Understanding Data Governance: Definitions, Layered Requirements, and Implementation Strategies

#01 What Is Data Governance

According to the Data Management Association (DAMA), data governance is a set of activities that exercise authority and control over data asset management.

The Data Governance Institute (DGI) defines it as a system that, through a series of information‑related processes, establishes decision‑making authority and responsibility distribution based on a consensus model describing who can act, with what information, when, where, how, and what action.

IBM defines data governance as the use of policies and standards to improve data availability, quality, and security, identifying data owners, security measures, and intended uses, ultimately maintaining safe, accessible, high‑quality data for deeper business insights.

Different organizations have varied interpretations and goals, but common objectives include:

Improving data quality, stability, and security through technical means.

Increasing data asset usage efficiency and reducing costs via standards and asset creation.

Enhancing business value through data mining, boosting core competitiveness and impact.

#02 Requirement Layering of Data Governance

Maslow’s five levels—physiological, safety, social, esteem, self‑actualization—are mapped to data governance as:

Stability, Security, Usability, Quality, Cost‑Value

Higher layers deliver greater value, but they depend on the foundational layers below.

#03 Stability Requirement

Data stability means data can be produced consistently and promptly, akin to Maslow’s physiological need.

Stability is measured by timeliness and reliability, often expressed as “X nines” of availability (e.g., 3‑9 = 99.9% uptime, 4‑9 = 99.99%, 5‑9 = 99.999%). In batch‑oriented data pipelines, reliability may be gauged by the frequency of job failures.

Latency is described using transaction‑date offsets (T+0, T+1, T+2, …) or hour‑based offsets (H+0, H+1) for near‑real‑time processing.

#04 Security Requirement

Data security covers permission management, sensitive data protection, and compliance, corresponding to Maslow’s safety need.

Security includes preventing data leakage and ensuring lawful, compliant usage, driven by regulations such as GDPR, China’s Data Security Law, and Personal Information Protection Law.

Key security controls:

Multi‑tenant isolation for compute and storage resources.

Role‑based access control for system accounts.

Network segmentation between internal/external systems.

Compliance requirements involve data retention periods and data masking techniques (redaction, static encryption, dynamic encryption) for personal and financial information.

#05 Usability Requirement

Usability means data is easy to query, understand, and standardize, analogous to Maslow’s social need.

Data Query: Build BI/OLAP platforms using engines like Kylin, Druid, ClickHouse, Doris, StarRocks, and visualization tools such as Superset, Grafana, DaVinci. Common metrics include TP90/TP95/TP99 response times and QPS.

Data Standards: Business standards (e.g., CTR, ROI calculations), technical standards (type, length, format, encoding, naming), and management standards (access procedures, deletion, onboarding).

Data Models: Evaluate reusability, coupling, and stability of data models to reduce maintenance cost.

Addressing these three aspects yields high usability and visible governance outcomes for end users.

#06 Quality Requirement

Quality focuses on accuracy, completeness, consistency, and validity—Maslow’s esteem level.

Achieving quality relies on data monitoring and scheduling, complemented by manual processes and governance policies.

Accuracy monitoring: Verify data ingestion standards and detect errors during processing.

Completeness monitoring.

Consistency monitoring: Compare results across data sources.

Validity monitoring: Generate quality reports via algorithms and assess improvements over time.

#07 Cost‑Value Requirement

Cost‑value addresses the economics of data production and the business value generated, reflecting Maslow’s self‑actualization need.

Cost Quantification: Use data lineage, metadata management, and log analysis to visualize job expenses.

Value Quantification: Measure governance impact on cost savings, ROI improvements, and risk reduction.

Cost Optimization: Identify low‑value, high‑cost jobs for throttling, schedule them during off‑peak periods, or migrate data to cheaper storage; prioritize high‑value tasks for optimization.

#08 Summary

The layered requirements of data governance are interdependent and iterative; higher layers bring more business impact but rely on solid foundations below.

Successful governance requires not only technical tools but also organized planning, clear processes, and institutional support.

Organizational Support: Unified governance goals, dedicated teams with authority, and enforceable policies.

Process Standards: Defined data ingestion, output, and standardization workflows.

Technical Tools: Robust platforms that enable end‑to‑end governance execution.

By aligning solutions with each requirement layer and evaluating ROI, organizations can make informed investments in data governance.