A single source of data means the organisation knows which record is authoritative for every important business entity. Customers, items, suppliers, prices, work orders and invoices cannot have several equal versions that are reconciled only before reporting. ERP does not need to originate every record, but there must be a clear rule for how each record enters the shared business context.
Fragmentation is not only duplicate entry. Every copy introduces its own update time, naming convention, status and owner. Two spreadsheets may contain the same amount while describing different business realities. Defining the identity and lifecycle of data is therefore more important than simply moving everything into one database.
Sales may track the agreed order value, the warehouse the quantity actually delivered and finance the issued invoice. All three values may be correct, but they are not interchangeable. A reliable ERP connects them through the same business event and preserves their stage. Management then asks which measure fits the decision, rather than which spreadsheet is correct.
A new database can centralise old inconsistencies. Without rules for creating customers, changing prices, closing orders or reversing documents, errors simply become available to every department more quickly. Migration must therefore include cleaning, mapping and a deliberate decision about what not to move. Historical data without meaning is often more expensive than data intentionally left in an archive.
A good result is not zero exceptions. It is an exception that is visible, owned and explainable without manually collecting evidence. Finance can trace a document to the operational event, operations understand the consequence of a change and reporting clearly states the period and data status it uses. The single source then becomes a management discipline rather than an IT slogan.
One version of the truth does not mean one number for every decision. It means a clear source, meaning and responsibility for every number.