@techwavedev/agi-agent-kit 1.5.3 → 1.6.1

package/templates/skills/extended/other/returns-reverse-logistics/references/decision-frameworks.md

@@ -0,0 +1,823 @@
+# Decision Frameworks — Returns & Reverse Logistics
+
+This reference provides the detailed decision logic, scoring matrices, financial models,
+grading standards, and disposition workflows for returns and reverse logistics operations.
+It is loaded on demand when the agent needs to make or recommend nuanced return-handling
+decisions.
+
+All thresholds, timelines, and cost assumptions reflect US retail and e-commerce operations
+with applicability to omnichannel, pure-play e-commerce, and brick-and-mortar environments.
+
+---
+
+## 1. Disposition Decision Trees by Product Category
+
+### 1.1 Decision Methodology
+
+Every returned item follows a decision tree that routes to the highest-value disposition
+channel. The routing decision is made after grading (see §5) and considers:
+
+1. **Recovered value** at each disposition tier (restock, open box, refurbish, liquidate, donate, destroy)
+2. **Processing cost** for each tier (inspection, repackaging, refurbishment, shipping)
+3. **Time-to-recovery** — cash tied up in returns inventory has a carrying cost
+4. **Regulatory constraints** — some dispositions are prohibited for certain categories
+5. **Brand protection** — premium brands may restrict secondary-market sales
+
+The **net recovery** for any disposition = `(Sale price at channel) - (Processing cost) - (Shipping cost) - (Channel fees)`.
+
+Always route to the disposition with the highest net recovery, subject to regulatory and
+brand constraints.
+
+### 1.2 Consumer Electronics
+
+Consumer electronics are the highest-value and most complex returns category. Serial numbers,
+firmware states, activation locks, and functional testing requirements add cost and time.
+
+```
+RECEIVE → Verify serial number matches RMA
+  ├── Mismatch → Flag for fraud review (swap fraud), HALT processing
+  └── Match → Visual inspection
+        ├── Grade A (no cosmetic defects, all accessories, original packaging)
+        │     └── Functional test (power on, screen, connectivity, battery health)
+        │           ├── Pass → Check activation lock / factory reset status
+        │           │     ├── Locked → Contact customer for unlock, hold 48 hrs
+        │           │     │     ├── Unlocked within 48 hrs → Restock as new
+        │           │     │     └── Not unlocked → Grade B (open box with disclaimer)
+        │           │     └── Clean → Restock as new (full margin recovery)
+        │           └── Fail → Route to refurbishment assessment
+        │                 ├── Refurb cost < 40% of refurb selling price → Refurbish
+        │                 ├── Refurb cost 40-60% → Liquidate as "for parts / not working"
+        │                 └── Refurb cost > 60% → Parts harvest or e-waste recycling
+        ├── Grade B (minor cosmetic wear, accessories complete, packaging damaged)
+        │     └── Functional test
+        │           ├── Pass → Repackage as "open box" or "renewed" (60-80% of retail)
+        │           └── Fail → Refurbishment assessment (same tree as above)
+        ├── Grade C (visible wear, scratches, missing non-essential accessories)
+        │     └── Functional test
+        │           ├── Pass → Sell through secondary channel at 30-50% of retail
+        │           └── Fail → Parts harvest if unit value > $100, else e-waste
+        └── Grade D (heavily damaged, non-functional, missing critical components)
+              └── Parts harvest if any component value > $15, else e-waste recycling
+```
+
+**Category-specific thresholds:**
+- Smartphones: Always verify IMEI against stolen device databases (GSMA) before restocking
+- Laptops: Battery health must be > 80% for Grade A restock; 60-80% triggers Grade B
+- Tablets: Check for MDM (mobile device management) profiles — enterprise tablets may have
+  remote-lock capability that surfaces post-sale
+- Headphones: Hygiene concern — all ear tips/pads replaced before resale ($2-8 per unit)
+- Smart home devices: Factory reset verified; linked account removal confirmed. A smart lock
+  that is still linked to a previous owner's account is unsellable and potentially a safety issue
+
+**Typical processing costs:**
+| Action | Cost per Unit | Time per Unit |
+|--------|--------------|---------------|
+| Visual inspection | $1.50-2.50 | 45-90 seconds |
+| Functional test (basic) | $3.00-5.00 | 2-4 minutes |
+| Functional test (full diagnostic) | $8.00-15.00 | 10-20 minutes |
+| Repackaging | $5.00-12.00 | 3-8 minutes |
+| Refurbishment (minor: screen clean, reset) | $15.00-30.00 | 15-30 minutes |
+| Refurbishment (moderate: component replacement) | $30.00-80.00 | 30-90 minutes |
+| Data wipe (NIST 800-88 compliant) | $5.00-10.00 | 5-15 minutes |
+
+### 1.3 Apparel and Footwear
+
+Apparel returns are high-volume, low-unit-value, and condition-sensitive. Odour, stains,
+and stretched fabric are the primary defects. Speed is critical because fashion depreciates
+rapidly — a trend item loses 10-20% of sellable value per month.
+
+```
+RECEIVE → Scan RMA / order lookup
+  └── Visual inspection (30-60 seconds)
+        ├── Tags attached, no signs of wear
+        │     ├── Original packaging intact → Restock as new
+        │     └── Packaging damaged/missing → Repackage, restock as new (tag is key, not box)
+        ├── Tags removed but no signs of wear
+        │     └── UV light + odour check
+        │           ├── Clean → Restock as "like new" or outlet (80-90% of retail)
+        │           └── Traces detected → Grade C, route to launder/clean assessment
+        │                 ├── Cleaning cost < $5 and item value > $30 → Clean and restock as outlet
+        │                 └── Cleaning not viable or cost > value threshold → Liquidate by weight
+        ├── Visible wear, stains, or damage
+        │     ├── Premium brand (retail > $100) → Assess repair viability
+        │     │     ├── Repair cost < 20% of outlet price → Repair and sell through outlet
+        │     │     └── Repair not viable → Liquidate (never destroy premium apparel; brand resale exists)
+        │     └── Standard brand → Liquidate by weight ($0.50-2.00/lb) or textile recycling
+        └── Heavily damaged, soiled, or biohazard
+              └── Textile recycling or destroy (biohazard requires specific disposal)
+```
+
+**Apparel-specific considerations:**
+- Seasonal timing is everything: a winter coat returned in February can restock for next season,
+  but the carrying cost of 8 months of storage may exceed liquidation recovery. Decision point:
+  if storage cost > (expected recovery next season × probability of sale) - liquidation value now,
+  liquidate immediately.
