The MDM governance framework I built at Canada Life is running through its fiftieth sprint. The QA reduction it enabled has been confirmed at 50%. Neither of those outcomes was primarily a technology achievement. They were governance achievements.
Every MDM failure I have encountered in Canadian financial services started the same way. The organization treated data quality as a technical problem, assigned it to an IT team, and waited for the data to get better. It did not get better. Data quality is not a technical problem. It is a people problem with a technical component.
Master data management is the practice of ensuring that an organization's critical data is consistent, accurate, and accountable. In insurance, critical master data includes policy records, client identity data, product definitions, and the reference data that drives pricing and claims.
When master data is inconsistent, the downstream effects are significant. A client who exists as two different records in two different systems generates duplicate correspondence, incorrect aggregation of exposure, potential regulatory reporting errors, and policy administration failures. These are not IT problems. They are business problems that originate in how people create, maintain, and transfer data.
The instinct in most organizations is to select an MDM platform and then figure out the governance. This is backwards. The platform can enforce rules, but it cannot define them. It can flag duplicates, but it cannot decide what constitutes a duplicate for your specific data model. It can route records for review, but it cannot define who is accountable for approving what.
Those are governance decisions. They have to be made by people who understand the business context: what does a unique client record mean in the context of group benefits? What level of address variance is acceptable before two records should be merged? What is the process for challenging a merge decision that a business user believes was wrong?
Before we selected any technology at Canada Life, we spent eight weeks defining these rules with business stakeholders. Eight weeks of workshops that many people thought was too long. It was not too long. It was the work that made everything else possible.
An MDM platform without governance rules is an expensive data consolidation tool. The rules have to come from the business. The technology enforces them, it does not create them.
The governance framework at Canada Life assigned three kinds of accountability that most MDM programs leave undefined.
Data stewardship: who is responsible for the accuracy of specific data domains? Not the IT team. The business owner of the data. In insurance, that means the team that owns the policy administration function for client identity data, the actuarial function for product reference data, and the compliance function for regulatory reference data.
Issue resolution: when a data quality issue is identified, who decides what the correct value is? The process for this has to exist before the platform goes live. If it does not, every disputed record becomes an escalation that consumes senior time and delays resolution.
Rule maintenance: data governance rules become outdated as the business changes. Who is responsible for reviewing and updating them? In most organizations, no one is explicitly accountable for this, which is how rules that made sense two years ago continue to be enforced on data that no longer matches the context the rules were designed for.
The program at Canada Life worked because of three things.
First, the business case was built on confirmed outcomes from earlier data migration waves, not on projected efficiencies. We had evidence that data quality problems were causing specific, measurable costs. That evidence made the governance investment an easy decision.
Second, the data stewards were senior enough to make decisions. Not junior analysts who had to escalate everything. People with enough business authority to say "this is the correct value" and have that decision stick.
Third, the governance model was kept simple. One page. Clear accountability. A defined process for every type of issue. Organizations that build complex governance models build models that no one uses. Simplicity is a design principle, not a shortcut.
The framework is on its fiftieth sprint. The QA reduction is holding. The reason it is holding is not the technology. It is the clarity of accountability and the quality of the governance rules that the business stakeholders defined before any platform was selected.