Customization and usage inventory
Every customization is a migration cost. Usage data shows which developments actually run; the unused ones do not travel. This is the step that shrinks scope most.
- Usage data covers at least the past year
- Unused developments are archived rather than migrated
- Customizations with a standard equivalent are flagged
- Developments with no owner are listed separately
Data quality and conversion
Master data quality is the most underestimated line in a migration. Duplicate supplier and material records carried into a new system continue growing there; cleanup belongs before the move.
- Master data cleanup completes before migration
- Conversion rules are proven with trial loads
- The scope of historical data is decided explicitly
- A reconciliation report follows every load
Cutover window and rollback
The most critical part of the plan is the window in which the business stops. It is reconciled with the business calendar, its duration is measured through rehearsal loads, and a threshold is set for the rollback decision.
- Window duration is measured with rehearsal loads
- Rollback threshold and decision owner are written down
- A manual-processing list is prepared for the cutover
- The first period close after go-live is supported separately
Stability after go-live
A migration does not end on go-live day. The first period close, interface behavior under load and support demand are planned separately, with extra capacity reserved for that stretch.
- Extra support is planned for the first period close
- Separate monitoring is set up for interface errors
- User questions are classified and fed back into training
- Open items are tracked to a closing date