An industrial company. A component. One wrong unit in the bill of materials, millimeters instead of centimeters. The ERP accepts the entry without complaint, because an ERP system isn't a proofreader. It's a multiplier. Whatever goes in comes back out at ten times the speed.
In this case: 400 incorrectly manufactured parts, a stopped production line, three days of delivery delay, and a customer complaint letter that ran four pages. The original mistake? A 30-second data entry.
The ERP as a nervous system, and what happens when a signal is wrong
An enterprise resource planning system is the central nervous system of a manufacturing company. It connects purchasing, production, warehousing, shipping, accounting, and sales, and passes information between all of them. The problem: it also passes on wrong information. Instantly, automatically, and without warning.
The error chain reaction
The most common ERP error sources, and who's really at fault
Wrong product descriptions
Product descriptions in an ERP often get written in a hurry, when a new product is created, during translation between country entities, during migrations from legacy systems. Missing or wrong descriptions lead directly to wrong sales materials, wrong quotes, and in the worst case, contract disputes (when the delivered product doesn't match what was contractually agreed).
Wrong bills of materials (BOM errors)
The bill of materials is a product's DNA. One wrong component, one wrong quantity, one missing part, and production builds something that shouldn't exist. BOM errors are especially dangerous because they often get carried forward across many product generations, until a delivery problem or a quality complaint eventually exposes the source.
No clear ownership of master data
This is the actual root cause. At most companies, any department can create master data, but nobody is ultimately responsible for its quality and completeness. Product management creates it, IT maintains it, sales adds to it, purchasing corrects it. The result: inconsistent data, duplicate items, and errors that never get fixed systematically.
At a MedTech company (a SAGE environment, flat hierarchy with no product structure), a master data review found: no clean product descriptions, missing components in the bills of materials, and nobody left who knew which product belonged with which connector. Everyone in the company knew about the situation. Nobody had fixed it. The customer, meanwhile, was wondering why they'd received the wrong product.
What good master data governance looks like
Master data governance isn't an IT project. It's a leadership decision. These building blocks make the difference:
- Name a data owner: for every master data domain (material, customer, supplier, BOM), there's one named person with decision authority.
- Define required fields: what has to be complete before release? An item without a full description, weight, and customs code shouldn't be creatable at all.
- Four-eyes principle for critical master data: bills of materials, price lists, and stock-keeping units get checked by a second person before activation.
- Regular data quality reviews: a quarterly assessment of completeness on critical fields, tracked as a KPI, not treated as an IT task.
- Clear change processes: who's allowed to change master data? Under what conditions? What happens to open orders when a BOM changes?
The real cost of poor data maintenance
Bad master data is rarely dramatic. It's a constant, quiet cost leak. Gartner estimates that companies lose an average of $12.9 million a year to poor data quality, through manual corrections, production stoppages, delivery delays, and complaint handling costs.
At small and mid-sized companies, the numbers look smaller. But the ratio holds: an hour spent on clean data entry saves an average of 8–12 hours of downstream correction work. The question isn't whether you can afford master data governance. The question is whether you can afford the alternative.
"In an ERP system, every error has children. Most of them you only notice once the grandchildren show up."
How do you spot a systemic data problem?
The early warning signs are subtle, but consistent:
- Purchasing regularly places follow-up orders because quantities don't add up
- Sales sends individual quotes instead of pulling from the system, because catalog prices are wrong
- Production keeps manual correction lists running alongside the ERP
- Accounting has recurring discrepancies in inventory valuation
- New employees learn: "It says that in the system, but in reality it's actually..."
If that sounds like your company: this isn't a normal state of affairs. It's a fixable system problem. And the first step is always an honest assessment, not new software.