6.4 Reference Data Governance & Standards

Reference Data Governance and Standards

Reference data provides the classifications, code lists, and lookup values that establish context across systems. Although reference data typically has a much smaller footprint than master data, its impact is enterprise-wide. A single incorrect code or an unmapped reference table can halt transaction flows, corrupt financial dashboards, or lead to regulatory compliance failures. Therefore, Reference Data Governance and adherence to international standards are critical disciplines within the DAMA framework.

---

Global and Industry Reference Data Standards

To promote interoperability and simplify integration, organizations should adopt international standards for common reference data domains. The CDMP exam heavily tests familiarity with standard ISO codes and industry classifications:

1. ISO 3166 (Country Codes)

This standard defines codes for the names of countries, dependencies, and special areas of geographical interest:

• ISO 3166-1 Alpha-2: Two-letter codes (e.g., US for United States, CA for Canada, GB for United Kingdom). This is widely used for internet top-level domains.
• ISO 3166-1 Alpha-3: Three-letter codes (e.g., USA, CAN, GBR), which are more readable and help avoid ambiguity.
• ISO 3166-1 Numeric: Three-digit codes (e.g., 840 for United States, 124 for Canada, 826 for United Kingdom). These are useful in systems that do not support Latin alphabets.

2. ISO 4217 (Currency Codes)

This standard defines three-letter codes for currencies (e.g., USD for US Dollar, EUR for Euro, JPY for Japanese Yen). The first two letters represent the ISO 3166 country code, and the third letter represents the currency name (e.g., US + D for Dollar).

3. ISO 5218 (Representation of Human Sexes)

This standard defines representation codes for human sexes:

• 0 = Not known
• 1 = Male
• 2 = Female
• 9 = Not applicable

4. ISO 639 (Language Codes)

This standard classifies languages using two-letter (ISO 639-1) or three-letter (ISO 639-2/3) codes (e.g., en or eng for English, es or spa for Spanish).

5. Industry-Specific Reference Codes

• NAICS and SIC: Used by government and businesses to classify industry establishments based on the primary economic activity (e.g., manufacturing, retail, services).
• ICD-10/ICD-11: International Classification of Diseases, maintained by the World Health Organization (WHO), used globally for medical diagnostics and billing.
• UNSPSC: Universal Standard Products and Services Classification, a hierarchical framework used for classifying products and services in procurement and spend analysis.

---

Reference Data Governance Lifecycle

Reference data requires formal governance policies to manage changes. Unlike transactional data, reference data changes are structural and affect database constraints, integration pipelines, and historical reporting.

Change Management and Approval

Any request to add, modify, or retire a reference code must go through a formal change request process. The request is evaluated by data stewards to assess the downstream impact. For instance, if an organization decides to retire an old product category code, the stewards must identify all active products using that code and map them to the new category before the change is approved.

The Principle of Soft Deletes

A core rule of reference data management is that codes should never be deleted from the database. If a code is deleted, historical transactions referencing that code will violate referential integrity constraints, leading to database errors or orphaned records. Instead, reference tables must implement a soft delete pattern. This involves using metadata attributes such as IsActive (Boolean flag) or EffectiveStartDate and EffectiveEndDate columns. When a code is retired, its EffectiveEndDate is set to the current date, and IsActive is set to false. This prevents new transactions from using the code while preserving historical records for reporting.

---

Mapping and Translation (Cross-walks)

In heterogeneous IT landscapes, different systems often use different codes to represent the same concept. For example, System A might use M and F for gender, System B might use 1 and 2, and System C might use Male and Female. To enable data integration and interoperability, organizations build Cross-walk Tables (or mapping tables) that translate local system codes to a standard enterprise reference code or an international standard.

Managing cross-walks requires active governance. Whenever a source system updates its local codes, the mapping tables must be updated in tandem to prevent integration failures.

Reference Data Distribution Architectures

To ensure consistency across the enterprise, reference data should be managed centrally in a Reference Data Management (RDM) system and distributed to subscribing applications. The two primary distribution patterns are:

• Push Architecture: The central RDM system pushes reference data updates to downstream applications via database replication, file transfer, or enterprise service buses. This ensures that downstream applications have local copies of reference tables, minimizing read latency.
• Pull Architecture: Consuming applications query the central RDM system in real-time via REST APIs or web services whenever they need to validate a code or display a dropdown list. This guarantees that systems always use the most up-to-date codes, though it introduces a network dependency.

---

DAMA Exam Traps and Study Tips

1. ISO Standards Memory: You must memorize the specific ISO standard numbers. The exam frequently asks which standard defines country codes (ISO 3166), currency codes (ISO 4217), representation of human sexes (ISO 5218), and language codes (ISO 639).
2. Deleting Reference Data: Watch out for questions asking how to handle retired reference codes. The correct answer is always to retire the code using start/end dates or status flags, never to perform a hard delete.
3. Ownership of Reference Data: Reference data is not owned by IT. Business data stewards and governance committees own the definitions and mappings, while IT is responsible for the distribution and technical implementation.