Integration Patterns: ERP, CRM, PIM, OMS, and Data Layer

Integration architecture is where commerce platform evaluations often fail. Buyers focus on feature lists and neglect to ask: "How does this actually connect to our ERP?"

ERP as the Operational Nucleus

Your ERP remains the system of record for:

• Customer master data and account hierarchies

• Contract terms, payment terms

• Price lists and discount schedules

• Product master data (for some companies)

• Order fulfillment and invoicing

B2B ERP integration is about orchestrating the entire transaction lifecycle, not just pushing orders. Commerce receives orders but syncs them to ERP for fulfillment. ERP generates invoices and updates payment status. Commerce reflects order status but ERP owns the authoritative record.

Integration patterns fall into two categories:

1. Master data sync (customers, products, prices) — usually batch, scheduled (nightly or hourly)

2. Transactional events (orders, shipments, returns) — near real-time via APIs or message queues

Propeller provides ERP connectors for SAP, Microsoft Dynamics, and other major systems. Our connectors handle the dual-sync pattern: batch master data sync (nightly product and pricing updates) plus event-driven order sync (order created in Commerce Hub triggers immediate POST to ERP).When orders flow from Commerce Hub to ERP, they include all necessary context: customer ERP account ID, contract reference, payment terms, requested delivery date. The ERP responds with its internal order number, and Commerce Hub stores the mapping. When customers track orders in the portal, they're seeing ERP data synchronized via webhook.

Direct Integrations vs Middleware/iPaaS

Point-to-point custom scripts between commerce, ERP, PIM, and 3PLs don't scale. You end up with brittle integration code scattered across systems, no error handling, no monitoring, and no documentation.

Case studies show companies consolidating integrations using iPaaS tools like Celigo, Boomi, or MuleSoft. A mid-market eyewear brand moved from direct Shopify–NetSuite scripts to Celigo prebuilt flows, eliminating system crashes and manual data entry.

For mid-market B2B, the right pattern is usually:

• Commerce platform exposes clean APIs and webhooks

• iPaaS tool orchestrates data flows between systems

• Prebuilt connectors reduce custom integration code

• Error handling and retry logic built into the middleware layer

• Monitoring dashboards show integration health and failures

Propeller's Commerce Hub is designed for this pattern. We emit webhooks for all major business events:

{
  "event": "order.created",
  "timestamp": "2026-03-01T10:12:45Z",
  "order": {
    "id": "ORD-102984",
    "externalErpId": null,
    "customer": {
      "id": "CUST-204",
      "erpAccountId": "ACCT-987"
    },
    "lines": [
      {
        "sku": "SPARE-12345",
        "quantity": 4,
        "uom": "PCS",
        "netPrice": 149.0,
        "currency": "EUR"
      }
    ],
    "totalNet": 596.0,
    "currency": "EUR",
    "paymentTerms": "NET30",
    "requestedDeliveryDate": "2026-03-15"
  },
  "signature": "HMAC-SHA256..."
}

This webhook payload includes everything your iPaaS needs to create the order in your ERP: customer ERP ID, line-item details, pricing, payment terms, delivery expectations. The HMAC signature ensures webhook authenticity.

We also provide REST endpoints for your iPaaS to push updates back: shipment tracking, invoice numbers, inventory changes. The integration is bidirectional and event-driven, not just one-way batch files.

PIM and Product Data Flows

For companies with large, complex catalogs — especially manufacturers with technical specs, compliance data, and regional variations — PIM systems centralize product information and syndicate to commerce and sales channels.

PIM owns:

• Product attributes (technical specs, dimensions, certifications)

• Digital assets (images, videos, 3D models, PDFs)

• Translations and localized descriptions

• Regulatory and compliance data

• Product relationships (accessories, spare parts, compatible models)

Commerce systems consume PIM data but shouldn't replicate full PIM functionality. The integration pattern:

1. PIM syndicates product data to commerce (nightly or on-demand)

2. Commerce stores the data optimized for customer-facing queries

3. Commerce reflects product updates within hours

4. PIM remains source of truth for product content

   Data Contracts, Events, and Idempotency

   Clean integration architecture requires:

   • Data contracts define exactly what fields are required, types, validation rules

   • Event-driven patterns use webhooks for near real-time updates and scheduled syncs for bulk master data

   • Idempotent APIs handle retries safely (calling "create order" twice with the same ID creates one order, not two)

   • Clear error semantics distinguish between retriable errors (ERP temporarily down) and permanent failures (invalid product SKU)

   Integration best practices emphasize using APIs plus webhooks for real-time events (orders, inventory changes) and scheduled syncs for bulk master data (catalog updates, price lists).

   Propeller's Commerce Hub emits webhooks for major business events: RFQ received, quote issued, quote approved, order created, shipment confirmed. Each webhook includes a signature for verification and supports retries with exponential backoff. On the receiving side, our APIs are idempotent — if your iPaaS posts the same order twice due to a network glitch, we detect the duplicate and return the existing order rather than creating a second one.

   Our AI Sales Hub consumes ERP-derived pricing and contract data using well-defined GraphQL queries, not direct database connections or file drops. This means your ERP can change its internal schema without breaking our integration, as long as the API contract stays stable.