Returns Process Design

The Full Decision Tree: Receive to Disposition

The design objective is to make the disposition decision as early as possible — ideally before the item arrives at the dock — because every step without a clear destination costs money and delays recovery.

Customer initiates return (in RMS: Loop, Happy Returns, Narvar)
↓ Drop-off or carrier pickup
↓ Hub consolidation (Happy Returns / Optoro Express)
↓ Returns receiving dock
↓ Unboxing / breakdown from transport carton
↓ System check-in (scan UPC or RFID into RMS / WMS)
↓ Grading / inspection
↓ Photography (if routing to secondary marketplace)
↓ DISPOSITION DECISION
├── Grade A → Repackage → Return to forward inventory as new
├── Grade B → Open-box / B-grade secondary channel
├── Grade C → Repair / refurbish queue
├── Grade D → Parts harvesting (CE, industrial)
└── Grade E → Certified recycling / destruction

Each branch requires its own downstream workflow, labor requirement, space allocation, and recovery economics. Treating returns as a single flow rather than five parallel flows is the primary process design error.

The Grading System: A Through E

Grade	Condition	Recovery Path	MSRP Recovery
A	Like new, all packaging intact	Immediate restock as new	80–100%
B	Minor cosmetic issues, functional	Open-box / secondary marketplace	40–65%
C	Functional but damaged/missing pkg	Repair / refurb before resale	20–40%
D	Non-functional but parts viable	Parts harvesting	5–15%
E	Damaged beyond economic repair	Recycling / destruction	0–5%

No universal grading standard exists. Amazon Renewed, Best Buy open-box, and Apple refurbished all use different rubrics — the imperative is to define grades explicitly, train to them consistently, and track grade distribution over time.

Grade distribution is diagnostic: 60% Grade A suggests heavy bracketing behavior or excellent packaging. 35% Grade C+ suggests carrier handling problems, product durability issues, or misleading product descriptions.

Touch Reduction: The $1–3 Rule

Every additional human touch on a returned unit costs approximately $1–3 in direct and indirect labor. Design principle: count the touches in your current process and eliminate each one that doesn’t add grade certainty or disposition accuracy.

Four touch-reduction disciplines:

1. Pre-sort at drop-off — If the drop-off network captures return reason and customer condition assessment, route obvious Grade A units to a fast-track restock lane before entering the full inspection queue.

2. Zone routing by disposition — Once graded, items physically move to a zone where only that disposition activity happens. Mixing workflows creates cross-contamination and adds unnecessary walk time.

3. System-first, hands-second — Scan into the RMS before picking the unit up to inspect. Know the item cost, category, and system-recommended disposition before human evaluation. Eliminates the “pick up, find information, put back down” redundant touch.

4. Auto-disposition for high-confidence cases — Rules engine auto-routes at 60–70% of volume (obvious Grade A and Grade E cases). Human inspection handles exceptions, high-value items, and flagged uncertainty (remaining 30–40%).

Inspection Time Standards by Category

Category	Inspection Time
Apparel (basic check)	20–45 seconds
Footwear	30–60 seconds
Small CE (accessories, chargers)	60–90 seconds
Complex CE (phones, laptops)	3–8 minutes
Furniture / large items	10–30 minutes

Three time costs frequently underestimated:

• Photography: +60–90 seconds per unit for secondary marketplace resale. At 500 units/day needing photography and 75 sec/unit: 10 labor hours at the photo station before any listing work. Model explicitly or face floor backup.
• Repackaging: +30–90 seconds per unit (new inner box + poly bag + label = 90 sec; just resealing = 30 sec).
• Mystery returns: 3–8 minutes per unit — packages with no RMA, wrong item, or damaged ID require manual research. Track separately from standard time benchmarks.

The Photography Bottleneck

Photography is the throughput constraint most operations miss until it’s too late. To sell Grade B/C inventory on any secondary marketplace, you need photography of individual unit condition — stock images are inadequate because buyers are paying specifically for condition transparency.

Solutions in order of capital intensity:

1. Manual photo station — One associate, camera on mount, ring light, neutral backdrop. 30–60 items/hour. Lowest capital; requires staffing allocation.

2. Automated photography conveyor — Items move on belt through lightbox with fixed cameras. 60–120 items/hour. Optoro SmartDisposition integrates AI that photographs and applies ML to assess cosmetic condition from images — reducing the human grading step for certain SKU types.

3. AI image grading — Camera + ML model inspects cosmetic condition as items pass a sensor. Current accuracy: ~70–80% matching human grades. Sufficient for easy cases; human review queue required for edge cases and high-value items.

If the photo station backs up — volume surge, software crash, SKU reconfiguration — the entire Grade B/C resale queue backs up. For fashion, idle time at the photo station has a direct markdown cost attached.

Auto-Disposition: Rules Engine vs. AI Hybrid

Rules-Engine Auto-Disposition: Define logical rules combining return reason code, item history, customer history, and time since purchase. Example: if return reason = “wrong size” AND category = “apparel” AND purchase < 7 days ago AND customer < 2 returns in past 90 days → route Grade A without physical inspection. Works for 40–50% of volume at well-structured retailers.

Optoro SmartDisposition: ML model trained on return images and item data routes to highest-margin next channel at point of return — before items reach the DC. Kept 94% of client inventory out of landfills in 2022. Integrates with in-store and locker networks.

Hybrid Operating Model: Auto-disposition handles 60–70% of volume; human inspection handles the remaining 30–40% (high-value, ambiguous condition, system-flagged uncertainty). Build measurement systems to track auto-disposition accuracy over time — a rules engine accurate six months ago may drift as SKU mix shifts.

Practical Rules

• Grade A units must re-enter forward inventory within 24–48 hours of dock arrival.
• Track grade distribution by SKU and return reason code weekly. Divergence between consumer-stated reason and warehouse-confirmed condition signals fraud or product quality issues.
• The inspection team is making financial decisions. Train and measure them as financial decision-makers, not as production workers.

See Returns Facility Engineering for the physical layout that absorbs these process flows. See Returns Throughput and Labor Modeling for labor staffing math. See Returns Management Software for the RMS that orchestrates the consumer-facing decision tree.