Ping An Life Insurance’s Data Middle Platform Construction Practice

At the 2023 Big Data Industry Development Conference, Zhu Sheng, General Manager of Ping An Life’s Data Management team, delivered a keynote titled “Practice of Data Middle Platform Construction at Ping An Life,” introducing the insurer’s approach to building a data‑centric middle platform.

Part 1 – Data Capability: The Foundation of Digital Transformation – In 2019 China recognized data as the fifth factor of production, prompting Ping An Life to launch a digital‑transformation project group and adopt a “4+3+1” strategy, where data drives product and channel upgrades. The company structures its data capability into three layers: data products, data foundation, and data governance.

Part 2 – Data Middle Platform: Architecture and Governance – The platform evolved from an Oracle‑based warehouse (2005) to a Hadoop‑based solution (2016) and finally to a full‑featured middle platform (2020‑present). Governance follows the DAMA framework, establishing policies, processes, and a real‑time data foundation (ODS, CDM, ADS layers). Data products are integrated to directly serve business needs, and maturity is measured via DCMM and CAICT assessments, technical metrics (latency, quality, reusability), and business KPIs such as NPS, active users, and claim‑automation rates.

Part 3 – Value Output: Business‑Scenario Applications – Four major use cases illustrate the platform’s impact: product innovation that shifts design from company‑centric to customer‑centric, digitalized operation management, a “Three‑Good Five‑Star” performance model, and rapid claim processing within minutes, all delivering measurable cost savings and service improvements.

Part 4 – Ongoing Operations: Talent and Culture – Success depends on skilled personnel; the team comprises data architects, analysts, and model engineers organized into seven groups. Ping An promotes a data‑culture through the “Data Elite Club,” encouraging external learning and knowledge sharing to sustain the platform’s evolution.

The presentation concludes by inviting further dialogue on data middle‑platform practices.