Migration projects are planned around the platform and the go-live date. The most expensive risk usually sits elsewhere: in the data the business uses every day.
Every ERP project reaches an uncomfortable moment. The functional team has finished the design, the integrator has prepared the test scenarios and the sponsor has approved the date. Then the data shows up: duplicate customers, materials coded with different logic in each plant, a chart of accounts with years of patches and suppliers nobody dares to deactivate.
This is neither a minor problem nor one limited to migrations. Gartner estimates poor data quality costs an organization an average of 12.9 million dollars a year. And according to its own surveys, 59% of organizations do not measure the quality of their data. If it is not measured, the problem only shows up when that data has to move to a new system.
Why the migration exposes it
An old ERP lives with dirty data because people compensate: they know which duplicate customer to use, which material code is the right one, which account to ignore. A new system has no such memory. It validates, rejects and, when it does not reject, spreads the error to every process it touches.
In the projects we support, when data cleanup was not addressed before design, it ended up consuming between 20% and 35% of total effort and stretched testing by several weeks. The costliest part was not the effort itself but what came after: reports that do not reconcile in the first months, audit findings and a finance team that loses trust in the new system.
Data as a workstream of its own
The alternative is to treat data as a separate workstream that starts before design and has its own owners:
- Inventory of critical master data. Customers, suppliers, materials, chart of accounts and assets, with their volume and real condition.
- Business owners. Each master needs a functional owner who decides what gets cleaned, merged or discarded. It is not a task for the technical team.
- Validation rules at the source. Load programs are built with the same rules the new system will apply, so errors surface early and not in integrated testing.
When not to clean everything
Not all data deserves the same effort. Cleaning ten years of history in order to migrate it is usually a poor investment. The sensible approach is to focus cleanup on what will operate in the new system, open balances and documents, active master records, and keep history archived and accessible outside the ERP.
How we approach it at Aunergia
In the SAP S/4HANA, Oracle Fusion or JD Edwards migrations we support, data readiness is a separate workstream that starts before design. We review critical master data, identify duplicates and inconsistencies, assign an owner to each and prepare loads with validation at the source. By the time the project starts, the data is ready to migrate.
Is your next migration planned around the go-live date, or around the data you will run on the day after?
Sources: Gartner, data quality research; Aunergia project experience.
Comentarios recientes