01
One customer, several versions of the truth
In many companies the same customer lives in several systems. CRM records it when sales opens the opportunity, ERP when the first order is created, billing when the invoice is issued and analytics when reports are consolidated. Each system captures data with its own rules and at its own time
Over time small differences appear that nobody decided: a tax ID without suffix, an abbreviated legal name, an address updated in one place only, a status that changed in billing but not in CRM
The consequences are operational, not technical. Duplicate customers are created, invoices go to old addresses, checks are repeated, reports from different teams disagree and every meeting starts by arguing which number is right. The costliest loss is trust: when nobody trusts the data, each team builds its own version
02
Not all data plays the same role
Governing data starts by telling it apart. The data describing a customer is not managed like the data recording a sale. This classification helps decide where rules, owners and controls belong
An error in master data multiplies: every order, invoice and report using it inherits the problem. That is why it is usually the first focus of a governance programme
03
Governing data is more than documenting rules
Data governance is the set of decisions, roles, policies, standards and controls that defines how an organisation's data is created, maintained and used. It connects what the company decides with what actually happens in operations
In practice two ways of exercising it coexist. Compare how a customer is created in each
Both approaches complement each other: passive gives visibility and active prevents. They are useful ways of explaining the topic, not universal normative categories. A mature programme usually combines catalogue and monitoring with controls where data is born or changes
04
Data quality needs owners
When data belongs to everyone, in practice it belongs to no one. Governance roles exist so every decision about data has someone who makes it and someone who carries it out
Indicative matrix. Actual distribution depends on each organisation's size and structure
05
From scattered data to the Golden Record
Master Data Management (MDM) is the discipline and technology that consolidates scattered master data into a trusted version and maintains it over time. If data governance defines the rules of the game, MDM is one of the mechanisms that enforces them on master data
Select each stage to see what happens to the example customer
Concepts worth knowing
- Duplicates
- Several records representing the same real entity, even if written differently
- Matching y merging
- Matching identifies two records as the same entity; merging joins them into one
- Survivorship rules
- Criteria to decide which value stays in each attribute when sources differ
- System of record and authoritative source
- The system where data is officially created and the source recognised as valid for each attribute
- Unique identifier
- A master key that recognises the entity across all systems
- Lineage and traceability
- Knowing where each value came from, how it was transformed and who changed it
- Data quality
- How complete, accurate, current, consistent and fit for purpose the data is
A Golden Record is not a table without duplicates. It is the governed master representation of an entity, sustained by rules, owners and traceability
06
Trusted data must stay trusted
Consolidating once is not enough. Every day customers are created, addresses change and commercial terms are modified. If those changes do not follow a governed flow, the Golden Record degrades within months
Walk through the customer creation and change flow and see who steps in at each point
The combination is the key: automatic rules handle the repetitive, business owners decide what needs judgement and quality controls confirm the system keeps working
07
How governed is your master data?
Think of one specific domain, such as customers or suppliers, and answer six questions. No answer is stored or sent
