What Is Master Data Management?
Master data management (MDM) is the practice of creating a “golden record” for key business entities that is consistent across all systems. Master data includes the core reference data that describes:
- Customers: Who buys from you
- Products: What you sell
- Vendors/Suppliers: Who you buy from
- Employees: Who works for you
- Accounts: Your chart of accounts
- Locations: Where you operate
MDM ensures these entities are defined consistently everywhere.
Why Master Data Matters
The Problem: Inconsistent Master Data
Without MDM, the same entity exists differently across systems:
Customer “Acme Corp” appears as:
- “Acme Corporation” in CRM
- “ACME CORP” in ERP
- “Acme Corp.” in billing system
- “Acme” in spreadsheets
Consequences:
- Can’t calculate total customer revenue
- Duplicate records inflate customer count
- Marketing sends multiple communications
- Support can’t see complete history
- Analytics are unreliable
The Solution: Master Data Management
MDM creates one authoritative version:
- Single customer ID links all instances
- Consistent attributes across systems
- Changes propagate everywhere
- Analytics reflect true picture
Master Data vs. Transactional Data
| Master Data | Transactional Data |
|---|---|
| Describes entities | Records events |
| Changes slowly | Changes constantly |
| Shared across systems | Specific to processes |
| Requires governance | Volume-focused |
| Examples: Customer, Product | Examples: Orders, Payments |
Master data provides context for transactional data. An order (transaction) references a customer and products (master data).
MDM Architecture Approaches
Registry Style
Systems keep their own data; MDM provides cross-reference:
- Links records across systems
- Doesn’t store master data itself
- Lowest disruption to implement
- Limited ability to enforce standards
Consolidation Style
MDM aggregates data for analytics:
- Creates golden record for reporting
- Source systems unchanged
- Read-only master for analytics
- Doesn’t fix source quality
Coexistence Style
MDM and sources both maintain data:
- Changes can originate anywhere
- Synchronization between systems
- Balance of control and flexibility
- Complex to implement
Centralized Style
MDM is the authoritative source:
- All changes go through MDM
- Sources subscribe to master
- Strongest data quality
- Highest implementation effort
MDM Process
1. Data Profiling
Understand current state:
- What master data exists?
- Where does it live?
- What’s the quality?
- How do systems differ?
2. Data Matching
Identify same entities across systems:
- Exact matching (ID, email)
- Fuzzy matching (name similarity)
- Rule-based matching
- Machine learning matching
3. Data Merging
Create golden records:
- Survivorship rules (which source wins)
- Attribute-level decisions
- Conflict resolution
- Manual review for uncertain matches
4. Data Stewardship
Ongoing maintenance:
- New record creation
- Change management
- Exception handling
- Quality monitoring
5. Data Distribution
Share master data:
- Push to source systems
- API access for applications
- Reporting and analytics
- Integration with data flows
MDM Challenges
Organizational: Who owns customer data? Sales? Marketing? Finance?
Technical: How to match records reliably across systems?
Process: How to handle ongoing changes and exceptions?
Quality: How to clean up years of accumulated duplicates?
Adoption: How to get systems to use master data?
How Go Fig Addresses Master Data
Go Fig helps with master data challenges:
Cross-system matching: Identify same entities across connected systems
Unified view: See consolidated master data in one place
Semantic layer: Define consistent entity attributes
Data quality alerts: Flag master data issues automatically
Excel integration: Work with master data in familiar tools
While not a full MDM platform, Go Fig provides practical master data capabilities for finance teams who need consistent customers, vendors, and accounts for reporting.
MDM Best Practices
- Start with high-value entities: Focus on customers or products first
- Define clear ownership: Single owner per data domain
- Establish governance early: Rules for creation and changes
- Invest in matching: Quality matching prevents duplicates
- Plan for exceptions: Not everything matches automatically
- Measure quality: Track duplicate rates and accuracy
- Build incrementally: Don’t try to boil the ocean