Master data management (MDM)
Master data management (MDM) is the set of policies, processes, and technology a company uses to create and maintain a single, authoritative record — the "golden record" — for its core business entities (products, customers, suppliers, locations) and distribute that record consistently across every downstream system. It is a governance discipline, not a data-quality or enrichment tool.
What MDM covers — and what it doesn't
MDM is not a single product. It's a discipline aimed at one problem: the same entity described differently in every system that touches it. The vendor who is "Acme Corp." in your ERP, "ACME Corporation" in the CRM, "Acme" in accounts payable, and missing entirely from the portal is the MDM problem in miniature.
Most organizations manage four or five MDM domains:
- Product master — SKUs, part numbers, attributes, hierarchies, packaging
- Customer master — accounts, contacts, billing addresses, credit terms
- Supplier/vendor master — vendor IDs, lead times, certifications, pricing tiers
- Location master — warehouses, distribution centers, stores, plant codes
Every domain has one type of record that has to be correct everywhere at once. Without a deliberate MDM practice, it rarely is. Orders ship to obsolete addresses. The same supplier appears under four names. A product exists in the ERP but not in the commerce platform, or vice versa.
What MDM does not cover: whether the data inside those records is complete, accurate relative to the physical product, or useful to a buyer. That's a separate problem, and a common source of disappointment for teams that invest heavily in MDM governance and still find their catalog underperforms.
MDM vs. PIM — overlapping but not the same
A Product Information Management (PIM) system manages the product domain's master data, which makes it a domain-specific implementation of MDM. The confusion arises because the two are sold by different vendors to different buyers — and neither camp rushes to clarify the overlap.
An enterprise MDM platform — Informatica MDM, Reltio, Stibo STEP, Syndigo — usually manages multiple domains across IT systems and is bought by IT and data governance teams. A PIM (Akeneo, Salsify, inRiver, Pimcore) is bought by catalog, merchandising, or e-commerce teams and focuses entirely on product: it adds channel syndication, digital asset management, and the enrichment workflows a generic MDM hub doesn't include.
In practice, most mid-market manufacturers and distributors use their PIM as their product MDM — one source of truth, one schema, one place where the product record lives. Enterprises with complex multi-domain data problems layer a proper MDM hub beneath the PIM to tie together ERP, CRM, and commerce. Neither arrangement is wrong; the important thing is that one system wins and the others sync to it.
The practical test: if someone makes a product change, how many places have to be updated? If the answer is more than one, you have an MDM problem regardless of how many systems you own.
How MDM is implemented — the three models
MDM implementations follow three broad patterns, each with different tradeoffs between control and integration overhead.
Consolidated. One central hub owns the master record. All systems write to and read from it. Authoritative and clean, but requires every system to integrate with the hub — high upfront investment, and a single point of failure if the hub goes down.
Coexistence. Each source system keeps its own local record; the MDM hub produces a "golden record" that reconciles and harmonizes them. Systems can still write locally. The hub broadcasts the authoritative view downstream. Most common in large B2B organizations where the ERP, PIM, and commerce platform all legitimately own parts of the product record.
Federated. No central repository. A shared ontology and governance rules let domain teams manage records in local systems while a semantic layer assembles a consistent view on read. More flexible, harder to enforce — works better in organizations where central IT authority is limited.
For B2B distributors and manufacturers, coexistence is typically the working model: the PIM is the golden record for marketing attributes and digital assets; the ERP holds pricing, inventory, and logistics data; middleware keeps them in sync. The risk is that over time the sync breaks in one direction or the other and you end up with two competing masters — which is the MDM problem reasserting itself.
Common mistakes — and the gap MDM leaves open
Treating MDM as a one-time project. New suppliers onboard continuously. Products get cloned, renamed, split into variants. SKUs migrate between categories. A master data program declared "done" degrades at exactly the rate your catalog changes. MDM is maintenance, not a migration.
Conflating governance with quality. MDM establishes who owns the record and where it lives. It does not fill missing attributes, fix thin descriptions, or validate that data is accurate relative to the actual physical product. A golden record that says length: [blank] is authoritative. It's also useless to a buyer.
Over-investing in the hub, under-investing in the data. Some organizations spend 18 months evaluating and implementing an MDM platform, then populate it with the same sparse supplier-copy data they had before. The hub enforces uniqueness and routing. It doesn't make the content better.
Declaring every system the master. The point of MDM is one golden record. Teams that maintain a "master" in the ERP, a "master" in the PIM, a "master" in the commerce platform, and a "master" in the CDP have solved governance by declaring every system authoritative — which means none of them are. The first conflict between systems exposes this.
The enrichment gap. This is the one that matters most in B2B e-commerce. A well-governed MDM program creates the right container. It doesn't fill it. The MDM layer ensures there's one product record with one part number, mapped to one category, owned by one team. Necessary — but not sufficient. For that product to rank on a marketplace, get cited by an AI assistant, or convert a buyer who never talks to a sales rep, the record inside that container has to answer buyer questions: dimensions, compatibility, application context, copy written for someone speccing a commercial HVAC install rather than a residential one. None of that comes from governance. It comes from enrichment — pulling what's missing from suppliers, the open web, and buyer signals, then writing it back into the authoritative record so every downstream channel inherits it.
The correct sequence: govern the record first, then enrich it. MDM without enrichment gives you a tidy filing cabinet. Enrichment without MDM gives you rich data scattered across conflicting systems. Together, they give you one complete, buyer-ready record that every channel draws from without having to fix the same SKU twice.
Frequently asked questions
What is the difference between MDM and a PIM?
A PIM (Product Information Management system) is a domain-specific MDM implementation focused on product records — it adds channel syndication, digital asset management, and e-commerce workflows that a generic MDM hub doesn't include. MDM is the broader governance discipline covering all master data domains: products, customers, suppliers, and locations. Most catalog and e-commerce teams work primarily inside a PIM; enterprise IT and data governance teams manage the MDM architecture that connects it to ERP, CRM, and other systems.
What is a golden record in MDM?
A golden record is the single, reconciled, authoritative version of a master data entity — a customer account, product SKU, or supplier record — that an MDM system produces by resolving conflicts across multiple source systems. It's the version every downstream system is supposed to trust. A golden record solves the consistency problem (one answer everywhere); it doesn't guarantee that answer is complete or accurate relative to the real-world product.
Does MDM improve product data quality?
MDM governs consistency and routing — it ensures one record exists and that changes flow to the right places. It doesn't make product data more complete or accurate relative to the physical product. Filling missing attributes, writing useful descriptions, and validating specs against reality is enrichment work that sits on top of the MDM layer. Teams that expect MDM to solve data quality are usually disappointed; the governance is working, but the data inside was thin to begin with.
What MDM platforms are common for B2B product data?
Enterprise MDM platforms with strong product domain support include Informatica MDM, Reltio, Stibo STEP, and Syndigo. For product-domain-specific needs, most B2B distributors and manufacturers treat their PIM (Akeneo, Salsify, inRiver, Pimcore) as the product master, integrated to an ERP for pricing and inventory. The right choice depends on how many non-product domains need governing — if the answer is "several," a dedicated MDM platform makes more sense than a PIM alone.
How does MDM relate to data syndication?
MDM governs the master record; syndication distributes it to downstream channels — retailers, marketplaces, partner portals. The two are complementary: strong MDM gives syndication one clean source to transmit, and syndication tools handle the format translation and delivery that MDM platforms don't. The risk is enriching only at the syndication layer (the feed) rather than writing improvements back into the master record — that fixes the channel-specific output but leaves the source thin, so every new channel inherits the same gaps.