+- Footwear: Check sole wear. A shoe worn on carpet for 5 minutes is different from one worn
+  on pavement. Sole scuffing = Grade C minimum. Check for orthotics left inside.
+- Swimwear and intimate apparel: Once hygienic liner is removed, non-returnable per health code
+  in most jurisdictions. If returned, destroy — do not restock or liquidate.
+- Designer/luxury: Authenticate before accepting. Counterfeits in the return stream are
+  increasing. Compare serial numbers, stitching quality, hardware weight against known
+  genuine samples.
+
+### 1.4 Home, Furniture, and Large Goods
+
+Returns of bulky items are expensive — return shipping alone can be $50-200+. The disposition
+decision often happens before the item physically returns.
+
+```
+CUSTOMER INITIATES RETURN →
+  ├── Item value < return shipping cost × 2.5
+  │     └── Offer returnless refund (customer keeps item, full refund)
+  │           Cost justification: returnless refund costs $X (product value).
+  │           Processing the return costs $X (product) + $Y (shipping) + $Z (processing).
+  │           If Y + Z > 40% of X, returnless is cheaper.
+  ├── Item value > threshold AND item is in original packaging
+  │     └── Schedule carrier pickup → Receive → Inspect
+  │           ├── Grade A → Restock (furniture typically restock at 85-95% due to assembly/box condition)
+  │           ├── Grade B → Sell as "open box" in-store (avoid re-shipping; sell from nearest location)
+  │           ├── Grade C → Donate locally (shipping destroyed items is negative ROI)
+  │           └── Grade D → Local disposal (donation or recycling based on materials)
+  └── Item is assembled
+        └── Generally non-returnable once assembled (policy). Exceptions:
+              ├── Defective → Offer replacement parts first, full return if unfixable
+              ├── Missing parts on arrival → Ship missing parts (cheaper than full return)
+              └── Customer insists → Accept but apply 25% restocking fee to cover disassembly/repackaging
+```
+
+### 1.5 Health, Beauty, and Personal Care
+
+Regulatory constraints dominate this category. Once opened, most health and beauty products cannot
+be legally resold due to FDA and state health department regulations.
+
+```
+RECEIVE → Seal integrity check
+  ├── Sealed / unopened
+  │     ├── Expiration date > 6 months out → Restock as new
+  │     ├── Expiration 3-6 months → Restock with markdown or outlet
+  │     └── Expiration < 3 months → Donate (tax benefit > markdown recovery)
+  ├── Opened but appears unused (seal broken, product visually intact)
+  │     └── DESTROY. Cannot verify non-contamination. No restocking of opened health/beauty.
+  │           Exception: Hard goods (hair dryers, electric razors) → treat as electronics tree
+  └── Used
+        └── DESTROY. Biohazard disposal if applicable (used cosmetics applicators, skincare).
+```
+
+**Special cases:**
+- Prescription items: Cannot accept return under any circumstances (most states). Direct
+  customer to pharmacy or manufacturer disposal programme.
+- Supplements/vitamins: Same as cosmetics — once opened, destroy. Sealed returns restock
+  only with lot number verification.
+- Sunscreen: Regulated as OTC drug by FDA. Opened sunscreen is destroyed. Expired sunscreen
+  (even sealed) is destroyed — never sell expired OTC.
+
+### 1.6 Books, Media, and Software
+
+High restock rate, low processing cost. The primary fraud vector is digital content extraction
+(reading/ripping then returning).
+
+```
+RECEIVE → Condition check
+  ├── New condition (no creasing, bending, markings)
+  │     └── Restock as new. Media: verify disc is present and matches case.
+  ├── Minor wear (slight cover bend, shelf wear)
+  │     └── Restock at minor discount or sell through marketplace as "very good"
+  ├── Moderate wear (highlighting, writing, water damage)
+  │     └── Liquidate through bulk book buyers ($0.10-0.50 per book) or donate
+  └── Software / digital media
+        ├── If activation key is unredeemed → Restock
+        ├── If key is redeemed → Cannot resell. Write off. Pursue refund from publisher if within terms.
+        └── Physical media with digital code (game + download) → Sell disc only at reduced price
+```
+
+---
+
+## 2. Fraud Detection Scoring Model
+
+### 2.1 Scoring Architecture
+
+The fraud scoring model assigns points based on observable signals at the time of return
+initiation (pre-receipt signals) and at the time of physical inspection (post-receipt
+signals). The two scores are summed for a composite score.
