The Banking SAP Migration Checklist: 9 Things Regulated Institutions Must Settle Before ECC Support Ends in 2027

1. Confirm your exact enhancement pack version — it decides your options

Extended maintenance eligibility depends on being onEHP 6 or above. Institutions still on EHP 5 or lower do not have the 2028–2030 runway as a fallback; the 2027 date is a hard wall. Pull the actual EHP level from your system, not from the last architecture deck. This single fact determines whether you have a 17-month runway or a 41-month one.

2. Name the regulatory reporting owner before you name the SI

Regulatory reporting is the workstream that breaks migrations in financial services, and it breaks for a structural reason: the SI is measured on go-live, the regulator is not. Those two clocks are in direct conflict, and unless someone inside the institution owns the mapping of every statutory and supervisory report to its source objects in the target system — with veto power on scope — the conflict resolves in favour of the go-live date every time. Name that person before you sign the SI contract, not after. If the answer is "the SI will figure it out," the answer is wrong.

3. Settle historical data retention before choosing a migration approach

Retention obligations in banking commonly run 7–10 years, and in some jurisdictions longer. You have three viable patterns: carry history into S/4HANA, retain a read-only legacy instance, or move history to a compliant archive with a documented retrieval SLA. Each has a different cost curve and a different examiner conversation. Pick one, price it, and document why.

4. Prove audit-trail continuity across cutover

The hardest question internal audit will ask is: "can you reconstruct a transaction that started in ECC and settled in S/4HANA?" Design the cutover so change logs, document numbers and approval records are traceable end-to-end across the boundary. This is a design decision, not a testing task — retrofitting it is expensive.

5. Inventory every interface, including the ones nobody owns

A typical ECC landscape in a bank exchanges data with core banking, payment rails and SWIFT, EDI partners, tax engines, treasury systems, CRM, regulatory reporting engines, and dozens of custom point-to-point interfaces built over two decades. Interface count — not module count — is the best early predictor of migration duration. Count them before you estimate.

6. Take your clean-core posture seriously now, not in phase 3

SAP's extensibility model now runs a four-level A–D maturity framework, giving you a structured path off modifications and toward ABAP Cloud and side-by-side BTP extensions. The reason this matters on day one rather than day 400:every SAP AI capability — Joule, embedded analytics, autonomous agents — is designed against a clean, standardized core. A migration that carries your modifications forward technically succeeds and strategically fails.

7. Baseline your close cycle before you touch anything — you get one chance

This is the item institutions skip and cannot recover. The moment cutover happens, your pre-migration baseline is gone: you can no longer measure it, only estimate it, and an estimate will not survive a CFO's scrutiny when you claim the benefit eighteen months later. Record four numbers now — days-to-close, manual journal entries per close, reconciliation exception rate, and restatement history. It takes a week. Skipping it means the entire financial case for the program becomes unfalsifiable, which is another way of saying unprovable.

8. Match the approach to your risk tolerance, honestly

Brownfield preserves history and process, greenfield buys a clean core at the cost of a re-implementation, and selective/bluefield splits the difference. In heavily regulated environments, risk tolerance — not ambition — usually decides, and phased or selective transitions dominate for a reason: they keep the evidence chain intact and the blast radius small.

The uncomfortable version of this conversation: brownfield is often chosen because it is the path of least organisational resistance, not because it is the right answer. It carries your modifications forward, which means you clear the 2027 deadline and inherit the clean-core problem intact — you have bought time, not capability. That is a defensible decision if you make it deliberately and budget for the remediation. It is an expensive accident if you drift into it.

9. Put internal audit and the regulator on the plan, early

Build audit checkpoints into the plan as gates, not as a post-go-live review. The institutions that struggle are the ones where internal audit first sees the migration design six weeks before cutover.