Master data governance for OEMs: keeping data clean after the cleanup
Cleaned data decays at a steady pace unless someone owns it. Owners, rules at the point of creation, automated checks and a monthly routine that keeps quality where the cleanup left it.
Not ready to talk yet? Take the 3-minute Data Foundation Check
TL;DR
- Data quality drops after every cleanup unless the creation process changes.
- Give every data object an owner with authority, not only a steward without a mandate.
- Check at creation, not afterwards: required fields, duplicate check, classification, approval.
- Measure with a small set of rules per object and review the trend monthly.
Why does clean data get messy again?
- New parts are created under time pressure, without classification or duplicate check.
- Free-text fields absorb everything that has no proper field.
- Nobody owns the object as a whole, only single fields.
- Rules exist in a document, not in the system.
Who owns what?
| Object | Business owner | Typical rules |
|---|---|---|
| Material / part master | Engineering standards (technical), purchasing (commercial) | Class and key attributes, no duplicates, unit rules |
| BOMs | Engineering / manufacturing engineering | No blocked items, effectivity on changes |
| Service BOM & supersession | Spare parts management | Positions, interchangeability codes |
| Customer master | Sales operations | Duplicate check, VAT ID, parent |
| Supplier master | Procurement | Duplicate check, approved bank data, tax data |
An owner is a role with authority
The data owner decides the rules and approves exceptions for an object. The data steward does the daily maintenance. Without owners, stewards end up enforcing rules nobody backs.
What does the routine look like?
Data-quality score (per object, monthly) = records passing all rules ÷ records in scope Example rules, material master: MAT-01 class and key attributes filled MAT-02 no duplicate (match score < 90) MAT-03 base unit in approved list MAT-04 purchased parts carry a manufacturer part no. MAT-05 spare parts have an image and where-used
What should a managed service cover?
| Task | Frequency |
|---|---|
| Run rules and update the score | Nightly or weekly |
| Work through exceptions | Weekly |
| Duplicate check on new records | At creation |
| Report to owners | Monthly |
| Review and adjust rules | Quarterly |
Checklist
- A named owner per data object
- Rules written per object and enforced at creation
- Automated checks running on a schedule
- Monthly score per object and owner
- An exception queue with a steward assigned
FAQ
What is master data governance?
The roles, rules and routines that keep master data such as parts, BOMs, customers and suppliers correct and consistent over time.
Why does data quality drop after a cleanup?
Because the processes that created the bad data haven’t changed. New records are still created under time pressure without checks, and quality drifts back.
What is the difference between a data owner and a data steward?
The owner decides the rules and approves exceptions for a data object. The steward maintains the data and works through exceptions day to day.
Which data-quality KPIs should we track?
A score per object based on a small set of rules, such as completeness of key fields, absence of duplicates and valid values, reviewed monthly.
Can data governance be outsourced?
The routine work can: running checks, fixing exceptions and reporting. Ownership and decisions on rules should stay inside the company.
How fast does your data drift after a cleanup?
Send us a part master extract under NDA. We’ll run our standard rules and show you the score per object.




