Business systems
Enterprise resource planning integration: who reconciles data that does not match?
A governance guide for finance, inventory, e-commerce and warehouse integrations.
Moving records is only the transport layer of an integration. Completion means the business can detect, explain and resolve missing, duplicated, delayed and rejected transactions. Each interface needs a business owner, technical owner, reconciliation rule, exception queue, recovery method and acceptance evidence.
From day-to-day administration
Details worth settling early
Name the business events
In a chain such as Shopify, Celigo, NetSuite and a warehouse system, an order, fulfilment, shipment, refund and stock adjustment are different events. Each needs an authoritative system, a matching key and someone responsible for exceptions.
Connector success covers transport
A successful flow shows that a record moved. Financial and stock control still needs totals, exception ageing and a way to prevent a corrected transaction from being replayed twice.
Migration changes the control model
Moving from NetSuite to a combination such as Cin7 and Xero changes where product, inventory and accounting authority sit. Reusing the old interface map without revisiting those responsibilities carries the old ambiguity into the new systems.
Design choices that matter
- Define the system of record for each business object and status.
- Separate transport success from business correctness.
- Design reprocessing and duplicate protection before cutover.
How to approach the work
- 01
Map business events
Document order, customer, item, inventory, fulfilment, invoice, payment, refund and adjustment flows in business language.
- 02
Assign authority
State which system owns each identifier, value and status and which owner approves exceptions.
- 03
Design control totals
Use counts, amounts, timestamps, sequence, status and sampled detail to prove completeness and accuracy.
- 04
Test failures deliberately
Include timeouts, invalid master data, duplicates, partial fulfilment, back-dated changes and unavailable destinations.
- 05
Operate the exception path
Give named staff a queue, runbook, escalation threshold, audit trail and reporting cadence.
- 06
Reconcile after change
Repeat end-to-end control totals after releases, mapping changes and major master-data updates instead of relying on the original cutover result.
What to test before handover
- System of record defined
- Owners named
- Control totals approved
- Failure cases tested
- Duplicate protection verified
- Replay runbook tested
- Cutover reconciled
Problems to catch early
- Relying on a green connector status as proof of correct accounts or stock
- Allowing both systems to own the same field
- Manually fixing data without recording cause and replay risk
Useful technical references
This is a general operational guide. Check applicable vendor documentation and professional advice before acting on a material decision.
Links and technical details checked 2 September 2026. Corrections can be sent to info@turnstoneai.com.