Data Governance Construction Path and Practice by Dipu Technology

This article shares Dipu Technology's practice and summary on data governance.

1. Data Governance Construction Path

Business digitization aims to create an integrated flow of business, information, and data. From strategic planning to data collection, analysis, and monitoring, information loss occurs at each layer; data governance seeks to minimize or compensate for this loss, ensuring data accurately reflects business facts.

During IT planning, businesses first map business flows, which manifest as process architectures, value streams, or capability frameworks. Information flows (forms, certificates, documents) are identified, and data objects are extracted to guide digital system design.

Because each department (HR, procurement, production, R&D) has its own digital demands, disparate systems develop localized understandings of data, making it impossible to simply stitch together legacy data to form a true business data model.

Therefore, data governance must bridge business understanding and data design, ensuring data can faithfully represent the entire business.

Data governance covers the flow from business to information, data, and database tables. Even if source systems cannot be changed immediately, governance actions are applied at the ODS layer of the data warehouse to create a reliable data image.

The concept of a "data image" describes detailed business processes; a true data image enables accurate KPI analysis and aligns business, information, and data flows.

In summary, data governance consists of two core actions: business data governance (creating a true data image) and analysis system governance (designing rational analytical structures based on that image).

2. Typical Enterprise Digital Platform Framework

The framework includes four layers: (1) Business systems as data sources, (2) Data middle‑platform for aggregation and modeling (source, detail, summary, application layers), (3) Self‑service data consumption for analysts and developers, and (4) Intelligent decision‑making (dashboards, visual modeling, AI applications).

Effective governance requires mechanisms for data standards, models, quality, and processes to keep the entire chain smooth.

3. Core Understanding of Data Governance

Three pillars are highlighted: (1) Data governance architecture design (data standards, master data, governance mechanisms), (2) Deep business data governance (asset catalog, data models, standards, distribution, quality improvement), and (3) Analysis data system design (metric management, performance indicator design, data capability supply such as tags and algorithm models).

4. Data Governance Implementation Path

The path consists of two parts: Governance activities (starting with data inventory to build a comprehensive data‑asset map, prioritizing assets by value and pain points, then standardizing and improving quality) and external enablement (establishing governance teams, roles, processes, and templates to sustain ongoing governance).

5. Starting Point – Data Asset Inventory

Data assets are the foundation for standards, quality, and security. An example from a manufacturing company shows how to identify production‑domain objects, create L1‑L3 hierarchical classifications, and derive conceptual, logical, and physical data models.

6. Phase‑wise Deepening of Data Asset Definition

From the L3 business‑oriented structure, further attributes are added to form logical entities, which are then mapped to physical tables, balancing business detail with system performance considerations.

7. Data Asset Responsibility Assignment

Roles such as data owners, change approvers, and entry custodians are defined, often at the L3 level, with escalation to higher levels when cross‑domain responsibilities arise.

8. Platform Support for Data Governance

Dipu provides an end‑to‑end data intelligence platform that supports data integration, governance (standards, quality, security), and data sharing.

Case Studies

1) A food‑processing company leveraged a report‑driven governance approach, combining business data governance with metric system design to deliver both immediate business value and long‑term data foundation.

2) An automobile manufacturer began with a data‑asset inventory and governance framework, later extending to a comprehensive business‑wide data map that guides IT planning.

Q&A

Q1: Why is master data important? A: It provides a single, reliable view of critical entities across domains, supporting integration and distribution.

Q2: Does governance create a new database? A: Initially it yields standards; over time a clean data layer emerges as source systems adopt those standards.

Q3: What if standards cannot be applied to legacy systems? A: Use mapping for existing data and enforce standards on new or changed data.

Q4: How to quantify governance value? A: By measuring improvements in digital initiatives, KPI reporting speed, data reuse, and the contribution of governed assets to business applications.

Thank you for attending the session.