Disposition rules: turning grades into decisions

Returns disposition rules are the logic that turns a condition grade into an action: they combine the grade with item attributes and route every returned parcel to a lane automatically, so no one has to decide item by item. A grade on its own is inert. It only earns its keep when it triggers a decision, and disposition rules are how that decision gets made in seconds instead of days. At areturnz, the platform applies these rules per return the moment a parcel is graded, an operator can override any outcome, and every override is logged.

This guide explains the four core disposition lanes, how to design rules from grade plus attributes, how rule precedence resolves conflicts, why override logging matters for audit, and how the automation collapses decision latency inside our roughly 48-hour median cycle.

Why a grade needs a rule

Our AI condition model assigns each return an A/B/C/R grade with a confidence score, detected tags, and a photo bundle. If you want the mechanics of the grade itself, see our breakdown of ABCR AI condition grading. The grade answers "what state is this item in." It does not answer "what do we do with it." That gap is where value leaks: a warehouse that grades fast but decides slowly still pays for dwell time, manual review, and missed resale windows. The real cost of a return is dominated by that decision latency and the labor around it, not the postage.

Disposition rules close the gap. A rule reads the grade and the item's attributes, then names a lane and a reason. Because the rule runs automatically the instant a return is graded, the item stops waiting for a human to look at it.

The four disposition lanes

• Restock - back to sellable inventory. Reserved for A-grade items (and clean B-grade for some categories) that pass hygiene and completeness checks. Speed here drives restock velocity, the metric that most directly protects margin.
• Liquidate - sold through secondary channels or resold under your own brand as open-box. Typical for B and C grades with cosmetic wear but full function. Recovering value under your label instead of a discount marketplace is covered in reselling returns under your brand.
• Donate - routed to charitable partners when an item is functional but not economical to resell, or when policy favors giving over dumping.
• Destroy / recycle - the last resort for R-grade, unsafe, or non-compliant items. Recycling beats landfill wherever the material stream allows, which is why our sustainability lane favors recovery over destruction.

How to design disposition rules

A durable rule set combines two inputs: the condition grade and item attributes. Attributes are the tie-breakers that a grade cannot capture on its own.

1. Start with the grade band. A restocks, B and C recover value, R exits. This is your default skeleton.
2. Layer attribute overrides. Hygiene-sensitive categories (intimates, cosmetics, opened consumables) cannot restock even at grade A. Hazmat and damaged-battery items cannot ship to liquidation. High-value items justify manual review even when the grade is confident.
3. Add seasonality and value thresholds. An end-of-season A-grade garment may liquidate rather than restock because it will not sell at full price before the season closes. A low-value C-grade item may donate rather than absorb liquidation handling cost.
4. Bind low-confidence grades to review. When the model's confidence sits under your threshold, route to a human queue instead of acting on a shaky grade. We unpack where to set that line in confidence scores on the line.

Example rule table

The table below shows representative rules mapping condition and attribute to a lane and the reason the platform records.

Rule precedence: resolving conflicts

Multiple rules can match one return, so precedence matters. Order your rules so that safety and compliance win first, hygiene next, then value and season, and finally the plain grade band. In the table above, a grade-A item that is also a damaged battery must land in destroy/recycle, not restock, because the safety rule outranks the grade rule. Design precedence as a top-down list where the first matching rule wins, and keep the destructive and compliance lanes at the top so a permissive grade rule can never accidentally ship something unsafe or unsellable.

The human override and why logging it matters

Automation handles the overwhelming majority of returns, but judgment cases remain: a borderline grade, a high-value SKU, a customer-service exception. An operator can override any lane the platform assigns. The override is not a silent edit. Every override is logged with who made it, when, the original lane, the new lane, and it attaches to the return's custody chain alongside the photos and grade.

That log is what makes the system auditable. When a brand asks why a specific parcel was destroyed instead of restocked, the answer is a record, not a recollection. Logged overrides also become training signal: a lane that operators reverse often points to a rule that needs retuning. Trustworthy decisions depend on that evidence trail, which is the same principle behind putting proof on every return.

How automation collapses decision latency

The reason disposition rules exist is speed. When a rule fires the moment a return is graded, the item moves to its lane without waiting in a decision queue. That is how areturnz holds a median cycle near 48 hours across 180K+ processed returns. Each avoided manual decision is dwell time removed and, for restock lanes, inventory returned to sale sooner. The rule engine, the grade, the photo bundle, and the webhooks (return.received, graded, dispositioned, restocked) fire in sequence so downstream systems act in near real time. The full event and payload contract lives in the node spec.

Frequently asked questions

What is a returns disposition rule?

A disposition rule is conditional logic that reads a return's condition grade and item attributes, then routes it to a lane such as restock, liquidate, donate, or destroy/recycle. It converts a grade into an automatic action and records the reason for the outcome.

How do restock, liquidate, donate, and destroy differ?

Restock returns an item to sellable inventory; liquidate recovers value through secondary or open-box channels; donate sends functional but uneconomical items to charity; destroy/recycle exits unsafe, non-compliant, or non-functional items, favoring material recovery over landfill.

Can an operator override an automatic disposition?

Yes. The platform applies rules automatically, but any operator can override the assigned lane. Every override is logged with the actor, timestamp, and before/after lanes, and it attaches to the return's custody chain for audit.

How do disposition rules speed up the returns cycle?

Because a rule fires the instant a return is graded, the item skips the manual decision queue and moves straight to its lane. Removing that decision latency is a core reason our median cycle sits near 48 hours.

What happens when the AI grade confidence is low?

Low-confidence grades should not trigger destructive or irreversible lanes. Bind them to a review queue so an operator confirms the grade before the item is dispositioned, protecting against acting on an uncertain read.

Related reading: Returns Evidence and Dispute Proof: The Complete Guide

Related reading: Building the Dispute-Ready Evidence Dossier