Ping An Financial Services' Big Data Platform Construction and Data Governance Practices

Ping An Financial Services, a subsidiary of Ping An Group, serves over 236 million individual customers and 185 enterprise clients, but rapid business expansion exposed limitations in its fragmented, siloed data architecture, leading to low data integration efficiency and high operational costs.

The company identified five core data challenges: fragmented architecture, slow data collection, limited processing capabilities, inefficient data usage, and dispersed data management. To overcome these, Ping An defined a four‑stage technical evolution: awareness formation, free development with commercial BI tools, unified platform construction, and intelligent upgrade toward a smart data middle‑platform.

During the unified platform stage, the company integrated multiple BI tools, standardized data models, and introduced a unified data standard system, improving development speed by 50%.

Ping An then built a modular data‑tool stack covering four layers—data collection (收数), data processing (整数), data usage (用数), and data governance (管数). The collection layer introduced a unified ETL engine with quality checks; the processing layer migrated from a PG‑based stream platform to StarRocks for high‑performance analytics; the usage layer provided tailored tools for business users, digital specialists, analysts, and engineers; the governance layer unified data standards, quality control, security, and role‑based access.

Data governance was structured around four pillars—data intake, standardization, quality, and data integration—implemented through standardized data models, metadata management, quality monitoring platforms, and a master‑data management system. Governance initiatives were driven by cross‑department collaboration, training, and alignment with business objectives, resulting in improved data discoverability, accuracy, and support for advanced analytics and AI applications.