Omnichannel Fulfillment Strategy: A CPG Operator's Framework

A CPG brand can be doing healthy revenue across Amazon, Walmart, DTC, and wholesale while losing the economics that made the growth worthwhile. One channel carries expensive marketplace fulfillment, another requires retailer-compliant replenishment, DTC absorbs pick-and-pack labor and parcel costs, and wholesale moves larger orders with very different service requirements. The team sees sales by channel, but not the full cost of serving each one. That's where an omnichannel fulfillment strategy earns its place. The objective isn't to offer more delivery options or promise faster shipping. It's to decide where inventory should sit, which node should fulfill each order, how much service the customer values, and whether the resulting contribution margin justifies the promise. Why Most CPG Brands Get Omnichannel Fulfillment Wrong A $5 million CPG brand selling through Amazon, Walmart, and its own site often reaches the same uncomfortable point. Revenue looks diversified, demand is present, and every channel manager has a growth plan. Yet contribution margin keeps compressing because each channel has developed its own inventory pool, routing habits, packaging standards, and reporting logic. Amazon inventory sits in one location, Walmart stock sits in another, and DTC orders ship from whichever warehouse has capacity. A wholesale buyer receives a separate allocation, sometimes based on an old forecast rather than current velocity. The brand may be carrying enough total inventory, but the wrong units are in the wrong nodes. Teams then pay for transfers, split shipments, expedited freight, or avoidable markdowns. This is a multichannel operating problem, not an omnichannel fulfillment strategy. Each channel may work independently, but the network doesn't make decisions as one system. A customer's location, the product's margin, available inventory, promised delivery date, and fulfillment cost should influence routing together. Instead, brands often let channel ownership determine the answer. Operator rule: Treat inventory as a network resource first, then apply channel priorities and reservations. Don't let a marketplace silo dictate the entire supply chain. The retail model has moved toward this networked approach. Forrester describes cross-channel experiences, live inventory visibility, and multiple ordering options as standard offerings among leading retailers, with buy online, pick up in store often preceding ship-from-store, ship-to-store, endless aisle, and buy online, return to store capabilities. A 2024 benchmark found retailers operating across an average of three channels, while adoption of omnichannel fulfillment tools rose from 31% in 2023 to 40% in 2024 (Forrester's omnichannel fulfillment benchmark). The customer-facing promise is now common. The economics behind it are not. The same Forrester reference reports that 61% of respondents considered multiple ordering and delivery options, including BOPIS, critical to their omnichannel brand strategy, while 60% said active stores were the final inventory location before delivery (Forrester's omnichannel fulfillment benchmark). Retailers clearly value flexibility, but flexibility without allocation rules can turn into expensive operational improvisation. The network has changed Traditional fulfillment assumed a central distribution center would receive inventory, process orders, and ship to customers or stores. Omnichannel networks add stores, third-party locations, lockers, marketplace fulfillment centers, and 3PL nodes to that design. A 2022 survey found that 56% of omnichannel retailers used live stores as the final inventory location before delivery, while 31% relied on distribution centers and 24% used third-party locations such as storage lockers (2022 omnichannel retailer survey). That diversification can shorten delivery distance and improve inventory utilization, but it also introduces more labor, more handoffs, and more opportunities for inaccurate availability. The brands that solve this well don't ask, “How do we make every channel faster?” They ask, “Which fulfillment path produces the required service level at an acceptable cost?” RedDog's discussion of omnichannel retail strategy is useful context because fulfillment belongs inside the broader retail operating model, not in a separate logistics workstream. Assessing Your Current Fulfillment Foundation Before adding nodes or changing providers, establish what the current network costs. Most brands have channel revenue reports, inventory reports, and carrier invoices, but they don't have a consistent contribution-margin view that connects all three. Start with a SKU-by-channel inventory map. For every item, identify on-hand units, available-to-promise units, reserved units, damaged units, inbound units, and the location of each quantity. Then ask whether Amazon, Walmart, Shopify, the WMS, and wholesale portals are using the same definition of “available.” If one system includes reserved inventory while another