Twelve SAP systems had grown up around twelve different ideas of what good master data looks like. Each plant maintained its own naming conventions, its own mandatory fields and its own spreadsheet of local exceptions. Records created in one region routinely failed validation in another, and nobody could say with confidence which version of a record was authoritative.
The data quality team spent most of its week correcting records after the fact rather than preventing bad ones from being created. Every correction was itself a manual SAP transaction, and every correction carried its own risk of introducing a new error.