; add nodes only when volume and product profile justify the added safety stock
• Place inventory in proportion to regional demand and rebalance monthly to kill split shipments
• Inbound consolidation, zone skipping, and accessorial audits recover freight cost that parcel negotiations never touch
• Treat optimization as a quarterly loop with one lever changed at a time, powered by unified order and inventory data

Go Further

The full PDF edition includes the cost-per-order worksheet, a network design decision tree, and the 90-day optimization plan. If you would rather have this executed than templated, Pi-Commerce designs and operates the entire flow, from network selection through daily fulfillment, as a 4PL partner with skin in your results. Explore our integrated supply chain services or contact us for a baseline assessment of your current network.