excludes it, the network is already making decisions from conflicting data. Build the true cost per order Calculate contribution margin by channel, not merely gross margin. A practical model starts with net sales and subtracts product cost, marketplace fees, fulfillment fees, pick and pack charges, shipping, storage, payment costs, advertising where you're evaluating channel profitability, returns processing, and expected damage or shrinkage. The important detail is granularity. A DTC subscription order, a single-unit Amazon order, a Walmart multipack, and a wholesale case shipment may all contain the same product but generate different costs. Use actual invoices and settlement data wherever possible. Estimates are acceptable for an initial baseline, but label them clearly and replace them with observed costs as the program matures. Packaging deserves its own audit. Amazon's 2026 U.S. FBA fee card applies different tables by size, shipping weight, price band, and order quantity. A small standard item weighing 2 ounces or less carries a fee of $2.43 during non-peak periods and $2.62 during peak periods, while a large standard item from 3+ pounds to 20 pounds carries $6.15 plus $0.08 per 4-ounce interval above the first 3 pounds during non-peak periods, and $6.69 plus the same increment during peak periods (Amazon's 2026 FBA fulfillment fee card). That isn't a minor packaging detail. A carton redesign, added protective insert, or dimensional change can push a SKU into another fee band and alter the channel decision. Audit execution quality Record order accuracy by node and channel. Use shipped orders as the denominator and track wrong SKU, wrong quantity, damage, late dispatch, and cancellation separately. A single blended accuracy number can hide a problem at one warehouse or store. Returns need the same treatment. Ask who authorizes the return, where it goes, how quickly it's inspected, whether it can be resold, and which channel receives the recovered unit. A structured approach to multi-channel inventory management can help teams identify gaps before they redesign routing. Finally, map the integrations. List the OMS, WMS, ERP, marketplace connectors, EDI or wholesale portals, carrier systems, returns platform, and analytics layer. For each connection, document the source of truth, update frequency, error handling, and owner. If a person exports a spreadsheet to keep a channel accurate, that workflow is a risk, not a system. Choosing the Right Fulfillment Models for Each Channel No single fulfillment model wins across every CPG channel. The right choice depends on velocity, margin, order profile, geographic demand, service commitment, and operational complexity. A fulfillment center works well when inventory is concentrated, demand is predictable, and the brand needs consistent process control. A 3PL makes more sense when the brand needs variable capacity, multiple locations, value-added services, or expertise it doesn't want to build internally. A hybrid network often works best once the brand has enough scale to separate high-velocity marketplace inventory from DTC, wholesale, and regional demand. For a useful operational distinction between storage space and an active fulfillment operation, review this Peak Transport warehouse guide. The difference matters because a warehouse that stores product may not have the systems, labor model, or service controls required to process marketplace and DTC orders. Channel Best Model Cost Driver When to Use Amazon FBA, FBM, or a controlled hybrid Fulfillment fees, storage, shipping weight, returns, and marketplace requirements Use FBA for eligible velocity and service needs. Test FBM when product economics, dimensions, or inventory control justify direct fulfillment. Walmart WFS, seller-fulfilled, or hybrid Dimensions, weight, fulfillment fees, and oversize exposure Use WFS where the fee and service profile support the item. Keep direct fulfillment available for products that become margin-dilutive through dimensional penalties. DTC In-house, 3PL, or regional hybrid Pick and pack, parcel shipping, packaging, returns, and customer-service handling In-house can work with concentrated demand and simple workflows. A 3PL is more practical as geography and order complexity expand. Wholesale DC, retail-compliant 3PL, or manufacturer shipping Case handling, freight, labeling, routing guides, chargebacks, and payment terms Prioritize consistency, fill rate, and retailer compliance over parcel speed. Stores Store fulfillment, BOPIS, or ship-to-store Labor availability, inventory accuracy, staging space, and handoff quality Use stores when local inventory is reliable and the service promise can be executed without disrupting core store operations. Amazon economics require order-level analysis Amazon FBA fees can look acceptable at the single-unit level and deteriorate when the product is bulky, heavy, low-priced, or return-prone. The reverse can also be true for bundles and multipacks. Amazon's fee guidance shows that a small standard item weighing 4 ounces