How the Securities Industry Turns Data Governance into a Service‑Oriented Advantage

Background of Securities Data Governance Service

The securities industry faces data‑governance pressures from external regulations, such as the Regulations on Information Technology Management for Securities and Fund‑Operating Institutions , which require robust data‑governance architectures to ensure data security and quality, and from internal needs for business analysis, marketing, risk management, and investment analysis that rely on high‑value data assets.

Vision and Goals

The overarching vision is to continuously improve data governance to protect and create value from both internal and external data assets.

Key Pain Points

Frequent data issues causing mistrust and affecting performance assessment, customer experience, and marketing.

Low participation from business units, with fragmented data‑issue management and lack of sustained online mechanisms.

Inconsistent data models due to numerous legacy and purchased systems that cannot be easily modified.

Under‑utilization of data‑asset platforms, as business users rely on traditional reporting dashboards instead of the data‑asset portal.

Service‑Oriented Data Governance Practice

1. Data‑Governance Construction Framework

A top‑level data‑governance group and working groups were established. Since 2021, the company has issued a 14‑year data‑governance strategy, data‑management policies, standards, quality, security, lifecycle, and metadata guidelines.

The work focuses on six domains: data standards, metadata, data architecture (early focus), and data security, data quality, data modeling (since 2021).

Three platforms support these domains: a data‑security operation platform, a data‑control platform (integrating metadata, quality, standards, asset portal, and architecture), and a data‑model platform linked with DevOps and JIRA for online model design and management.

2. Data‑Governance Service Thinking

Previously, many governance tasks were formalistic and lacked business or IT adoption. The new approach collects real‑world data problems from asset‑management, risk‑management, development, and testing teams, analyzes requirements, and designs solutions that bridge governance, application, and security platforms.

Two guiding principles are adopted: moving from back‑office to business‑oriented services, and shifting from pure control to service‑oriented thinking.

3. Data‑Model Management

Model governance combines institutional policies and tooling. Standards and checks enforce uniform character sets (UTF‑8) and other rules across MySQL, SQL Server, and other databases. The model tool integrates with the control platform to verify production models against design specifications, flagging non‑compliant changes.

Key functions include model design (conceptual, logical, physical with standard references), model management (merging, root‑word library, version control), and model analysis (comparison with production, tagging, rule checks).

4. Data‑Model Service Practice

Standard libraries are linked to model tools, allowing developers to reuse established data‑standard definitions (e.g., account codes) across downstream warehouses and reports, ensuring consistency.

5. Data‑Quality Framework

Data quality is addressed at system and organization levels. System‑level management resolves issues directly with developers, while organization‑level management quantifies issues, tracks trends, and implements preventive mechanisms.

The quality‑management platform provides problem creation, classification, work‑order handling, monitoring (basic data, indicator data, table‑structure, cross‑system, and task monitoring), dashboards, scoring, and reporting.

6. Data‑Quality Assurance Mechanism

Pre‑emptive processes define problem intake, analysis, solution design, remediation, and post‑implementation review, supported by alerts (email, SMS) and severity grading.

7. Data‑Standard Framework

Based on the industry model DG‑SDOM, eight thematic data‑standard groups were built, covering brokerage, securities‑fund, risk management, and regulatory reporting.

8. Data‑Standard Service Practice

Standards are exposed through the data‑control platform to BI, reporting, and smart‑BI tools, allowing users to view metric definitions, source systems, and responsible teams directly in report headers.

9. Metadata Framework

The metadata system collects metadata from business systems, warehouses, and data marts, providing version control, parsing, and pre‑release management. It supports impact analysis, hot‑cold analysis, cost governance, cross‑environment comparison, and data‑map applications.

10. Metadata Service Practice

Weekly comparisons between production and test environments (e.g., 500 production tables vs. 490 test tables) detect structural inconsistencies, prompting developers to align schemas before release.

11. Data‑Security Framework

Since 2021, a security governance framework centers on personal client data, covering data rights, risk assessment, and classification/leveling, with organizational, technical, and operational layers.

12. Data‑Security Service Practice

A security‑operation platform implements classification, risk monitoring, dynamic/static masking, and visual dashboards to showcase sensitive data and risk status.

13. Value of Service‑Oriented Data Governance

Improved data consistency and accuracy through unified standards.

Enhanced compliance for regulatory reporting (over 10,000 metric definitions).

Data‑asset valuation via classification, metadata, standards, and models.

Increased business efficiency by streamlining data‑asset access and demand evaluation.

Summary and Outlook

The organization now possesses a comprehensive, normalized data‑governance capability and moves toward DataOps, aiming to standardize data‑governance processes, quantify data‑asset value, and contribute to industry‑wide standardization initiatives.

Q&A

Q1: Are business units involved in data‑model design, and how is consistency ensured across lines?

A1: Business participates in logical‑model reviews; the model tool can directly reference the company‑wide data‑standard library (e.g., 10,000+ metric definitions), ensuring consistent field definitions across systems.