Home Insights Preparing for an AI-Ready S/4HANA Migration
Platform AI ROI
Preparing for an AI-ready S/4HANA migration
Last reviewed: 29 September 2026
Key takeaways
- A like-for-like migration carries old workarounds into a new system and wastes the chance to design AI in.
- Map the manual work around your current SAP system before you design the target.
- Keep the core clean and put AI in the processes around it, where most manual effort sits.
An AI-ready migration is an ERP migration planned so that the new system’s processes, data and integrations support AI from day one, rather than adding AI afterwards.
For many SAP customers, the end of mainstream maintenance for SAP ECC at the end of 2027 is forcing decisions about S/4HANA. Optional extended maintenance runs to 2030 at additional cost, but most organisations are now planning their move.
The temptation is to treat the migration as a technical upgrade. That is the most expensive way to do it.
Why like-for-like is a missed opportunity
Over years, organisations build a layer of manual work around their ERP: spreadsheets for planning, email for orders and approvals, custom reports, re-keying between systems. A like-for-like migration moves the ERP and leaves that layer intact — or rebuilds it as custom code. Meanwhile, the new platform ships AI and automation capabilities that could remove much of it.
Step 1: Map the work around SAP
Before designing the target system, map where people spend time around your current SAP processes: what arrives by email, what is re-keyed, which spreadsheets exist and why, which reports are pulled manually, where approvals wait. This "shadow process" map is the most valuable input to an AI-ready design.
Step 2: Decide what AI should replace
For each area of manual work, decide whether it should be:
eliminated
by standard S/4HANA functionality;
automated
by embedded AI or automation on the SAP platform;
handled by AI outside the core
for example, extracting orders or invoices from documents before they enter SAP;
kept
, because the value of changing it is low.
Step 3: Keep the core clean
A clean core — standard processes, minimal custom code inside the ERP — makes upgrades easier and lets you adopt new vendor AI features as they arrive. Put variation and AI-driven steps in extension and integration layers, connected through standard interfaces.
Step 4: Fix the data you will need
AI depends on master data: customers, materials, suppliers, prices. A migration forces data cleansing anyway; prioritise the fields your planned AI use cases rely on, and set up governance so data stays clean after go-live.
Step 5: Build governance into the design
Define from the start which automated and AI-driven steps may post transactions, under which technical users, within which limits, and how they are logged. Retrofitting controls after go-live is far harder.
Step 6: Plan adoption, not just cutover
People need to learn new processes and trust new automation. Build role-based training and adoption tracking into the migration plan, and measure whether the manual work you mapped in step 1 actually disappears.
Quick wins before migration
You don’t have to wait. Many AI use cases — order intake, invoice matching, collections prioritisation — work with ECC today through standard interfaces and move across when you migrate. Delivering one or two before migration builds capability and shows which workarounds to drop.
Find the value already inside your platforms before you migrate.
Frequently asked questions
Will AI features in S/4HANA replace the need for other AI tools?
They will cover many tasks inside SAP. Work that crosses SAP and other systems — email, documents, CRM — usually still needs integration and AI outside the core.
Does an AI-ready migration take longer?
The design phase takes a little longer; the overall programme is often shorter because less custom code is rebuilt.
Where should we start?
With the shadow-process map. It takes a few weeks and shapes every other decision.