The technology question is the easy one. What separates MDM programs that last from those that stall is a set of decisions taken early, about scope, ownership and measurement. These are the eight that matter most, drawn from the programs APGAR has delivered for international groups.

The Challenge : why MDM Programs stall? .

Where Programs Lose Momentum

Few MDM programs fail outright. Most lose momentum, and they lose it in recognizable ways: a scope that grows faster than the delivery, a governance model nobody in the business recognizes as their own, a first release that improves data nobody had complained about, and a steering committee that cannot say what the program has produced.

Technology Is Not the Problem

None of these is a technology problem. Each is a decision that was taken implicitly at the start, and revisited too late. The eight practices below are those decisions, made explicit.

Business alignement: Start with business outcomes and deliver value progressively

1. Start from a business outcome, not a data domain

The first question is not which domain to govern. It is which business outcome the program is accountable for. Master data management supports four distinct types of goal: running the business better, growing it, transforming it, and reducing the technical debt created by disparate business languages across departments.

Naming the goal first changes what gets built. A program accountable for reducing regulatory risk designs different controls from one accountable for time-to-market on new products. A program that names no goal at all designs for completeness, which is the most expensive option and the least defensible at budget time.

2. Sequence one domain at a time

Multidomain ambition is right; multidomain delivery is not. The programs that keep their funding deliver one domain, prove it, and use the proof to open the next. The programs that stall try to model customers, suppliers, products and finance structures simultaneously, and reach the first release eighteen months later with nothing a business user can see.

Choosing the first domain is a judgement about value and feasibility together. Reference data — the shared lists of values every other domain depends on, units of measure, country codes, payment terms — is often the fastest quick win, because it is small, uncontroversial, and immediately visible in every consuming system.

 

Governance

3. Design governance around your decision rights, not your tool’s

Every MDM platform ships with a governance model. It is rarely yours. Adopting the predefined governance of the tool means approval paths that follow the software’s assumptions about who decides what, rather than the way your organization actually decides.

The consequence is predictable and it is always the same: the business finds a way around the workflow. Governance designed around your real decision rights, and configured to fit them, is slower to specify and far more likely to survive.

4. Name data owners in the business, not in IT

An MDM program needs named accountability for each domain, and that accountability has to sit where the definition is decided. When nobody in the business is accountable for what constitutes a finished good, or the same customer, the definition drifts by department and no platform will hold it together.

Naming owners is the easy part. What makes it work is the rest: identifying them, training them, mentoring them through the first months, and giving them a workspace they can actually use without an ERP licence. Ownership that exists only in a RACI chart is not ownership.

 

Data & Technology

5. Keep the data model business-oriented and the integration native

A master data model built to mirror the ERP inherits the ERP’s limits: it holds the attributes transactions need, and has no natural home for the ones the CRM, the sourcing system or the ESG reporting require. A model built purely for the business, with integration treated as a later problem, creates a second reconciliation exercise.

The workable position is both: a business-oriented data model, and preconfigured APIs that keep the integration with the transactional systems smooth. That combination is precisely what an accelerator provides.

6. Embed quality controls at creation, not after

Data quality remediation is a permanent cost. Quality at creation is a one-off design effort. A record created in five minutes to unblock an urgent process, with a third of its mandatory attributes populated, will surface as a costing error, a failed payment or an audit finding months later — and someone will fix it by hand.

Embedding validation, standardization and completeness rules at the point of entry moves the effort upstream, where it is cheapest, and changes what data stewards spend their time on.

 

Value & Adoption

7. Calibrate governance to each object’s impact

The most common governance failure is uniformity. A single global approval path applied to every object, in every country, is the path most often bypassed — because a local controller who cannot work with it will find another way, and the exception becomes the rule.

Governance should be proportionate: a validation path matched to the object’s real impact, and stewardship granularity defined so that roles and authorizations can be adjusted by country, zone or legal entity. Configurability here is not a luxury feature. It is what makes the rules survive contact with the organization.

8. Measure and communicate value from the first release

Programs lose funding not because they deliver nothing, but because what they deliver is reported in metrics nobody outside the data team values. Completeness rates and duplicate counts describe the work; they do not describe the outcome.

“Make authoritative master and reference data easy to discover and consume. Make trusted data products easier to access than uncontrolled copies.”

Photo Forbes Bruno BILLY

Bruno BILLY

President & CEO, APGAR North America

FAQ about MDM best practices

What are the best practices for an MDM project ?

Cosed

Start from a named business outcome, deliver one domain at a time, design governance around your own decision rights rather than the tool’s, name data owners in the business, keep the data model business-oriented, embed quality controls at creation, make approval paths proportionate to impact, and measure value from the first release.

Where should an MDM program start ?

Cosed

With the domain that combines visible value and low controversy. Reference data — the shared lists of values every other domain depends on — is often the fastest quick win, because it is small, uncontested and immediately visible in every consuming system.

Who should own master data ?

Cosed

Named owners in the business, where the definition is actually decided, not in IT. Ownership only works when owners are identified, trained, mentored through the first months and given a workspace they can use without a transactional system licence.

Why do MDM projects fail ?

Cosed

Rarely for technical reasons. They stall when scope grows faster than delivery, when the governance model is the tool’s rather than the organization’s, when the first release improves data nobody complained about, and when results are reported in metrics executives do not value.

How do you avoid governance being bypassed ?

Cosed

By making it proportionate. A single global approval path applied to every object in every country is the one most often circumvented. Validation paths matched to each object’s impact, with roles adjustable by country, zone or legal entity, survive contact with the organization.

How do you measure the success of an MDM program ?

Cosed

By mapping master data improvements to the business outcomes they affect, then quantifying the impact in four categories: risk reduction, revenue growth, cost savings and corporate governance. Completeness rates and duplicate counts describe the work, not the outcome.

Business value : turn better Master Data into measurable business outcomes.

Connect Data to Business Outcomes

MDM creates value when improvements in master data can be connected to outcomes the business already tracks. Rather than focusing only on technical indicators such as completeness rates or duplicate counts, organizations need to demonstrate how trusted master data contributes to real business performance.

Measure What Matters

The impact can be measured across four categories: risk reduction, revenue growth, cost savings and corporate governance. Designing this value measurement from the first release makes the contribution of MDM visible and easier to communicate to stakeholders.

APGAR Expertise: build an MDM program designed for lasting business value .

Go beyond technology

Successful MDM programs go beyond technology. They combine clear business objectives, effective governance, trusted data and ownership models that reflect how the organization actually works.

Build MDM for lasting value

APGAR supports organizations in designing and implementing MDM programs that connect business priorities with governance, data quality and measurable outcomes.

Planning an MDM program? 

Build a focused MDM roadmap, establish the right governance model and deliver measurable business value from the first release.