Tencent Data Governance Practices and the WeData Platform

Introduction – The talk presents the current stage of Tencent's data governance, practical experience, and the WeData data‑governance platform.

Data Governance Challenges

Management challenges: data scattered across many business units and reporting locations, making unified management difficult.

Technical challenges: ensuring data quality after collection.

Business challenges: missing metadata prevents unified auditing and measurement.

Data Governance "Maslow" Hierarchy – Issues are classified by timeliness, quality, usability, security, and cost.

Internal Practices – Business Landscape – Tencent operates dozens of business groups (BGs) with thousands of product lines, EB‑scale data storage, and thousands of analysts.

Three Stages of Data Governance

Stage 1: Data assetization – turning data into valuable assets.

Stage 2: Cost reduction and efficiency – lowering resource consumption while maintaining asset value.

Stage 3: Platformization – abstracting common practices into a data‑governance platform.

Practice 1 – News Data Assetization – Faced with inconsistent data standards and low data reuse, Tencent News unified data models, upgraded warehouses, and built 250 models, 52 dimension tables, and 270 application tables, achieving >95% completeness, 73% reuse, and <5% cross‑layer references.

Practice 2 – PCG Data Cost Governance – Defined cost scope (collection, generation, analysis, application platforms) and reduced both resource usage and unit cost, cutting absolute big‑data cost by at least 10% despite a 30%+ month‑over‑month cost increase.

Practice 3 – Platformization of Governance – Four phases: overview, asset details, governance solutions, and execution; a scoring system evaluates conformity, security, quality, cost, and usage.

WeData Platform Architecture – The platform consists of two parts: agile data production (modeling, integration, development, services) and efficient data governance (asset, quality, security, metadata).

WeData Services – Data flow/storage layer and processing layer (aggregation, development, operations, API services).

Governance Tools

Standardization tools: metric, dimension, and metadata management; warehouse planning and physical materialization.

Quality tools: rule definition, real‑time ETL monitoring, offline checks, alerting, and periodic quality reports.

Security tools: sensitive data identification, privacy protection (static/dynamic masking, encryption, watermarking), and audit of data access/export.

Metadata asset tools: discovery, technical and business metadata linkage, data catalog, lineage, change tracking, and data temperature.

Metadata Management – Improves search, understanding, and application of data by providing global discovery, usage metrics, and clear business context.

Governance Outcomes – Achieved cost reduction, increased data reuse, and established an enterprise‑wide rating system to prune cold or duplicate data.

Q&A

Metadata security tags are independent and not propagated downstream.

Business metadata includes normative, quality, security, cost, and usage dimensions.

TBDS is the underlying data‑processing engine; WeData is the governance layer built on top of it.

Thank you for attending; follow DataFunTalk for more content.