Definition
Master data management, or MDM, is the governance, processes, and tools used to maintain consistent and authoritative records for core business entities across systems. In procurement and supply chain, those entities include suppliers, items, locations, and organizational units.
MDM establishes which records the business can rely on, who can change them, and how approved changes reach connected systems. It is an ongoing operating discipline rather than a one-time exercise in cleaning a spreadsheet.
Why consistent names are not enough
Two supplier names may refer to the same legal entity, different subsidiaries of one parent, or unrelated businesses with similar names. Those distinctions affect both commercial analysis and transaction controls.
For sourcing analysis, a parent-level view may reveal total exposure or negotiating scale. For purchasing and payment, the business may need separate legal entities, sites, tax registrations, currencies, and payment instructions.
A useful MDM design preserves these relationships instead of collapsing every related record into one supplier. A group-level analytical rollup should not erase the identity of the actual contracting party or payee.
Establish authority by data domain
VPP recommends assigning business ownership before selecting matching rules or automation. Procurement, Finance, Operations, and other functions may have different responsibilities within the same supplier or item record.
- Supplier identity: Define evidence for the legal entity, parent relationship, operating site, and approved business status. Name similarity alone is not sufficient proof.
- Item identity: Govern specifications, units of measure, packaging conversions, and approved alternatives. Similar descriptions do not establish interchangeability.
- Commercial attributes: Identify where effective contract terms and purchasing conditions are maintained. Not every negotiated term belongs directly in a supplier master.
- Payment attributes: Apply restricted access, independent verification, and appropriate separation of duties. An analytical match must not trigger an unreviewed bank-account change.
Ownership should cover creation, modification, suspension, and retirement of records. A named data steward needs a defined review path and an accountable business decision-maker, not just a queue of unresolved tickets.
A hypothetical supplier and item crosswalk
Consider the same component initiative used across this editorial group. Two ERPs identify the approved supplier as A-101 and V-007; validated legal-entity documentation establishes that both refer to the same supplier.
The analysis can map those identifiers to one authoritative supplier identity while retaining each source-system key. If the systems describe the component in individual units and cases of 100, a separately validated item crosswalk supplies the conversion.
| Record layer | Controlled treatment |
|---|---|
| Source supplier IDs | Retain A-101 and V-007 for traceability |
| Legal supplier identity | Map both IDs to the verified contracting entity |
| Parent relationship | Maintain as a separate analytical hierarchy |
| Item and packaging | Validate the component specification and 100-unit case |
| Payment instructions | Preserve the approved workflow; do not infer changes |
The component's on-contract transactions total 50,000 individual units and $452,000 of recorded expenditure. Connecting the supplier records makes that exposure visible across systems, but it does not create another $452,000 of spend.
Likewise, merging supplier identities is not the same as deleting duplicate invoices. Transaction deduplication requires document-level evidence, including source system, company, document number, line, and relevant adjustment relationships.
Put changes through a repeatable lifecycle
A practical MDM workflow captures a proposed change, validates its evidence, obtains the required approval, distributes the approved record, and checks that downstream systems accepted it. The workflow should also preserve the prior value and effective timing.
Change controls matter after onboarding as much as during it. A supplier acquisition, a new distribution site, an item-packaging change, or a deactivated entity can make a once-correct record unsuitable for a later transaction.
Do not assume that a record is current because it passed an initial review. Define review triggers and responsibilities according to the attribute's risk and the business process it supports.
Measure fitness for decisions, not just filled fields
Completeness is useful, but a fully populated record can still be wrong. Quality measures should test accuracy, consistency, timeliness, and whether the data supports the intended transaction or analysis.
Useful operating measures include unresolved identity conflicts, aged change requests, transactions using inactive records, failed downstream updates, and high-risk changes awaiting independent verification. Report these by accountable owner and material exposure where possible.
Avoid a single enterprise score that hides a critical supplier or item problem. A missing optional description and an incorrect payment destination should not carry the same operational significance.
Frequently asked questions
- Does MDM require one ERP? No. A governed identity model and reliable crosswalks can support multiple systems, provided ownership, synchronization, and exceptions are controlled.
- Is a golden record always one physical database record? It is an authoritative view of an entity, not necessarily a requirement to replace every operational record. The implementation should preserve necessary legal, site, and system distinctions.
- Can AI approve master-data changes? AI can assist matching and exception prioritization. Approval rights and evidence requirements should remain explicit, especially for sensitive or high-impact attributes.
Build a baseline the business can keep using
Within the VPP Sourcing Control System, a reliable baseline depends on governed supplier and item identities rather than repeated analytical repairs. VPP connects that foundation to spend visibility, purchasing execution, and performance monitoring so that improved data remains useful after the initial assessment.