Cross Channel Parity

Daily Scan Process

The cross-channel parity check runs as a daily cron job. The scan follows this sequence:

Step 1: Build Product Map

For each active product in Shopify:
    Query Supabase products table for:
        - shopify_product_id
        - etsy_listing_id (if listed on Etsy)
        - amazon_asin (if listed on Amazon)
        - current Shopify price
        - current inventory (from inventory_extended)
    Build cross-reference map: {SKU -> {shopify, etsy, amazon}}

Step 2: Check Each Parity Dimension

For each product with listings on 2+ channels, check:

Price Parity

Expected Etsy price = Shopify price * etsy_markup_factor (from channel-config)
Expected Amazon price = Shopify price * amazon_markup_factor (from channel-config)

IF actual_etsy_price differs from expected by >5%:
    FLAG as price_drift
    severity = "warning" if <10%, "critical" if >=10%

IF actual_amazon_price differs from expected by >5%:
    FLAG as price_drift
    severity = "warning" if <10%, "critical" if >=10%

Inventory Parity

available_to_sell = total_on_hand - fba_allocated - in_transit_reserved - safety_stock

IF etsy_listed_quantity > available_to_sell:
    FLAG as oversell_risk, severity = "critical"
IF amazon_listed_quantity > (available_to_sell - etsy_reserved):
    FLAG as oversell_risk, severity = "critical"
IF any channel shows "in stock" but available_to_sell = 0:
    FLAG as phantom_stock, severity = "critical"
IF any channel shows "out of stock" but available_to_sell > 0:
    FLAG as missed_sales, severity = "warning"

Description Parity

For each product, extract core facts from Shopify description:
    - Material
    - Dimensions/size
    - Origin/sourcing
    - What's included (mallet, cushion, etc.)

Compare against Etsy and Amazon descriptions:
IF core fact is missing from a channel listing:
    FLAG as content_drift, severity = "warning"
IF core fact contradicts between channels (e.g., different dimensions):
    FLAG as content_conflict, severity = "critical"

Image Parity

IF Shopify primary image updated more recently than Etsy/Amazon:
    FLAG as image_stale, severity = "low"
IF Etsy listing has fewer than 5 images:
    FLAG as insufficient_photos, severity = "warning"

Decision Tree: Auto-Fix vs. Escalate

1. Is the drift in INVENTORY?
   → Inventory sync failures are time-sensitive (overselling risk)
   → Phase 1: FLAG as critical, draft correction, queue for human approval
   → Phase 2 (graduated): Auto-fix inventory sync within 15 minutes
   → ALWAYS log the correction in skill_invocations

2. Is the drift in PRICING?
   → Price drift <5%: Log only, include in daily report
   → Price drift 5-10%: WARNING — draft correction, queue for ceo approval
   → Price drift >10%: CRITICAL — draft correction, queue for ceo approval with alert
   → NEVER auto-correct pricing, even in Phase 2

3. Is the drift in DESCRIPTIONS?
   → Missing core fact: Draft updated description, queue for approval
   → Contradicting facts: CRITICAL — flag for immediate review
   → Style/formatting differences: ACCEPTABLE (channels have different content needs)
   → NEVER auto-correct descriptions in Phase 1

4. Is the drift in IMAGES?
   → Stale images: Low priority — include in weekly report
   → Missing images: Warning — flag for content team

Output Format

Daily Parity Report

{
  "report_date": "YYYY-MM-DD",
  "scan_summary": {
    "products_scanned": 0,
    "products_with_multi_channel": 0,
    "products_in_parity": 0,
    "products_with_drift": 0
  },
  "drift_items": [
    {
      "sku": "TS-BOWL-HH-7IN",
      "product_name": "Hand-Hammered Singing Bowl 7 inch",
      "channels_affected": ["shopify", "etsy"],
      "drift_type": "price_drift | inventory_drift | content_drift | image_stale",
      "severity": "critical | warning | low",
      "details": {
        "dimension": "price",
        "shopify_value": "149.00",
        "etsy_value": "149.00",
        "expected_etsy_value": "166.88",
        "drift_pct": 12.0
      },
      "recommended_action": "Update Etsy price to $166.88 (12% markup per channel-config)",
      "auto_fixable": false,
      "requires_approval_from": "ceo"
    }
  ],
  "corrections_applied": [],
  "corrections_pending_approval": []
}

Correction Request (queued for approval)

{
  "correction_id": "parity-YYYYMMDD-001",
  "sku": "TS-BOWL-HH-7IN",
  "channel": "etsy",
  "field": "price",
  "current_value": "149.00",
  "proposed_value": "166.88",
  "rationale": "Etsy price should be Shopify base ($149) * 1.12 markup = $166.88 per channel pricing rules",
  "severity": "critical",
  "requires_approval": true,
  "approval_target": "ceo"
}

Phase 1 Behavior

All corrections require human approval before execution:

• Inventory corrections: Queue for operations-manager's review (operational). If oversell risk is imminent, also alert ceo.
• Price corrections: Queue for ceo's review via dashboard approval queue.
• Description corrections: Queue for ceo's review. Show current vs. proposed text.
• Image updates: Queue for ceo's review. Low priority — included in weekly report, not daily alerts.

Escalation thresholds (trigger immediate alerts, not just daily report):

• Any oversell risk (inventory on channel > available stock)
• Price drift > 10% on any active listing
• Contradicting product facts between channels

Phase 2 Graduation Criteria

After demonstrating reliable performance (target: 90+ days, 200+ corrections, <2% error rate):

• Auto-fix: Inventory sync corrections (within 15 minutes of detection)
• Remains human-approved: All pricing changes, all description changes, all image changes
• Auto-demote to Phase 1: If any auto-correction is reverted by a human, or if error rate exceeds 2% in a 30-day window

Model Routing

• Daily scan and data comparison: Haiku 4.5 (structured data comparison, pattern matching)
• Description diff analysis: Sonnet 4.6 (needs judgment about whether differences are acceptable channel optimization vs. actual drift)
• Correction drafting and report generation: Sonnet 4.6 (needs to write clear rationale and proposed corrections)

Dependencies

• References: agents/shared/channel-config/SKILL.md (fee structures, markup rules, SKU conventions), agents/shared/product-knowledge/SKILL.md (product facts for validation), agents/shared/escalation-matrix/SKILL.md (approval routing)
• Data sources: Supabase products table, inventory_extended table, Shopify API, Etsy API, Amazon SP-API
• Referenced by: agents/ecommerce/skills/etsy-content-optimization/SKILL.md, operations agents (inventory sync)