Automated Return Disposition

When a return lands on the dock, the disposition decision has to happen fast and consistently. Restocking a damaged unit wastes labor and shelf space. Writing off a saleable item burns margin. Liquidating too early leaves money on the table; holding too long ties up inventory that will never sell at full price.

Automated return disposition scores each returned unit against condition signals, recovery cost, current demand, and expected resale value, then recommends one path: restock, repair, liquidate, recycle, or write off. The model does not grade merchandise or move units. A returns supervisor or returns associate still inspects the item, confirms condition, and sends it down the chosen lane.

The value is ranking and consistency under volume. High-velocity return centers process mixed categories, incomplete paperwork, and uneven grading skill across shifts. A shared scoring frame reduces “gut feel” variance between supervisors and makes exceptions easier to audit later.

Disposition quality depends on four input groups. If any required group is missing, the system should return empty output rather than guess.

Condition. Graded state of the unit after inspection: new/unopened, open-box excellent, light cosmetic damage, functional defect, missing parts, hygiene or safety fail. Photos, checklist answers, and category-specific rules (apparel, electronics, consumables) all belong here. Condition is the primary gate for restock eligibility.

Cost to recover. Labor and materials to inspect, clean, rebag, repair, or rebox, plus inbound freight already spent and outbound cost if the unit must move to another node. A cheap SKU with high touch cost often loses to liquidation even when the unit looks fine.

Demand. Sell-through rate, weeks of supply, active promotions, seasonality, and whether the SKU is still in the active assortment. High demand for a clean open-box unit supports restock or open-box listing. Dead or discontinued SKUs favor liquidation or recycle even when condition is strong.

Resale value. Expected net recovery on the recommended channel: full-price or open-box on the primary site, marketplace, outlet, B2B liquidator, or scrap credit. Compare that recovery to the cost of holding and processing. Write-off is appropriate when net recovery after cost is at or below scrap and the unit cannot safely re-enter commerce.

Supporting context helps but does not replace the four groups: original order value, return reason code, warranty or manufacturer RMA eligibility, hazmat or recall flags, and channel rules (marketplace returns vs store returns). Related workflows such as Refund Policy Compliance Checker and Return Fraud Risk Scoring run alongside disposition; they answer different questions (refund eligibility and abuse risk) and should not silently override a condition-based disposition without a documented policy rule.

The recommendation is a ranked policy, not a black-box label. Practitioners should be able to explain why a unit landed on a given path.

Restock when condition meets sellable standards, recovery cost is low relative to margin, demand is healthy, and the unit can return to the correct location or virtual inventory without policy breach (sealed consumables, regulated goods, and hygiene categories often block restock regardless of appearance).

Repair when the defect is known, parts and bench time are available, and expected post-repair resale value minus repair cost beats liquidation and restock-as-is. Electronics and appliances are common candidates; fashion rarely is unless alteration cost is trivial.

Liquidate when the unit is saleable on a secondary channel but not worth primary restock: cosmetic damage, open-box with soft demand, discontinued SKUs, or recovery cost that erodes primary margin. Choose liquidation over write-off when a liquidator or outlet bid clears a floor after handling cost.

Recycle when the unit fails safety, hygiene, or brand standards for resale, contains recoverable materials, and local compliance favors material recovery over landfill. Batteries, certain packaging, and damaged hard goods often fall here once resale is blocked.

Write off when no safe or economic recovery path remains: severe damage, contamination, total loss, or scrap credit that does not justify further handling. Write-off should be explicit so finance and inventory systems stay aligned.

Tie-breakers should be configurable: prefer restock for hero SKUs during stockouts; prefer liquidation when DC capacity is constrained; force recycle or destroy on recall and safety holds irrespective of value.

The model recommends; staff still grade and send. That split keeps liability and quality control with people who can see the unit.

Human-in-the-loop steps

1. Associate completes condition grading and captures required photos or checklist fields.
2. System scores only when condition, cost, and demand inputs are present and valid for the category.
3. Supervisor sees the recommended path, the top drivers (for example, “high touch cost vs open-box ASP”), and any policy blocks.
4. Supervisor accepts, overrides with a reason code, or requests re-grade.
5. Downstream systems receive the final disposition for putaway, RMA, liquidation ASN, recycle, or write-off.

Empty output is required when condition grade is missing or incomplete, recovery cost cannot be estimated for the category, or demand/resale signals are unavailable for that SKU and location. Do not default to restock or write-off. Surface a clear “insufficient inputs” state so the associate finishes data capture or routes the unit to a manual exception queue.

Overrides should be logged with who, when, and why. Patterns in overrides (for example, frequent “restock” overrides on a liquidate recommendation for one category) are training and policy signals, not noise to suppress.

Disposition sits downstream of intake and reason capture. Return Reason Classification structures why the customer returned the item; those codes inform condition expectations and whether a manufacturer claim or quality loop is warranted, but they do not replace physical grade.

Fraud and policy tools protect refund and abuse decisions. A high fraud score may change refund treatment while disposition still follows condition and safety rules for the physical unit. Keep those decision objects separate in the UI so supervisors do not conflate “deny refund” with “destroy unit.”

Operational KPIs for this page’s outcome (speed) include time from receive to final disposition code, percent of units with a first-pass recommendation accepted, override rate by reason, and aged returns still sitting in “pending grade” or “insufficient inputs.” Accuracy KPIs belong beside them: restock reverse rates, liquidation recovery vs quote, and write-off dollars per return.

Start with a narrow category set where cost and demand feeds are reliable, require empty output on missing inputs, and keep supervisors in the accept/override loop until override rates stabilize. Expand categories only when grading checklists and cost tables are ready for those SKUs.