Master Data Management for ERP: Customers, Products, and Suppliers
By Apex Horizon Digital
ERP workflows depend on master data that appears simple until the same customer has four records, one product uses three units, or a supplier changes bank details without review. These defects spread into orders, inventory, invoices, reporting, and integrations. Master data management is the operating discipline that defines trusted records, assigns ownership, controls changes, and measures quality. It does not require a large committee. It requires a practical data dictionary, named stewards, validation rules, and a workflow that makes the approved path easier than an unofficial spreadsheet.
Key takeaways
- Treat customer, product, and supplier records as governed business assets, not clerical entries.
- Assign one accountable owner for every domain and define which teams may propose, approve, and execute changes.
- Measure completeness, uniqueness, validity, consistency, and timeliness with visible exception queues.
2. Assign ownership and a lightweight governance workflow
A domain owner is accountable for policy and quality. Data stewards handle daily review, while requesters provide evidence and technical administrators maintain rules. Write a responsibility matrix for create, review, approve, change, merge, deactivate, and restore actions. Separate sensitive changes such as supplier bank accounts, customer credit limits, tax identifiers, and product valuation attributes from routine updates.
Use a request workflow with reason, supporting document, affected records, effective date, reviewer, and audit trail. Urgent changes still need an expedited controlled path. Service targets help prevent teams from creating shadow records because approval feels slow. Review the queue weekly for bottlenecks, repeated rejection reasons, and changes that should be automated through verified source systems.
- Policy owner and daily steward
- Requester, reviewer, and executor
- Sensitive-field approval path
- Service target and escalation
3. Set required fields and validation by lifecycle stage
Not every field must be complete at initial capture, but each workflow stage needs an entry standard. A sales prospect may need minimal details, while a customer approved for invoicing needs legal name, tax treatment, billing address, payment terms, and credit controls. A draft product may lack a final price but cannot be released for purchase without unit, category, lead time, and accounting mapping. Define these gates by status.
Validate format, range, reference, and combination rules. Postal code can depend on country, unit conversions must be positive, and an inactive supplier should not receive new purchase orders. Avoid defaults that conceal uncertainty, such as assigning an unknown tax code or warehouse. Route invalid submissions to an exception queue with a clear explanation and owner.
- Minimum fields by lifecycle status
- Format and reference validation
- Cross-field business rules
- Visible exception handling
4. Prevent duplicates and manage merges safely
Duplicate detection should combine normalized names with stronger signals such as tax number, telephone, email domain, address, bank account, manufacturer code, or barcode. Configure exact and possible matches separately. A possible match should prompt review rather than block legitimate branches. Record why a steward declared records equivalent or distinct so later reviewers understand the decision.
Merging is a controlled transaction, not a simple deletion. Select the surviving identifier, preserve source references, move allowed relationships, resolve conflicting values, and retain an audit trail. Test how open orders, invoices, inventory, pricing, contracts, and integrations behave. Some records should be linked rather than merged because legal or reporting distinctions matter.
- Normalized and strong matching fields
- Exact versus possible-match rules
- Survivor and cross-reference policy
- Relationship and audit preservation
5. Operate a quality scorecard and change review
Track completeness, uniqueness, validity, consistency, timeliness, and unresolved exceptions by domain and responsible team. Report operational consequences too, such as orders blocked by missing terms, invoices rejected because of tax data, stock adjustments caused by unit errors, or payments paused for bank-detail review. Consequence-based measures keep governance connected to business value.
Review high-risk changes, duplicate trends, aging exceptions, rejected integration records, and fields frequently corrected after creation. Retire fields nobody uses and strengthen rules where defects repeat. The core artifact is a living data dictionary paired with ownership, approval, duplicate, and lifecycle rules. Good master data makes every ERP module and connected application easier to trust.
- Quality dimensions by domain
- Exception aging and owner
- Operational impact measure
- Rule and field retirement review