+
+**Thresholds:**
+| Composite Score | Action |
+|----------------|--------|
+| 0-30 | Process normally. No additional review. |
+| 31-50 | Flag for passive monitoring. Process refund but add customer to watch list for 90 days. |
+| 51-64 | Enhanced inspection. Hold refund until physical inspection is complete and matches RMA description. |
+| 65-79 | Supervisor review. Hold refund. Detailed inspection with photo documentation. Supervisor approves or escalates. |
+| 80-100 | Fraud review team. Refund on hold. Customer contacted for "verification" (never say "fraud"). LP notified. |
+
+### 2.2 Pre-Receipt Signals (scored at return initiation)
+
+| Signal | Points | Logic |
+|--------|--------|-------|
+| Customer return rate > 30% (rolling 12 months, ≥ 5 orders) | +15 | High return rate alone isn't fraud, but it is a risk multiplier. Exclude exchanges from rate calculation. |
+| Return rate > 50% | +25 | Replaces the +15 above. At this rate, the customer is almost certainly bracket-shopping or abusing policy. |
+| Return initiated < 48 hours after delivery confirmation | +5 | Could be legitimate (wrong item, didn't match description) or bracket shopping. Mild signal. |
+| Return reason is "defective" but product category has < 2% defect rate | +10 | "Defective" is used to avoid restocking fees. True defect claims on low-defect products are suspicious. |
+| Return reason changed between online initiation and customer service contact | +10 | Inconsistency suggests the customer is constructing a narrative. |
+| Customer account age < 30 days | +10 | New accounts used for return fraud or testing fraud viability. |
+| Multiple returns in same week (3+) | +10 | Cumulative with return rate signal. Suggests bracket shopping or wardrobing batch. |
+| Return from an address different than the original shipping address | +10 | Excludes gift returns (flagged as gift at order). Otherwise indicates potential organised activity. |
+| Item is in a high-shrink category (electronics, designer, cosmetics) | +5 | These categories have higher fraud incidence. Mild base signal. |
+| No-receipt or no-order-match return | +15 | Receipt fraud is the primary risk vector. Match to payment method or loyalty ID. |
+| Order was placed with a promotion/coupon > 30% off | +5 | Price-arbitrage returns are more common on heavily discounted purchases. |
+| Customer has previously been flagged for fraud review (any outcome) | +15 | Prior flags, even if resolved as legitimate, indicate a pattern worth monitoring. |
+
+### 2.3 Post-Receipt Signals (scored during physical inspection)
+
+| Signal | Points | Logic |
+|--------|--------|-------|
+| Serial number mismatch (does not match the unit sold to this customer) | +40 | Near-certain swap fraud. Verify against order record before escalating — packing errors at fulfilment can cause legitimate mismatches. |
+| Product weight differs > 5% from expected for SKU | +25 | Indicates missing components, swap with lighter/cheaper item, or empty packaging. Weigh before opening. |
+| IMEI/MEID on returned device doesn't match sold device | +40 | Definitive swap fraud for mobile devices. Cross-reference IMEI from order with IMEI on returned device. |
+| Product shows wear inconsistent with stated return reason | +15 | "Changed my mind" return on a laptop with 200 battery cycles suggests extended use. |
+| Tags removed on apparel/footwear with "didn't fit" reason | +10 | Tags removed is consistent with wearing, not just trying on. |
+| Cosmetic traces on apparel (makeup, deodorant, perfume) | +15 | Wardrobing indicator. UV light reveals traces invisible to naked eye. |
+| Packaging has been repacked (tape over tape, non-original inner packaging) | +10 | Could indicate swap (customer repacked a different item) or simply customer repackaging for return. Context-dependent. |
+| Security/RFID tag removed | +20 | Tags that are not customer-removable (hidden tags, sewn-in RFID) should still be present. Removal suggests the item was worn/used in a retail environment. |
+| Product firmware/software shows usage history inconsistent with claim | +15 | Laptop claiming "unopened" but with 6 months of OS updates installed. |
+| Multiple units of same SKU returned (3+) in single return | +10 | Could be legitimate (sizing across colours) or reseller return. Check original order for bulk discount. |
+
+### 2.4 Score Adjustments and Overrides
+
+| Condition | Adjustment |
+|-----------|-----------|
+| Customer lifetime value > $10,000 and net LTV positive | -15 points (floor at 0) |
+| Customer is a verified loyalty programme member (2+ years) | -10 points |
+| Return is an exchange (not refund) | -10 points |
+| Return reason is verified fulfilment error (wrong item shipped) | Set score to 0, process immediately |
+| Return was pre-approved by customer service with case notes | -10 points |
+| Customer has filed a chargeback simultaneously with this return | +20 points (escalate regardless of score) |
+
+### 2.5 False Positive Management
+
+False positives destroy customer relationships. Every customer flagged by the fraud scoring
+system who turns out to be legitimate represents a risk of customer attrition. Manage through:
+
+1. **Never communicate "fraud" to the customer.** Use neutral language: "additional processing
+   time," "verification of your return," "quality review."
+2. **Time-box the review.** Flagged returns must be resolved within 5 business days. If the
+   review cannot conclusively determine fraud within 5 days, process the refund. The cost
+   of a false positive held for 3 weeks exceeds the cost of most fraudulent returns.
+3. **Track false positive rate monthly.** Target: < 3% of total returns flagged are
+   confirmed false positives. If rate exceeds 5%, recalibrate the scoring model.
+4. **Feedback loop:** Every fraud review outcome (confirmed fraud, confirmed legitimate,
+   inconclusive) feeds back into the scoring model calibration. Signals that generate
+   high false-positive rates have their point values reduced.
+
+---
+
+## 3. Vendor Recovery Framework
+
+### 3.1 Return-to-Vendor (RTV) Process
+
+RTV is the primary mechanism for recovering costs on defective products. The process:
+
+```
+Identify RTV-eligible unit (defective, vendor-caused quality issue, mispick at vendor DC)
+  │
+  ├── Check vendor agreement for RTV terms
+  │     ├── RTV window: Typically 90 days from retailer receipt of return (not customer purchase date)
+  │     ├── Minimum shipment value: Usually $200-500 per RTV shipment
+  │     ├── Documentation requirements: Varies by vendor (photos, defect codes, customer complaint data)
+  │     └── RTV authorisation: Some vendors require pre-approval; others accept "open RTV" under agreement
+  │
+  ├── Accumulate RTV-eligible units by vendor
+  │     ├── Stage in designated RTV area (separate from general returns inventory)
+  │     ├── Track aging — units approaching 90-day window need priority shipment
+  │     └── Batch by vendor to meet minimum shipment thresholds
+  │
+  ├── Obtain RTV authorisation (if required)
+  │     ├── Submit RTV request with: SKU, quantity, defect description, photos, customer return rate data
+  │     └── Vendor has 5-10 business days to approve/deny (per most vendor agreements)
+  │
+  ├── Ship to vendor return facility
+  │     ├── Use vendor-provided shipping label (if applicable — vendor pays)
+  │     ├── If retailer pays shipping: deduct from credit claim or use lowest-cost carrier
+  │     └── Track shipment and confirm delivery at vendor facility
+  │
+  └── Track credit issuance
+        ├── Vendor credit should appear within 30-45 days of vendor receipt
+        ├── If no credit at 30 days: first follow-up (email to vendor returns dept)
+        ├── If no credit at 45 days: escalate to vendor account manager
+        ├── If no credit at 60 days: debit memo against next PO (per vendor agreement terms)
+        └── If vendor disputes: provide defect documentation as evidence. Escalate to vendor management.