or less costs $7.34 for a 1-unit order, $4.98 per unit for 2 units, $4.10 per unit for 3 units, and $3.64 per unit for 4+ units (Amazon's multi-unit FBA fee guidance). That makes basket construction part of fulfillment strategy. A profitable multipack may not be a profitable single unit, and an advertised bundle must be evaluated against packaging, referral, storage, and return behavior rather than product margin alone. Walmart and DTC need different controls Walmart Fulfillment Services also prices around item dimensions and weight, with an added $3 surcharge for Oversize 1 items that exceed at least one defined dimensional threshold (Walmart Fulfillment Services fee guidance). Audit the packaged product, not the product specification. A carton can cross a cutoff even when the item itself hasn't changed. For DTC, compare more than provider rates. Consider pick density, zone distribution, branded packaging, subscription handling, kitting, customer-service requirements, and return disposition. Wholesale should usually remain a distinct workflow because retail routing guides, case quantities, labels, appointments, and chargebacks create different failure costs. The routing rule should be explicit: prioritize margin contribution, then inventory velocity, then service-level requirements, with channel reservations protecting contractual or marketplace commitments. Don't route every order to the closest node if that node is expensive, inaccurate, or needed for a higher-value demand stream. Building the Tech Stack and Integration Layer An omnichannel fulfillment strategy fails quickly when the technology layer treats inventory as separate channel buckets. The foundation should be a single shared inventory pool with controlled reservations. That doesn't mean every channel can consume every unit. It means the business maintains one network view, then applies rules for marketplace commitments, wholesale allocations, safety stock, and service-level protection. The inventory layer must distinguish physical stock from sellable stock. Reserved units, damaged units, quality-hold units, inbound inventory, and available-to-promise inventory should have separate statuses. Channel reservations then protect the units that a promise depends on without creating permanent fragmentation. Assign every system a job The OMS should capture and orchestrate orders. The WMS should manage warehouse execution, including receiving, picking, packing, and shipping confirmation. Marketplace and commerce integrations should transmit orders, inventory, tracking, cancellations, and status updates to Amazon, Walmart, Shopify, and wholesale systems. A distributed order management layer can sit within or alongside the OMS to apply routing logic. That logic might consider inventory location, shipping cost, promised delivery date, product age, node capacity, and channel priority. The point isn't to create an elaborate model on day one. It's to replace human-by-human routing with documented rules that the system can execute consistently. A practical independent benchmark recommends inventory synchronization in under 5 minutes, no manual intervention per channel, and written order-accuracy targets above 99.8% (omnichannel fulfillment operating benchmarks). Those are operating controls, not marketing claims. If a brand can't meet the sync requirement, it should lower available-to-promise inventory or tighten reservations rather than publish an availability promise the network can't support. Measure exceptions, not just averages Create an exception queue for oversells, delayed acknowledgments, failed tracking updates, inventory mismatches, routing fallbacks, and cancellations. Each exception needs a reason code and owner. A dashboard showing total orders without exception detail tells management that volume moved, not whether the network worked. Track accuracy at the unit and order level. An order containing several items can be partially wrong even if most units are correct. Review accuracy by facility, shift, SKU family, and order type so the team can distinguish a training problem from a barcode, slotting, or integration problem. For broader guidance on connecting commerce, data, and operational systems, use this CPG ecommerce tech stack framework. The integration layer should reduce manual work, not create another reporting environment that operators must reconcile every morning. The Hidden Costs and Execution Risks Brands Underestimate A store can sit closer to the customer and still produce a worse contribution margin than Amazon FBA, Walmart WFS, a 3PL, or a centralized facility. The missing calculation is labor and disruption. Store associates must find the unit, pick and pack it, print the label, stage the order, and hand it to the carrier while keeping shelves replenished and customers served. A 2025 optimization study found that ship-from-store can improve profits as price, fulfillment cost, market share, and customer distribution change, but the result depends on planning assumptions that rarely hold perfectly (2025 