Order-to-Cash

The problem

Sales activates an order in the CRM, then someone rekeys it into the ERP for fulfillment and invoicing. Errors creep in, sales cannot see billing status without asking finance, and month-end close means reconciling two systems line by line.

How it runs on Neblex

A Salesforce change starts the flow when an order is activated, through a Change Data Capture stream or the polled order trigger. An upsert by external ID finds the existing NetSuite customer or creates one, and field mapping with type-aware AI suggestions aligns items, quantities, pricing, and terms with the NetSuite sales order schema.

The flow creates the sales order and downstream billing records as a coordinated sequence with per-step retries. If a later step fails, saga-style compensation unwinds partial writes, and the run can be replayed from the exact step that broke, so the two systems do not drift into a half-synced state.

Poll-based change detection watches NetSuite for fulfillment, invoice, and payment status, and writes each change back to the Salesforce order. High order volume runs through chunked batch processing with checkpoints, and promotion approvals gate every flow change from dev to prod.

Step by step

1. 1
   Capture the orderA Change Data Capture stream or the polled order trigger fires when the order is activated.
2. 2
   Match the customerDedup rules resolve the NetSuite customer so no duplicate records are created.
3. 3
   Create the sales orderField mapping with type-aware AI suggestions builds the NetSuite order from the CRM data.
4. 4
   Guard against partial writesPer-step retries and saga-style compensation keep failed runs from leaving orphaned records.
5. 5
   Track fulfillment and billingPoll-based change detection picks up fulfillment, invoice, and payment status in NetSuite.
6. 6
   Write status backThe Salesforce order reflects ERP status so sales and finance read the same record.

Platform capabilities used

• Webhook triggers
• Dedup with lookups and upserts
• Field mapping with AI suggestions
• Saga-style compensation
• Poll-based change detection
• Chunked batch processing with checkpoints

Common questions