Inventory Modeling Guide

What is Stock Segmentation?

• Customer-facing product differences, where buyers explicitly choose between variants (for example, size or color)
• Business context, such as sales channel or target market, where the same stock must be exposed differently
• Future inventory context, where stock is expected to become available at a later point in time and must be tracked separately
• Physical characteristics of inventory, such as condition (e.g., perfect / damaged stock), manufacturing batch, or expiry date
• Supply chain characteristics, such as supplier or country of origin, which require stock to be managed and allocated differently

Mental Model

Segmentation Patterns and Recommended Practices

Pattern 1: Product Variant Segmentation

• The customer explicitly chooses between alternatives (size, color, configuration)
• Inventory must be tracked separately per option (position)
• Availability does not depend on business context

• Apparel:
  T-shirt in Small / Medium / Large with different inventory per size
• Electronics:
  Laptop in Silver / Space Gray where customers select color
• Furniture:
  Sofa in Leather/Fabric where material choice affects availability

• Simple and explicit
• Fully supported across all workflows
• No additional aggregation complexity

• Does not support dynamic or contextual segmentation
• Not suitable for regulatory or operational differences

• The same SKU must be allocated differently by sales channel or region
• Segmentation depends on non-customer-facing attributes

Pattern 2: Inventory Catalog Isolation

• Inventory data comes from fundamentally different sources
• Ingestion, ownership, or SLAs differ
• Reconciliation between sources is undesirable

• Multi-region retailer:
  Separate catalogs for North America vs. EMEA due to different suppliers and fulfillment SLAs
• Third-party logistics:
  Client's owned inventory vs. 3PL-managed consignment stock with separate ingestion pipelines
• Franchise model:
  Corporate-owned inventory vs. franchisee-owned inventory with different ownership and reconciliation rules

• Strong isolation of data pipelines
• Clear operational boundaries

• Increased integration complexity
• Limited flexibility once catalogs diverge

• Segmentation is purely business-driven
• Inventory must be shared across contexts

Pattern 3: Segmented Inventory Quantities

• Segmentation reflects physical or temporal properties of stock
• Multiple segments must coexist within the same SKU and location
• Segmentation participates in allocation, sourcing or recall logic

• Physical Segmentation:
  • Perfect stock vs. damaged stock sold at discount
  • Pharmaceutical:
    Same medication with different expiry dates requiring separate tracking
• Supply Chain Segmentation:
  • Multi-supplier fashion:
    Same dress from Supplier A (fast delivery) vs. Supplier B (bulk order)
  • Electronics:
    Components from USA vs. China requiring country-of-origin tracking for tariffs
  • Luxury goods:
    Different manufacturing batches for product recall capability
• Future Inventory Segmentation:
  • Pre-orders:
    Reserved inventory from upcoming shipment (PO001, arriving March 1)
  • Seasonal:
    Summer collection inventory in-transit vs. current spring inventory on-hand

• Enables segmentation without multiplying catalogs or variants
• Preserves physical and lifecycle differences at the inventory level
• Allows multiple segments to coexist within the same Inventory Position
• Aligns segmentation with sourcing and allocation decisions

• Requires a more expressive inventory data model
• Increases complexity of aggregation and sourcing logic
• Requires custom extensions to define the segmentation logic

• Segmentation is purely customer-facing
• Differences only affect availability presentation, not physical stock
• A simpler (e.g., Virtual Catalog-based) approach is sufficient

Pattern 4: Virtual Catalog Segmentation

• The same physical stock should present different availability views
• Availability differs by sales channel (for example, B2C vs B2B)
• Business rules affect ATS calculation, not physical stock

• Marketplace seller:
  Same warehouse inventory shown differently on .com vs. marketplaces with different buffer rules
• Member tiers:
  Same inventory with different availability for Gold members (see full stock) vs. Silver members (limited view)
• Sales channel inventory allocation:
  Physical stores see 90% of stock, online sees 80% (10% held as safety stock)
• Regional websites:
  US website shows availability after US shipping buffers; UK website applies UK shipping buffers to same stock

• Strong isolation between business contexts
• Clear ownership of rules per channel
• Mature and widely adopted pattern

• Each additional Virtual Catalog increases operational overhead
• Potential data duplication
• Does not scale well for high-cardinality segmentation (for example, per individual country)

• Segmentation attributes are numerous or frequently changing
• Segmentation reflects physical stock reality

Pattern 5: Virtual Segments

• Segmentation is required at the availability layer
• Segments are dynamic
• Creating one Virtual Catalog per segment is not viable

• Time-Based Availability:
  • Wave 1 chairs available February 1 (200 units), Wave 2 available March 15 (300 units)
  • Fashion pre-orders:
    Spring collection shows "50 available now, 200 more arriving April 1"
  • Electronics launch:
    iPhone pre-orders against future shipments with specific delivery dates
• Channel Segmentation (High Cardinality):
  • Geographic markets:
    Same product with availability segmented across 50 country-specific sites
  • Membership levels:
    Bronze/Silver/Gold/Platinum customers each with different allocations
  • B2B wholesale:
    10 different partner tiers each seeing different availability
• Dynamic Segments:
  • Promotional campaign:
    "First 100 customers" segment that depletes as orders come in
  • Loyalty program:
    "VIP early access" segment before general availability
  • Bundle availability:
    Product available standalone vs. as part of bundle (different quantities)

• Reduces the need for Virtual Catalog proliferation
• Supports high-cardinality or dynamic segmentation
• Complements inventory-level segmentation

• Requires careful design of aggregation and querying

• Must align with sourcing and allocation expectations
• Requires custom extensions

• segmentation reflects physical stock differences
• a small number of static Virtual Catalogs is sufficient

Common Anti-Patterns

• using virtual or pseudo-locations to represent stock types, such as condition, manufacturing batch, supplier, or channel allocation
• duplicating inventory catalogs to separate stock that belongs to the same product and physical location
• forcing physical or lifecycle stock differences into product variants, when those differences are not customer-facing
• modeling physical stock differences only at the availability layer, without representing them in the inventory model
• introducing complex custom allocation or sourcing logic to compensate for missing segmentation in the data model

How to Choose the Right Approach

• Is the difference explicitly chosen by the customer? → Product Variant Segmentation
• Does the difference represent a physical or lifecycle property of stock? → Segmented Inventory Quantities or Inventory Catalog Isolation
• Does the difference affect how availability is presented, rather than the stock itself? → Virtual Catalog Segmentation or Virtual Segments
• Is the segmentation high in cardinality or expected to change frequently? → Virtual Segments