+```
+
+### 3.2 Defect Rate Monitoring and Claims
+
+Beyond individual RTV, monitor defect rates at the SKU level to identify systemic quality
+issues that trigger formal defect claims:
+
+| Defect Rate (per SKU, rolling 90 days) | Action |
+|----------------------------------------|--------|
+| < 2% | Normal. Process individual RTVs. No escalation. |
+| 2-5% | Alert vendor management. Request root cause analysis from vendor. Continue selling but monitor weekly. |
+| 5-8% | Formal quality complaint. Demand corrective action plan within 14 days. Consider pull from active sales pending vendor response. |
+| 8-15% | Pull from active sales. Formal defect claim for all returns above the 2% baseline. Negotiate credit or replacement. |
+| > 15% | Full stop-sale. Vendor compliance violation. Chargebacks for all returns + lost margin + customer service costs. Consider vendor termination. |
+
+**Defect claim calculation:**
+```
+Total returns for SKU in period: 500 units
+Expected baseline return rate (non-defect): 8% of units sold (industry avg for category)
+Units sold: 4,000
+Expected returns: 320
+Excess returns attributable to defect: 500 - 320 = 180 units
+Claim = 180 × (wholesale cost + inbound freight per unit + return processing cost per unit)
+       = 180 × ($24.00 + $1.80 + $7.50)
+       = 180 × $33.30
+       = $5,994.00
+```
+
+Add consequential costs if the defect caused customer service escalations, negative reviews
+that required response, or marketplace listing suppression.
+
+### 3.3 Vendor Chargeback Schedule
+
+For vendor-caused issues beyond defects (packaging failures, mislabelling, wrong items shipped
+from vendor DC), apply chargebacks per the vendor compliance programme:
+
+| Violation | Chargeback | Notes |
+|-----------|-----------|-------|
+| Wrong item shipped from vendor DC | 100% of product cost + return shipping + $25 processing fee | Requires photo evidence of received vs ordered |
+| Mislabelled product (UPC doesn't match contents) | $50 per incident + product cost if unsellable | Creates inventory accuracy issues downstream |
+| Packaging failure (product damaged due to inadequate packaging) | 100% of product cost + return processing | Requires photos of packaging condition at receipt |
+| Missing components (accessory, manual, warranty card) | Cost of sourcing replacement component + $10 processing | If component unavailable, full product cost |
+| Counterfeit or unauthorised product | 300% of product cost + $500 penalty per incident | Zero tolerance. Notify brand protection. |
+| Late shipment from vendor causing customer-facing delay | Customer credit issued + $15 processing | Must document customer complaint and credit |
+| Incorrect hazmat/regulatory documentation | $250 per incident + cost of regulatory remediation | Regulatory liability makes this non-negotiable |
+
+### 3.4 Vendor Recovery ROI Model
+
+Not all vendor recovery is worth pursuing. The ROI model:
+
+```
+Recovery ROI = (Expected recovery - Recovery cost) / Recovery cost
+
+Where:
+  Expected recovery = Claim amount × Collection probability
+  Recovery cost = Labour (documentation, communication, follow-up) + Shipping (if RTV) + Relationship cost
+
+Labour cost estimates:
+  - Simple RTV with existing authorisation: $15-25 per batch
+  - Defect claim requiring documentation assembly: $75-150 per claim
+  - Disputed claim requiring escalation and negotiation: $200-500 per claim
+
+Collection probability by vendor tier:
+  - Tier 1 (top 20 vendors, strong relationship): 85-95%
+  - Tier 2 (mid-tier, established relationship): 65-80%
+  - Tier 3 (small/new vendors): 40-60%
+  - International vendors (no US entity): 25-45%
+```
+
+**Decision matrix:**
+| Claim Amount | Tier 1 Vendor | Tier 2 Vendor | Tier 3 Vendor | International |
+|-------------|---------------|---------------|---------------|---------------|
+| < $100 | Offset against next PO | Offset against next PO | Write off | Write off |
+| $100-500 | Batch RTV | Batch RTV | Batch if > $200 total | Write off, note for contract |
+| $500-2,000 | Standard RTV/claim | Standard claim | Standard claim with escalation plan | Claim if > $1,000 |
+| $2,000-10,000 | Standard claim | Standard claim + account mgr | Account mgr + formal notice | Pursue with local agent if > $5,000 |
+| > $10,000 | VP-level engagement | VP-level + legal review | Legal review | Legal counsel in vendor's jurisdiction |
+
+---
+
+## 4. Return Policy Exception Matrix
+
+### 4.1 Exception Decision Framework
+
+When a return falls outside standard policy, the decision to grant an exception depends on
+a structured evaluation, not individual judgment calls. This matrix standardises the
+exception decision.
