GULDMANN.

Architecture · Data · Intelligence

Master Data Management

PROCESS-DRIVEN MDM · DATA OWNERSHIP · INFORMATION QUALITY

Master data management is not primarily a data problem. It is a business operating problem expressed through data.

When customer, supplier, product or other core information is unreliable, the instinct is often to start with the data: profile it, cleanse it, deduplicate it, consolidate it and build a better repository. Those activities may be necessary. But they rarely explain why the same problems keep returning.

Data is created and changed through business processes. If those processes have unclear responsibilities, conflicting definitions or weak controls, the information will reflect that reality. A central master data platform cannot compensate for an organisation that has not decided who is responsible, what a record means or which process outcome matters.

My approach is process-driven and deliberately pragmatic: start where poor information causes real pain, fix the decisions and responsibilities behind it, deliver value, build trust and scale with discipline.

Start with the process, not the domain

Customer, supplier, product and location are useful labels, but they are not a strategy. Begin by understanding where information is created and used, which process depends on it, and what happens when it is wrong, missing or late.

A critical data element is only critical in relation to something: a process, a decision, a control, a customer promise or a regulatory obligation. Without that context, criticality becomes another abstract classification exercise.

The principles behind process-driven MDM

1. Follow the business consequence

Find the operational friction caused by poor information: rework, delays, failed automation, compliance exposure, bad decisions or lost customer confidence. Use that consequence to define where to begin and what success should mean.

2. Make ownership real

Ownership is not a name in a RACI chart. It is the authority and accountability to make decisions about information and the process that creates it. Attribute-level responsibility may be distributed across the business, but someone must be accountable for resolving conflicts and maintaining the overall outcome.

3. Build quality into creation

Prevent avoidable defects where information enters or changes. Validation, enrichment, evidence and business rules are most effective when they support the person or system creating the information, rather than leaving another team to repair it downstream.

4. Govern what matters

Apply stronger controls to information with meaningful business consequences. Define the decisions, thresholds and escalation paths that are needed. Avoid turning every attribute into a committee discussion or every change into a central ticket.

5. Keep the operating model proportionate

Governance should be the minimum structure required to make decisions and sustain improvement. A small, capable central function can set direction and resolve cross-business issues while ownership remains close to the processes and people who understand the information.

MDM is not a central data queue

MDM should not become a bottleneck through which every data change must pass. Nor should it become an endless programme of domain definitions, committees and policy documents. Its purpose is to make reliable information part of how the business operates.

That requires clear decision rights, usable business rules, process-aware controls and feedback from the places where information is actually used. Technology matters, but it should enable this operating model rather than dictate it.

What AI changes, and what it does not

AI creates new opportunities to interpret unstructured information, identify possible matches, enrich records and assist people at the point of creation. It can make tasks that once required extensive rule-writing or manual review faster and more adaptable.

But AI does not decide which business definition should prevail, who is authorised to approve a change or what evidence is sufficient. Those are questions of meaning, ownership and accountability. In practice, AI makes a sound MDM foundation more valuable because more decisions can now be supported or automated, and the consequences of getting them wrong can travel further and faster.

Start here

Why AI Turns Product Data Quality from a Clean-up into a Creation Habit
How AI changes the economics of product data quality, while making creation-time discipline even more important.

Master Data at Birth: Why Golden Records Should Be Created, Not Reconciled
Why quality and identity should be addressed before records spread across the landscape.

Stop Chasing Data, Start Fixing Processes
A shift from treating data as the problem to understanding the processes behind it.

Critical Data Elements Are Not About Data
Why criticality only makes sense when connected to business use and consequence.

Golden Records Are Beliefs, Not Mirrors
On trust, evidence and the assumptions hidden inside data consolidation.

Master Data as Applied Epistemology
Understanding what we know, why we believe it and how decisions are made.

The wider argument

Master data management succeeds when it helps the business operate better. The measure is not the size of the repository, the number of attributes governed or the volume of cleansing completed. It is whether the processes that depend on information become more reliable, efficient and accountable.

Start where the pain is. Deliver value quickly. Build trust. Scale with discipline.

For the broader architectural context, explore Architecture. For the changing role of AI in enterprise information, visit AI & Intelligence. You can also browse the full article archive.