Disposition Technology and Fraud Prevention

SmartDisposition and Image AI

The most valuable decision in a returns DC is the disposition decision: where does this item go next? Every routing error costs money — underrecovered value, unnecessary handling, or inventory past its optimal sell window.

Optoro SmartDisposition (Now Blue Yonder)

Ingests product category, condition grade, SKU-level demand, seasonal timing, and current inventory levels across available channels. Routes each unit to highest-value next destination: full restock, outlet channel, branded recommerce, third-party liquidation, or donation. Routing is dynamic — adjusts channel prioritization based on real-time supply and demand, not a static waterfall rule table.

Rules-based routing breaks down as SKU catalog and channel mix evolve (maintenance burden). ML-driven routing updates based on actual recovery outcomes, learning which channel generates the best margin for which product type under which timing conditions. Operational as part of Blue Yonder following August 2025 acquisition.

Clarity X-Ray AI (ReturnPro)

X-ray imaging + proprietary AI classifies product condition and verifies component presence without unboxing — each unit processed in 3.2 seconds. Primary use case: catching empty box scams and component substitution at point of receipt before the refund processes. A returned electronics unit that looks intact externally may contain bricks or a worthless substitute component.

Flagged units don’t route to open-box resale at 70% of MSRP — they route to fraud review, then to parts auction at 10–15% of MSRP.

B2B Liquidation Channels

B-Stock Solutions

Largest B2B liquidation network in the U.S. Runs private branded auction storefronts for nine of the top ten U.S. retailers: Amazon, Walmart, Target, Home Depot, Samsung, Whirlpool. Auction format = market-driven price discovery via real-time competitive bidding.

• Average recovery: ~31% of MSRP
• Platform fee: ~10% of final bid price
• Electronics lots with per-unit retail >$100 recover 83% more than lots with retail <$35
• For computers/tablets: smaller lots under $6,000 recover 30% more than $50,000–$100,000 lots
• Smartwatches recover approximately 7x more than printers in liquidation

Lot structure decisions directly affect recovery. Configuring lots by product type and size is a financial optimization, not a logistics convenience.

Liquidation.com (Liquidity Services)

Primary B-Stock competitor. FY2024: $363M revenue, $1.4B GMV (SEC 10-K). Average recovery: ~24% of MSRP. Longer-established platform with roots in government surplus. Both deserve evaluation; optimal choice depends on product category and buyer network.

Branded Resale at Scale: Trove

If liquidation is the bottom of the recovery stack, branded resale is the upper tier.

Trove provides white-label recommerce operating system for Patagonia (Worn Wear), REI (Re/Supply), Lululemon (Like New), Levi’s (SecondHand), Arc’teryx (ReGear), Eileen Fisher (Renew). Active in 700+ U.S. stores by 2022 (Lululemon across 390 stores, REI across 170). REI’s recommerce grew 86% YoY in 2021.

Trove model: Brand handles customer interaction, gets ESG credit. Trove takes 15–25% revenue share in exchange for intake software, pricing intelligence, optional physical processing, storefront technology, and listing management.

Recovery economics: 30–50% of original MSRP after platform fees — structurally higher than liquidation (15–31%), but requires operational investment in trade-in programs and recommerce storefronts.

Trove + Recurate (August 2024): Recurate acquisition added peer-to-peer resale — brands can facilitate direct C2C transactions without taking physical possession. Trove is now the only platform offering both brand-operated B2C resale and C2C P2P in a single system.

Dynamic pricing engines: Static percentage-of-retail pricing for recovered merchandise guarantees a suboptimal outcome. Optoro SmartDisposition, Trove’s AI pricing engine, and ReturnPro’s re:Source AI system all use real-time market data to set and adjust recommerce pricing. Every day at a static price point, recovery is left on the table.

Returns Fraud: Scale and Typology

Scale

U.S. retailers lost $101 billion to returns fraud in 2023 — 13.7% of all returns (NRF / Appriss Retail). Holiday fraud rate: 16.5%, or $25 billion of $148 billion in holiday return volume.

At a brand doing $500M revenue at 20% return rate: 13.7% fraud rate = $13.7M in annual fraud loss before detection costs.

Fraud Typology

Type	Description	Detection Difficulty
Wardrobing	Buy for event, return after (69% of shoppers admit; up 38% in 2024)	Low without behavioral analytics
Empty box scam	Return empty packaging or weight-matched filler	Low without X-ray or weight verification
Return of stolen merchandise	Affects 44% of retailers	Moderate (requires receipt validation)
Returnless refund abuse	Claim non-receipt to get refund without returning	High (fastest-growing online fraud 2024 per Appriss)
Wardrobing sub-type	Average loss $45–$120/incident; detection rate <15% without dedicated tooling	Low

The Returnless Refund Reality

59% of retailers now implement keep-it / returnless refund policies — up from 26% the prior year. Economics are straightforward: if reverse shipping + processing costs exceed maximum recoverable value, the retailer loses less by issuing the refund and telling the customer to keep or dispose.