+
+**Step 1: Is the exception request covered by an automatic override?**
+
+| Condition | Action | No Further Analysis Needed |
+|-----------|--------|--------------------------|
+| Product is defective (verified or reasonably claimed) | Accept return, full refund, no restocking fee | Yes |
+| Fulfilment error (wrong item shipped, wrong quantity) | Accept return, full refund, prepaid return label | Yes |
+| Product is subject to active recall | Route to recall programme (not returns) | Yes |
+| Customer is in top 5% by LTV and request is first exception in 12 months | Grant exception, standard refund | Yes |
+| State or federal law requires acceptance (lemon law, cooling-off period) | Comply with applicable law | Yes |
+
+**Step 2: If not an automatic override, score the exception request:**
+
+| Factor | Score Range | Description |
+|--------|-----------|-------------|
+| Days past policy window | 1-30 days: +2 / 31-60 days: +5 / 61-90 days: +8 / >90 days: +12 | How far outside the standard window |
+| Product condition at return | Grade A: 0 / Grade B: +2 / Grade C: +5 / Grade D: +10 | Worse condition = higher cost of exception |
+| Customer LTV | Top 20%: -5 / Middle 60%: 0 / Bottom 20%: +3 | Valuable customers get more latitude |
+| Return reason credibility | Compelling story with evidence: -3 / Plausible: 0 / Weak: +5 | "My house flooded" with photos vs "I forgot" |
+| Precedent risk | Private resolution: 0 / Customer mentioned social media: +5 / Customer has large following: +8 | Public exceptions become policy expectations |
+| Product restockability | Restockable as new: -3 / Open box: 0 / Liquidation: +3 / Destroy: +5 | Restockable items cost less to accept |
+
+**Step 3: Interpret the exception score:**
+
+| Score | Decision | Authority Level |
+|-------|----------|----------------|
+| < 0 | Grant exception. Cost is minimal, customer value is high. | Returns associate |
+| 0-5 | Grant exception with standard refund. | Team lead |
+| 6-10 | Grant as store credit (not original payment refund). | Team lead |
+| 11-15 | Partial credit (50-75% of purchase price) as store credit. | Returns manager |
+| 16-20 | Deny with empathetic explanation and alternative offer (exchange, discount on next purchase). | Returns manager |
+| > 20 | Deny. Offer to connect with manufacturer warranty if applicable. | Returns manager |
+
+### 4.2 Common Exception Scenarios with Recommended Resolutions
+
+| Scenario | Typical Score | Recommended Resolution |
+|----------|--------------|----------------------|
+| 5 days past window, Grade A, loyal customer | -3 | Full refund to original payment |
+| 45 days past window, Grade B, average customer | +7 | Store credit for purchase price |
+| 90 days past window, Grade C, low-value customer | +16 | Deny, offer 15% discount on next purchase |
+| 10 days past window, Grade A, customer cited family emergency | -1 | Full refund to original payment |
+| Within window, Grade C, customer claims defect but inspection shows user damage | +10 | Store credit at 50% (goodwill), document for fraud scoring |
+| 60 days past window, brand-new customer, first order | +13 | Partial credit (50%), welcome them back with incentive |
+
+---
+
+## 5. Grading Standards by Product Category
+
+### 5.1 Universal Grading Criteria
+
+All categories share these baseline grade definitions. Category-specific addenda
+follow in §5.2.
+
+#### Grade A — Like New
+- Zero signs of use beyond initial unboxing
+- All original accessories, manuals, and packaging materials present
+- Original packaging in good condition (minor shipping wear acceptable)
+- Passes all applicable functional and safety tests
+- Can be restocked and sold as new without any additional processing beyond re-shelving
+
+#### Grade B — Good / Open Box
+- Minor cosmetic imperfections (light surface scratches, small scuffs) that do not affect function
+- Original packaging may be damaged, opened, or missing outer sleeve/shrink wrap
+- All essential accessories present (charger, main cable); non-essential items (stickers, pamphlets) may be missing
+- Fully functional — passes all applicable tests
+- Requires repackaging or "open box" labelling before resale
+
+#### Grade C — Fair
+- Visible cosmetic wear, scratches, dents, or staining that are noticeable at arm's length
+- Missing accessories that affect the completeness of the product (but not its core function)
+- Functional but may have minor performance degradation (battery at 60-80%, worn but operational buttons)
+- Not suitable for primary retail channel — routes to outlet, marketplace, or liquidation
+- May be viable for refurbishment if cost justifies it
+
+#### Grade D — Salvage / Parts
+- Non-functional, heavily damaged, or missing critical components that render the product unusable
+- Structural damage (cracked screens, bent frames, water damage indicators triggered)
+- May have value for parts harvesting or materials recovery
+- Routes to parts extraction, recycling, or destruction
+
+### 5.2 Category-Specific Grading Addenda
+
+**Consumer Electronics:**
+- Grade A additional requirement: battery health > 80% of design capacity (measurable via diagnostic)
+- Grade B threshold: battery 60-80%, cosmetic scratches visible only under direct light
+- Functional test required for all grades: power on, display, connectivity (WiFi/Bluetooth/cellular), speakers, cameras, ports
+- Data wipe verification mandatory before any resale disposition
+
+**Apparel:**
+- Grade A: tags attached (original retail tags, not just care labels)
+- Grade B: tags removed, but no wear indicators; passes UV and odour check
+- Grade C: visible wear, minor staining treatable with professional cleaning, or slight fabric stretching
+- Grade-reducing odours: tobacco smoke, pet odour, heavy fragrance, body odour
+- Automatic Grade D: mould, mildew, pest contamination
+
+**Footwear:**
+- Sole inspection is primary grading factor: unworn soles = Grade A, indoor-only wear marks = Grade B, outdoor wear = Grade C
+- Grade B requires: no toe box creasing deeper than 2mm, no heel counter collapse
+- Include insole inspection: customer orthotics must be removed, original insole must be present
+
+**Home Goods / Small Appliances:**
+- Grade A: unused, all packaging foam/wrapping in place
+- Functional test: operate through one full cycle (coffee maker: brew cycle, vacuum: run for 60 seconds, blender: blend ice test)
+- Missing filters, bags, or consumable accessories: Grade B (replaceable at $3-10 cost)
+- Cosmetic damage on surfaces visible during normal use: Grade C
+
+---
+
+## 6. Liquidation Channel Selection
+
+### 6.1 Channel Overview
+
+When product is routed to liquidation, selecting the right channel significantly affects
+recovery rates. The wrong channel can mean the difference between 20% recovery and 5%.
