A single duplicated supplier record doesn't stay in one place. It travels — through twelve systems, across hundreds of reports, into the decisions your teams make every day.
It usually starts with something innocuous. A supplier onboarded twice — once by the procurement team, once by finance, slightly different naming conventions, slightly different address fields. Neither team notices. The system accepts both. The MDM repository, if there is one, wasn't consulted at the point of entry.
Six months later, that record isn't just a duplicate in a database table. It has participated in hundreds of transactions. It has shaped purchase order workflows, seeded vendor master files, influenced payment terms calculations, and fed the analytics pipeline that generates your supplier performance dashboards. The record that "nobody bothered to fix" has become the invisible foundation of a significant portion of your procurement reality.
The hidden cost isn't the record. It's everything the record touched.
When we audit data quality issues inside enterprise environments, the first thing we map is not the record itself but its contamination footprint. One bad record — a duplicate, a wrong hierarchy assignment, a stale attribute — typically has three cascading layers of impact.
The first layer is system propagation: ERP modules that synchronized the corrupted entity, CRM records built on that entity's contact data, data warehouse pipelines that ingested it as a valid dimension. Each synchronization point becomes a new contamination node.
The second layer is process distortion: workflows designed around the assumption that entities are distinct now run on a reality that isn't. Payment thresholds get applied twice. Risk scores are calculated on a split entity. Segmentation logic fires on phantom categories. None of this is flagged as an error — it just runs wrong, quietly, for months.
The third layer is the most expensive: decision contamination. Every report generated from contaminated data, every KPI calculated against duplicated entities, every strategic decision anchored in analytics that carry the error. By the time the bad record is found, it has influenced decisions at multiple organizational levels.
"The record costs almost nothing to fix. The contamination it leaves behind — in systems, processes, and decisions — costs significantly more."
The 847-hour figure is not abstract. When we break it down by activity type across a remediation engagement, the distribution is consistent: roughly 35% of effort goes to identifying all contamination points (many teams don't know how far the record traveled); 40% to correcting downstream systems and recalculating affected outputs; and 25% to validating that the correction propagated correctly and didn't introduce new inconsistencies.
The identification phase is almost always underestimated. Data lineage documentation in most enterprises is partial at best. Teams know what they own — they rarely know what consumed data from what they own. Tracing the contamination map is manual, time-intensive, and almost always reveals more touchpoints than the initial estimate.
The contamination map of a single duplicated supplier entity — from origin node to full enterprise footprint.
Finance teams budget for data migration. They budget for MDM tool licenses. They almost never budget for the ongoing cost of unresolved data quality — because it doesn't arrive as a line item. It arrives as lost hours, inflated headcount in data remediation roles, and a persistent background noise of "we can't trust this report."
The 47K€ figure per record is a conservative estimate based on fully-loaded remediation hours. It doesn't include the cost of decisions made on bad data — and decisions made on contaminated analytics are where the real damage accumulates. A procurement strategy built on a split supplier risk profile, a customer retention campaign targeting a phantom segmentation, a board presentation anchored in KPIs inflated by duplicate entities — these are costs that never appear on a remediation invoice but are real nonetheless.
The ROI of MDM, properly framed, is not the cost of the MDM system. It's the cost of all the bad records you didn't create, all the cascades that didn't happen, and all the decisions that were made on data you could actually trust.