A $20 item with $15 reverse logistics cost and $10 maximum recovery: better P&L as a returnless refund. Amazon, Walmart, Target, Chewy, and Wayfair all issue returnless refunds on items typically valued $20–$300, depending on customer LTV and product condition. Requires per-SKU cost modeling; blanket policies leave money on the table.

Detection Platforms

Platform	Approach	Claimed Impact
Appriss Retail Engage	AI at point of return authorization; cross-retailer behavioral database enables network-level fraud signal sharing	8–12% reduction in return dollars
Riskified	480+ attributes per transaction; guarantees order approvals (chargeback covered if Riskified-approved order fraudulent)	594% three-year ROI (Forrester TEI)
Signifyd Intelligent Returns	Classifies returns as fraud/abuse/wardrobing/bracketing/legitimate via ML	Up to 75% reduction in fraudulent returns; $0.06–$0.12/transaction
Forter	Processes $200B+ in annual online commerce; payment fraud, returns abuse, account takeover, policy manipulation	Per-transaction guarantee model
Happy Returns Return Vision AI	Behavioral risk scoring at drop-off point; flags before item enters network	Non-optional for high-value CE/luxury

Optimal fraud strategy: Personalized, not uniform. Allow-list high-LTV customers. Scrutinize new/zero-history accounts. Implement SKU-level dynamic policies. Aggressive gatekeeping creates false positives that damage loyal customers — 80% of shoppers have stopped shopping at a retailer because of return policy changes.

Hardware: Dimensioners, Sortation, AMRs

Dimensioners

Every returned item needs weight and measurement: fraud verification (matches product spec?), freight billing, WMS routing.

• Cubiscan — market leader; countertop scan stations to in-line conveyor-mounted units with WMS integration
• Cargo Spectre — 3D scanning with ML algorithms; WMS-agnostic; more cost-effective for SMB/mid-market; runs on 110V standard outlets

Automated Sortation for Returns

OPEX Sure Sort / Sure Sort X: Purpose-designed for returns. 2,400 items/hour with a single operator. Six-sided barcode scanning handles any item orientation — critical in returns where items arrive without consistent packaging orientation. Sure Sort X extends to items up to 20 lbs. Modular (add capacity via iBOTs without facility reconfiguration). OPEX positions Sure Sort X against AMRs: same throughput, smaller footprint, less manual intervention.

BEUMER BG Sorter (cross-belt): Industrial standard for 10,000+ UPH. Cross-belt preferred over tilt-tray for polybag items (lower spill risk) and mixed-dimension SKU profiles characteristic of e-commerce returns.

Shoe sorters: Workhorse for high-volume lower-diversity returns (footwear, apparel with consistent sizing).

Robotics in Returns

Returns operations are the hardest robotics application domain: every unit arrives in unknown condition, unknown packaging, unknown contents. Practical robotics entry point today: post-grading AMR putaway — after a human grades and assigns a destination, AMRs move the item to the right location.

• Locus Robotics Origin AMRs at nGroup: 229% improvement in putaway productivity (124 → 285 lines/hour) on RaaS at ~$1,500/robot/month
• 6 River Systems (Chuck): comparable AMR capability on lease model with native Shopify integration

Autonomous grading without human oversight is not yet standardized. Most facilities deploy humans with camera-capture stations. The next 3 years will see significant investment in vision-based autonomous condition assessment.

Reason Code Design

25–50 codes total in reason code groups. A 200-code scheme degrades to random selection under production pressure.

Build product-type specific subcodes:

• Apparel: “Sleeve length” under “Fit”; “Fading” under “Damaged”
• Electronics: “Screen damage” under “Damaged”; “Missing charger” under “Incomplete”
• Furniture: “Assembly issue” under “Defective”; “Missing part” under “Incomplete”

Generic codes like “Quality issue” are not actionable.

Separate consumer-entered codes from warehouse-confirmed codes. High divergence between stated and confirmed reason = fraud signal or product quality problem. Map every code to resaleable vs. not resaleable and feed to merchandising weekly.

Reference: Microsoft Dynamics 365 disposition code architecture separates reason codes (why customer returned) from disposition codes (what happens to the item). That separation is non-negotiable in any WMS-based returns build.

The 8-KPI Returns Dashboard

KPI	Description
Gross return rate	Returns as % of total sales by category
Net return rate	Gross rate adjusted for returnless refunds and denials
Recovery rate	Revenue recovered as % of MSRP across all grades and channels
Cost per return	Fully-loaded processing cost per unit (all 9 cost components)
Days to disposition	From dock arrival to final channel routing
% auto-routed	Share of volume handled by rules engine without human inspection
Fraud rate	Fraudulent returns as % of total returns (NRF benchmark: 13.7%)
Customer NPS on returns	Net Promoter Score specifically for the return experience

All 8 metrics reviewed together weekly by operations, finance, and merchandising. Returns cannot be managed from a single vantage point.