+
+| Channel | Best For | Typical Recovery (% of retail) | Fees | Min Lot Size | Speed to Cash |
+|---------|----------|-------------------------------|------|-------------|---------------|
+| B-Stock (owned auctions) | Electronics, home goods | 12-25% | 10-15% of sale | 1 pallet | 2-4 weeks |
+| Direct Liquidation | Mixed general merchandise | 8-18% | 15-20% of sale | 1 pallet | 2-6 weeks |
+| Bulq (owned by Optoro) | Small lots, mixed goods | 10-20% | Built into marketplace | 1 box (small lots) | 1-3 weeks |
+| Regional liquidators | Bulky/heavy items, furniture | 5-15% | Negotiated | Varies | 1-4 weeks |
+| Wholesale to dollar stores | Low-value general merchandise | 3-8% (often per-pound) | None (buy outright) | Truckload preferred | Immediate |
+| Online marketplace (eBay, Amazon Warehouse) | Individually valuable items ($50+) | 25-50% | 12-15% + shipping | Single unit | 1-8 weeks |
+| Charity donation | Items not worth liquidating, brand-sensitive | $0 (tax deduction at FMV) | None | No minimum | Immediate |
+
+### 6.2 Channel Selection Decision Tree
+
+```
+Is the individual unit value > $50?
+  ├── Yes → Is the item in Grade B or better condition?
+  │     ├── Yes → Sell individually on marketplace (eBay, Amazon Warehouse). Highest recovery.
+  │     └── No → Is the brand premium/recognisable?
+  │           ├── Yes → Auction on B-Stock (brand buyers pay premium). Recovery 15-25%.
+  │           └── No → Direct Liquidation or regional liquidator. Recovery 8-15%.
+  └── No → Is there a full pallet of same or similar category?
+        ├── Yes → Auction as category-sorted pallet on B-Stock or Direct Liquidation.
+        │     Recovery improves 30-50% vs mixed pallets.
+        └── No → Accumulate by category until pallet quantity reached.
+              If aging > 30 days, mix into general pallet and liquidate.
+              Holding cost exceeds sort premium beyond 30 days.
+```
+
+**Critical liquidation rules:**
+1. Never mix electronics with non-electronics on the same pallet. Electronics buyers won't bid on mixed pallets.
+2. Never include recalled products, counterfeit items, or hazmat in liquidation lots. Liability exposure is unlimited.
+3. Remove all customer personal data before liquidating electronics. Data breach from a liquidated device creates legal exposure.
+4. Photograph every pallet before shipping to liquidation. Disputes about condition are common.
+5. Manifest every pallet (list of SKUs, quantities, conditions). Manifested pallets sell for 20-40% more than unmanifested.
+
+---
+
+## 7. Refurbishment ROI Model
+
+### 7.1 When to Refurbish
+
+Refurbishment is only viable when the economics justify it. The decision model:
+
+```
+Refurbishment ROI = (Refurbished selling price - Refurbishment cost - Fulfilment cost) / Refurbishment cost
+
+Decision thresholds:
+  ROI > 100%: Always refurbish. High-value recovery.
+  ROI 50-100%: Refurbish if capacity exists. Good return on investment.
+  ROI 25-50%: Refurbish only if liquidation alternative is particularly poor (< 8% recovery).
+  ROI < 25%: Liquidate. The refurbishment effort isn't justified.
+```
+
+### 7.2 Refurbishment Cost Benchmarks by Category
+
+| Category | Common Defects | Typical Refurb Cost | Typical Refurb Selling Price | Typical ROI |
+|----------|---------------|--------------------|-----------------------------|-------------|
+| Smartphones (flagship) | Screen scratches, battery degradation | $40-80 (screen polish, battery replace) | $350-550 (65-75% of new) | 300-500% |
+| Laptops | Battery, cosmetic damage, slow storage | $50-120 (battery, SSD upgrade, clean) | $400-800 (55-70% of new) | 200-400% |
+| Tablets | Screen scratches, battery | $30-60 | $200-400 (60-70% of new) | 200-350% |
+| Small appliances | Cosmetic, missing parts | $10-25 (clean, replace accessory) | $30-60 (50-65% of new) | 100-200% |
+| Power tools | Battery, switch wear | $20-45 (battery, switch replacement) | $60-120 (55-65% of new) | 100-200% |
+| Headphones (premium) | Ear pad wear, cosmetic | $8-15 (new pads, clean) | $80-200 (60-75% of new) | 400-800% |
+| Game consoles | Cosmetic, controller wear | $15-30 (clean, replace controller pads) | $150-300 (60-70% of new) | 300-500% |
+
+### 7.3 Refurbishment Capacity Planning
+
+Refurbishment requires dedicated space, trained technicians, and parts inventory. The
+capacity model:
+
+- **Space:** 1 refurb station = approximately 80 sq ft (workbench + test equipment + parts storage)
+- **Throughput:** 1 technician handles 8-15 units/day for electronics, 20-30 units/day for small appliances
+- **Parts inventory:** Maintain 30-day supply of top 20 replacement parts by volume (batteries, screens, cables, ear pads, filters)
+- **Break-even:** A refurb station breaks even at approximately 5-8 units per day at average ROI of 150%. Below this volume, outsource to a third-party refurbisher.
+
+### 7.4 Outsource vs In-House Decision
+
+| Factor | In-House | Outsource |
+|--------|----------|-----------|
+| Volume > 40 units/day | Preferred (economies of scale) | Viable but more expensive |
+| Volume < 40 units/day | Only if margin justifies | Preferred (avoid fixed overhead) |
+| Brand certification programme exists | Required for "certified refurbished" branding | Must verify third-party is certified |
+| Product requires proprietary tools/software | In-house (IP control) | Only with NDA and audited facility |
+| Seasonal volume spikes | Core volume in-house, surge outsourced | Flexible capacity |
+| Data security requirements | In-house (direct control over data wipe) | Requires NIST 800-88 certification |
+
+---
+
+## 8. Return Processing Workflow by Channel
+
+### 8.1 E-Commerce Returns (Ship-Back)
+
+The standard e-commerce return flow. This is the highest-volume channel for most
+retailers and the one with the most automation opportunity.
