How AI Can Help Returned Products Reach Their Next Destination Faster

A US retail customer returns a $100 holiday sweater in early December. If the retailer treats every return the same way, the item follows a default path: routed to a central facility, queued for standard inspection and repackaging, eventually shipped to a discount partner in January. The retailer recovers about 50% of the product’s worth.

If instead the retailer uses an AI-driven decision engine that recognizes the customer as a trusted buyer with a reasonable return history, predicts no refurbishment is needed, identifies the item as in-season merchandise, and routes the sweater directly to a nearby store for resale within days, recovery climbs to roughly 75%. The product, the customer, and the return event are identical. The difference is the dispositioning decision — what to do with the item and where to send it next.

This is the central insight from McKinsey & Company’s February 2026 reverse logistics research: among 30 supply chain executives surveyed, more than half cite dispositioning as their greatest returns management challenge, because most of the total cost of a return is concentrated at this stage. And of those same 30 leaders, all but five rely on basic data or no data to make dispositioning decisions. The decision that determines most of the cost is being made with the least intelligence.

What Dispositioning Actually Is

Dispositioning is the operational decision sellers face the moment a return is initiated: what to do with the product and where it should move next. Options vary by item condition, customer profile, product margin, seasonality, and resale potential — restock for direct resale, refurbish for second-life sale, route to third-party liquidation marketplace, recycle, donate, or harvest for parts.

Also Read: Returns Management Is the Ultimate Test for Your TMS. Here’s Why

Historically, dispositioning has been manual — costly inspection at a central facility before deciding the pathway, often through slow linear processes that erode product value with every day the item sits queued. The cost concentration comes from cumulative effects: shipping to default intermediary locations, manual inspection labor, queue time during which seasonal value decays, redistribution shipping if the disposition decision changes downstream. Each step adds cost; each delay erodes recovery.

US retail returns reached $849.9 billion in 2025 per NRF and Happy Returns data — 15.8% of annual retail sales — with McKinsey estimating $1 trillion in 2024 through different methodology. US retailers spend an estimated $200 billion annually recovering value from these returns. The dispositioning decision sits at the cost concentration point, which makes it the highest-leverage moment for intelligence in the entire reverse logistics flow.

The Data Gap

Twenty-five of thirty supply chain leaders McKinsey surveyed rely on basic data or no data for dispositioning. That gap is significant because most US retailers already have the data they need.

They know their customers — return history, lifetime value, claim accuracy. They know their products — margin profile, seasonality, shelf life, resale potential, defect history. They know their supply chain — shipping costs, inspection capacity, redistribution timelines. They know their network — store locations, inventory positions, demand forecasts by zone. The data exists in customer systems, product catalogs, supply chain platforms, retail operations data, and demand planning systems.

What’s missing is integration. Dispositioning decisions get made in isolation from this data — manually by warehouse staff inspecting items, or by software running on rules disconnected from current operational state. The decision gets made, but without the intelligence the operation already has.

Also Read: AI in Reverse Logistics: Turning Returns into a Competitive Advantage

What Dynamic AI Dispositioning Requires

A dynamic disposition model combines customer, product, supply chain, and operational data into a single AI-driven decision engine that routes each returned item to its highest-value outcome in real time. The architectural components:

Real-time data integration across customer history, product attributes, demand forecasts, network capacity, and logistics cost. Decision at initiation — the dispositioning pathway determined the moment the return is initiated through the customer-facing portal, not after the item enters the network. Risk-adjusted routing weighing the trusted-customer profile against the high-risk return-fraud profile to inform inspection requirements. Demand-aware routing that restocks when demand still exists and routes to second-life channels when it doesn’t. Integration with active logistics operations that can pick up returns on existing delivery routes (round-trip optimization), route directly to stores with available capacity, or move items to the right re-commerce endpoint without unnecessary intermediary stops.

Per McKinsey, the bigger issue across US retail isn’t the absence of solutions but limited and uneven adoption — many retailers use dispositioning software in isolation rather than as part of an end-to-end returns process.

What Changes Operationally

Three scenarios show the operational difference dynamic dispositioning creates.

An unknown customer returns a high-defect-rate item. Static process routes to central inspection regardless. Dynamic dispositioning flags the customer/product combination for risk-adjusted inspection, but routes the inspection to the geographically closest facility rather than the default central location — saving transit time and reducing the queue impact.

A trusted customer returns in-season merchandise. Static process routes to central facility, inspects, repackages, ships to discount partner months later. Dynamic dispositioning routes directly to nearby store for resale within days, recovering 75% rather than 50% of product value (per McKinsey’s example).

A customer returns an out-of-season item with seasonal value tied to next year. Static process discounts or liquidates immediately. Dynamic dispositioning identifies seasonal holding economics, routes to appropriate storage with the right inspection level, and re-releases when demand returns — capturing value the immediate-liquidation pathway destroys.

In each case, the dispositioning decision determines whether returns generate cost recovery or cost destruction.

Also Read: AI-Powered Reverse Logistics ROI – The Returns Architecture

The Evaluation Framework for US Retailers

For US VPs of Operations, Heads of Supply Chain, and Heads of Returns evaluating dispositioning architecture in 2026, six dimensions matter:

Decision-at-initiation capability, customer/product/supply chain data integration depth, real-time demand-aware routing, risk-adjusted inspection routing, integration with active logistics operations (round-trip optimization, store routing, multi-carrier), end-to-end visibility across the forward and reverse flow. Per CSCMP State of Logistics Report research on US last-mile economics, the operational maturity gap across these dimensions correlates materially with overall returns cost performance.

The real question for US retailers: given that dispositioning is the cost concentration point of US retail returns and the decision is currently being made with the least data, are we treating dispositioning as a core capability with integrated architecture — or as a back-end process running on manual judgment and isolated software?