omnichannel fulfillment optimization research). Store inventory may improve the customer promise while raising labor cost, interrupting replenishment, or creating inaccurate availability when cycle counts fall behind. Pickup satisfaction depends on execution Availability does not create a good pickup experience by itself. MIT's 2025 omnichannel performance survey reported order fulfillment as a challenge for 50% of respondents and returns management as a challenge for 47% (MIT Omnichannel Supply Chain Lab survey). Home delivery remained dominant at 74%, while click-and-collect, in-store shopping, and curbside pickup each sat around the mid-50% range in that survey. A 2026 retail experience study identified the moments shoppers value most: wait time on arrival at 39.1%, smooth and confusion-free pickup at 38.4%, and same-day product availability for pickup at 33.2%. BOPIS should go live only when the store can locate, stage, verify, and hand over orders consistently. A commerce platform can enable the feature, but it cannot fix weak store labor planning or unreliable inventory records. Service-level test: Define the promise around the customer's actual handoff experience, not the timestamp when the order was accepted. Returns add cost across every channel. The business pays for reverse transportation, inspection, restocking, disposal, customer service, and potentially another outbound shipment. Shrinkage and damage can turn a returned unit into a write-off instead of recovered inventory. Store picking also carries an opportunity cost when employees leave replenishment or customer service to process online orders. Measure these costs against net revenue and channel contribution margin, not against shipping speed alone. Optimization models still need judgment Centralized joint inventory and fulfillment planning can outperform decentralized methods. One algorithmic study reported average cost reductions of 11.81% for short horizons and 3.93% for longer horizons versus prior methods (joint inventory and fulfillment planning study). The finding supports centralized decisions, but operators still need to review the assumptions behind each routing rule. Forecasts miss. Store counts become stale, carriers change performance, and promotions shift the order mix. Models need buffers, fallback nodes, cutoff rules, and manual escalation for exceptions. The MIT Omnichannel Supply Chain Lab frames the work as integrating online and offline distribution into the broader go-to-market strategy, rather than adding faster delivery options (same MIT Omnichannel Supply Chain Lab survey). Each rule should be tested against contribution margin by SKU and channel before it is scaled. Your Phased Implementation Roadmap A practical rollout follows RedDog's Foundation, Optimization, Amplification sequence. The order protects contribution margin. Automating inaccurate inventory data or scaling an unprofitable routing rule makes correction harder. Foundation from days 1 to 30 Audit inventory by SKU, location, channel, and status. Map OMS, WMS, ERP, marketplace, commerce, carrier, and returns connections. Establish contribution margin by channel, then baseline order accuracy, cancellations, late shipments, returns, and storage. Do not proceed until the team can identify sellable inventory, calculate each channel's service cost, and confirm which system owns every key data field. Optimization from days 31 to 60 Set routing rules around margin, velocity, geography, inventory age, service requirements, and channel reservations. Compare Amazon FBA, Walmart WFS, 3PL, in-house, and direct-ship economics at SKU level. Pilot one controlled change, such as regional ship-from-store fulfillment, a DTC 3PL lane, or a marketplace bundle with different order economics. A control-tower approach helps monitor exceptions and network performance across nodes. Kagool's overview of implementing a control tower offers context for organizing that visibility. Amplification from days 61 to 90 Scale only models that meet their margin and service gates. Expand channel coverage gradually, add inventory buffers where results support them, and retire workflows requiring repeated manual correction. Use weekly inventory reviews, monthly margin analysis, and quarterly network optimization to maintain governance. Prioritize channels by contribution margin and operational complexity, not revenue alone. The strongest network is one the team can operate accurately as demand, inventory, and carrier conditions change. Reddog Consulting Group helps CPG founders and operators review channel contribution margin, marketplace performance, inventory velocity, and fulfillment routing across Amazon, Walmart, DTC, and wholesale. Book a free 30-minute strategy call with Reddog Consulting Group to identify margin leakage and build a practical fulfillment optimization plan.

Unleashing Insights

/

fulfillment models

Omnichannel Fulfillment Strategy: A CPG Operator's Framework

Posted on August 24, 2026

Contact

Marketplaces

Reddog Consulting Services

Connect with us