+
+```
+Customer initiates return on website/app
+  │
+  ├── Automated policy check (within window? excluded category? customer in good standing?)
+  │     ├── Auto-approve → Generate RMA + prepaid return label
+  │     ├── Auto-deny → Display denial reason + alternatives
+  │     └── Manual review queue → Agent reviews within 4 hours
+  │
+  ├── Customer ships product
+  │     └── Tracking monitored for: label scan (confirms customer shipped), delivery to return centre
+  │
+  ├── Receiving at return centre
+  │     ├── Scan RMA barcode → pulls order record, expected product, customer profile
+  │     ├── Initial sort: sealed/unopened → express lane (15-second visual, Grade A, restock)
+  │     └── Opened/used → standard inspection lane
+  │
+  ├── Standard inspection (see §5 for grading criteria)
+  │     ├── Serial number verification (electronics only, but expanding to luxury goods)
+  │     ├── Visual inspection + functional test (category-dependent)
+  │     ├── Fraud scoring (post-receipt signals added to pre-receipt score)
+  │     └── Grade assignment: A / B / C / D
+  │
+  ├── Disposition routing (see §1 for category-specific trees)
+  │     ├── Grade A → Restock queue
+  │     ├── Grade B → Open-box / repackaging queue
+  │     ├── Grade C → Liquidation staging or refurbishment assessment
+  │     └── Grade D → Parts / recycling / destruction
+  │
+  └── Refund processing
+        ├── Refund triggered upon grade assignment (do not wait for disposition completion)
+        ├── Restocking fee applied if applicable (calculated at grading, not at refund)
+        └── Refund to original payment method → customer notification sent
+```
+
+**Key timing targets:**
+| Step | Target | Stretch Goal |
+|------|--------|-------------|
+| RMA generation (auto-approve) | < 5 minutes | Instant |
+| Return label delivery to customer | Immediate (email) | Immediate |
+| Customer ship-back | < 7 days from RMA | < 5 days |
+| Receiving scan at return centre | Day of delivery | Same as carrier delivery scan |
+| Inspection + grading | < 24 hours of receipt | < 4 hours |
+| Refund processing | < 24 hours of grading | Same day as grading |
+| Refund visible to customer | 3-5 business days | 1-2 business days |
+| Total RMA-to-refund cycle | < 14 days | < 7 days |
+
+### 8.2 Buy Online, Return In-Store (BORIS)
+
+Cross-channel returns are operationally more complex but have higher customer satisfaction
+and lower total cost (no return shipping). The critical risk is price discrepancy.
+
+```
+Customer arrives at store with online-purchased product
+  │
+  ├── Associate initiates BORIS return in POS
+  │     ├── Scan product barcode
+  │     ├── Look up online order (by order number, customer email, or loyalty account)
+  │     │     ├── Order found → POS displays actual purchase price from online order
+  │     │     │     └── CRITICAL: Refund at actual online purchase price, NOT store shelf price
+  │     │     └── Order not found → Customer provides order confirmation (email/app)
+  │     │           ├── Verified → Manual price entry at confirmed online price
+  │     │           └── Cannot verify → Process through online returns channel (mail-back)
+  │     │                 Do NOT guess the price. Do NOT use store shelf price.
+  │     └── Verify return eligibility (window, excluded categories)
+  │
+  ├── In-store inspection
+  │     ├── Visual + functional check (same criteria as return centre)
+  │     ├── For electronics: serial number check against online order
+  │     └── Grade assignment
+  │
+  ├── Refund processing
+  │     ├── Refund to original online payment method (not store credit, not cash)
+  │     │     Exception: customer paid with a gift card → store credit acceptable
+  │     ├── Restocking fee applied if applicable
+  │     └── Customer receives refund confirmation email
+  │
+  └── Inventory disposition
+        ├── If Grade A and product is in store assortment → Restock on store shelf
+        ├── If Grade A but product is online-only → Ship to return centre or nearest DC
+        ├── If Grade B/C → Ship to return centre for open-box/liquidation processing
+        └── If Grade D → Local disposal (do not ship non-functional product to return centre)
+```
+
+**BORIS-specific risks:**
+1. Price discrepancy (online vs store) → Mitigated by mandatory online order lookup
+2. Return of promotional/bundled items → Verify if the item was part of a BOGO or bundle; refund the proportional amount
+3. Store inventory adjustment → Ensure the store's inventory count correctly reflects the returned unit
+4. Different return windows → Online and store may have different windows; honour the more generous one
+
+### 8.3 In-Store Purchase, Return In-Store
+
+The simplest return flow. The POS has the transaction record, pricing is definitive,
+and the product doesn't need to be shipped.
+
+```
+Customer arrives with product + receipt (or POS lookup via card/loyalty)
+  │
+  ├── POS transaction lookup → Confirms purchase price, date, payment method
+  ├── Window check → Within return policy period?
+  ├── Inspection at return counter
+  │     ├── Quick visual for obvious damage/use
+  │     ├── Electronics: power-on test, serial number check
+  │     └── Grade assignment (usually Grade A or B at point of sale)
+  │
+  └── Refund to original payment method → Receipt printed → Customer exits
+```
+
+**Target transaction time:** < 5 minutes for standard returns. This is the benchmark
+that drives customer satisfaction — long return lines at the service desk are the
+#1 complaint in retail returns.
+
+### 8.4 Returnless Refunds (Customer Keeps Product)
+
+For items where the cost of return exceeds the recovery value. The decision model:
+
+```
+Return shipping cost estimate > 40% of product value?
+  ├── Yes → Evaluate returnless refund
+  │     ├── Product value < $50 → Auto-approve returnless refund
+  │     ├── Product value $50-100 → Supervisor auto-approve
+  │     ├── Product value $100-200 → Manager review (consider partial return — just the defective component)
+  │     └── Product value > $200 → Case-by-case (may justify return shipping for high-value)
+  │
+  └── No → Standard return process
+```
+
+**Returnless refund abuse prevention:**
+- Track returnless refunds per customer. More than 3 in 6 months triggers review.
