We've migrated more than 30 SAP ECC tenants to Oracle Fusion Cloud. The pattern is repeatable enough that we've codified it into a 12-step playbook. The secret isn't a magic tool — it's knowing in advance which 4 mappings will burn 60% of the project, and having pre-built crosswalks ready for them.
The 12 steps, by stage
We split the work into four stages — Discover, Map, Load, Cutover — and 12 steps within them. Each step has a deliverable; nothing moves to the next without it.
Discover (steps 1–3)
- Step 1 — Inventory: every SAP module in scope (FI, CO, MM, SD, HR, PP, etc.) with row counts.
- Step 2 — Profile: data quality scorecards for the top 30 tables.
- Step 3 — Sign-off: customer's data owner approves the in-scope inventory.
Map (steps 4–7) — and the 4 mappings that burn the project
These four mappings consistently consume 60% of the mapping effort. Front-load them.
- Step 4 — Company Code → Business Unit + Legal Entity. SAP's Company Code is a single concept; Oracle splits it across Business Unit, Legal Entity, and Ledger. The split is rarely 1:1.
- Step 5 — GL Natural Account → Oracle COA. SAP GL accounts are flat; Oracle wants a Chart of Accounts hierarchy. Don't assume a 1:1 GL mapping — use this opportunity to clean up the COA.
- Step 6 — Vendor Account Group → Oracle Supplier Classification. SAP's vendor account groups have to be re-encoded into Oracle's supplier classification + business relationship type.
- Step 7 — Storage Location / Plant → Oracle Inventory Org + Subinventory. Oracle's inventory hierarchy is more granular; expect to split many SAP locations.
Load (steps 8–10)
- Step 8 — Master data load: suppliers, customers, items, COA, employees. This must complete and reconcile before transactions.
- Step 9 — Open transactions: open POs, open SOs, open invoices, open AR.
- Step 10 — Historical balances: GL trial balance, AP/AR aging snapshots.
Cutover (steps 11–12)
- Step 11 — Rehearse the cutover three times. Each rehearsal must produce the same control-total report.
- Step 12 — Cut over for real. Run the same scripts. Compare to the rehearsal.
What to do differently when SAP is the source
Two SAP-specific things to plan for:
- SAP date fields are stored as YYYYMMDD strings in many tables. Oracle Fusion expects ISO dates. Build the conversion into Stage 4.
- SAP customer/vendor numbers are zero-padded to 10 digits in some tables, unpadded in others. Standardise during cleanse, not during load.
The Syntra ETL team has run more than 100 enterprise data migrations to Oracle Fusion Cloud. We share what we learn so your migration is faster, cleaner, and more boring than ours were.