The three routes, plainly stated
A brownfield conversion carries the existing system forward — configuration, history and custom code — and converts it in place. It is the shortest route to being on S/4HANA and the surest way to inherit two decades of accumulated process debt.
A greenfield build starts from standard processes and migrates only the data that matters. It gives the cleanest end state and demands the most from the business, because every deviation from standard has to be argued for rather than assumed.
Selective data transition sits between them: a new system built deliberately, into which chosen company codes, history and configuration are moved. It buys flexibility at the price of tooling complexity and a longer design phase.
- Brownfield: fastest path, inherits customization and process debt
- Greenfield: cleanest core, heaviest change-management load
- Selective transition: most control, most complex to execute
Custom code is the honest inventory
Before choosing, run the code inventory and be ruthless about it. Typically a large share of custom objects has not executed in years, another share duplicates functionality now standard in S/4HANA, and only a minority encodes something genuinely distinctive about how the company operates.
That third category is the argument for clean core discipline: keep it, but move it out of the core onto SAP BTP or an integration layer where it can be upgraded independently. Extensions inside the core are how organizations end up unable to take releases.
Integration and data quality set the timeline
The ERP is the smaller half of the program. Interfaces to CRM, MES, warehouse, banking, tax and reporting systems, plus the master data those interfaces depend on, usually determine when the cutover can actually happen.
Start data remediation early and treat it as a business workstream with named owners for material, vendor, customer and finance master data. Cleansing during the cutover window is the most common cause of a delayed go-live.
Sequence for provable increments
Long ERP programs lose sponsorship when nothing is visible for a year. Structure the work so that each phase ends in something demonstrable: a validated design against real transactions, a converted sandbox running month-end, a pilot company code live, then the wider rollout.
Rehearsed cutovers matter more than plans for them. Several full dress rehearsals with real data volumes, timed to the hour, are what turn a cutover weekend into an operational exercise rather than a gamble.