+- High-value returnless refunds (> $100) are flagged for post-refund audit.
+- Products declared "defective" for returnless refund should still be counted in the
+  SKU defect rate, even though the physical product doesn't return.
+- Consider asking the customer to send a photo of the defect as lightweight verification.
+
+---
+
+## 9. Seasonal Return Planning
+
+### 9.1 Holiday Return Surge (January)
+
+January is to returns operations what December is to fulfillment. Plan for:
+
+| Metric | Normal Month | January Peak | Multiplier |
+|--------|-------------|-------------|-----------|
+| Return volume | 100% baseline | 250-350% | 2.5-3.5x |
+| Return processing backlog | < 24 hours | 48-96 hours | 2-4x |
+| Fraud attempts | Baseline | 180-220% | 1.8-2.2x |
+| Customer service contacts about returns | Baseline | 300-400% | 3-4x |
+| Gift receipt returns (% of total) | 5-8% | 25-35% | 4-5x |
+
+**Holiday return planning checklist:**
+1. Staff: Bring temporary inspection staff online by Dec 26. Target: 2x normal inspection capacity by Jan 2.
+2. Space: Reserve additional staging area for the return volume. Last-mile sorting should be simplified (Grade A express lane for sealed/tagged items).
+3. Policy: Extended holiday return window (typically Nov 1 - Dec 31 purchases returnable through Jan 31) means that returns trickle in over 4 weeks rather than concentrating in the first week. Model the curve.
+4. Gift receipts: Train all associates on gift receipt pricing rules. The #1 January return error is refunding at current price instead of purchase price.
+5. Fraud: Increase fraud scoring thresholds by 10% during January (reduce false positives — many legitimate gift returns trigger fraud signals).
+6. Liquidation: Pre-negotiate January liquidation capacity with liquidation partners. You'll have 3-4x normal Grade C/D volume, and everyone else will too.
+
+### 9.2 Category-Specific Seasonal Patterns
+
+| Category | Peak Return Period | Key Driver | Planning Action |
+|----------|-------------------|-----------|----------------|
+| Consumer electronics | Jan 2-15 | Holiday gifts, "not what they wanted" | Pre-stage functional test stations; serial number verification capacity |
+| Apparel | Jan 5-31 (extends longer) | Gift sizing, holiday party returns | UV/odour inspection throughput; wardrobing detection focus |
+| Fitness equipment | Jan 15 - Mar 1 | New Year's resolution abandonment | Large-item return logistics; returnless refund thresholds |
+| Outdoor/sporting goods | Mar-Apr (post-ski), Sep-Oct (post-summer) | Season-end returns | Seasonal markdown timing; storage vs liquidation decision |
+| School supplies | Sep 1-15 | Over-purchasing for school | High-volume, low-value; fast processing is key |
+
+### 9.3 Markdown-Driven Returns
+
+When products go on markdown or clearance, returns of the same product purchased at
+full price increase. Customers are not returning because of dissatisfaction — they're
+returning to repurchase at the lower price (return arbitrage).
+
+**Detection:** Monitor for returns where the same customer repurchases the same SKU within
+7 days of the return. If the repurchase price is lower, flag as potential price-match return.
+
+**Preferred handling:** Offer a price-match credit instead of processing a return and
+repurchase. The price-match credit costs the markdown difference; the return-and-repurchase
+costs the markdown difference plus processing cost plus potential disposition loss.
+
+---
+
+## 10. Data-Driven Return Reduction
+
+### 10.1 Root Cause Analysis by Return Reason
+
+Returns are a symptom. Reducing return rates requires treating the cause:
+
+| Return Reason | Root Cause Investigation | Typical Fix | Expected Reduction |
+|--------------|------------------------|------------|-------------------|
+| "Didn't fit" (apparel) | Poor size guidance, inconsistent sizing, inadequate photos | Size recommendation engine, fit model photos, detailed measurements | 15-25% reduction in "didn't fit" returns |
+| "Not as expected" | Product photo/description doesn't match reality | Lifestyle photos, video demos, customer photos in reviews, AR preview | 10-20% reduction |
+| "Defective" | Manufacturing quality, shipping damage, product design flaw | Vendor quality scorecard, packaging improvement, design feedback loop | Variable — depends on root cause |
+| "Changed my mind" | Impulse purchase, bracketing | Cooling-off period messaging, wish list instead of cart, fit technology | 5-10% reduction |
+| "Better price found" | Competitive pricing, price transparency | Price-match guarantee, automated price alerts | 8-15% reduction |
+| "Arrived too late" | Shipping delays, inaccurate ETAs | Improved delivery estimates, proactive delay notifications | 20-30% reduction in lateness returns |
+
+### 10.2 SKU-Level Return Rate Monitoring
+
+| Return Rate (rolling 90 days) | Action |
+|-------------------------------|--------|
+| < 5% | Normal. No action required. |
+| 5-10% | Review product listing for accuracy. Check reviews for recurring complaints. |
+| 10-15% | Flag for merchandise review. Audit listing photos/description. Check sizing data. |
+| 15-25% | Escalate to category manager. Consider adding warnings to listing. Review vendor quality. |
+| > 25% | Stop-sell review. The product may have a systemic issue that no listing fix can solve. |
+
+### 10.3 Return Cost Allocation Model
+
+Allocating return costs to the business units that drive them creates accountability:
+
+| Return Cause | Cost Allocation |
+|-------------|----------------|
+| Defective product | Vendor (via RTV/defect claim) or Merchandise (if vendor approved) |
+| Wrong item shipped | Fulfillment operations |
+| Damaged in shipping | Carrier (via shipping claim) or Packaging engineering |
+| Poor product description | E-commerce / Content team |
+| Sizing issue | Merchandise / Product development |
+| Customer changed mind | Cost of doing business (absorbed by margin model) |
+| Fraud | Loss prevention budget |