Building Data Products and a Data Middle Platform at NetEase Yanxuan: Practices and Lessons

In June 2022, NetEase Yanxuan data expert Gu Ping shared the company's journey of constructing a comprehensive data product system and data middle platform, highlighting challenges and solutions from 2017 to 2020 when the platform evolved from zero to a mature state.

Yanxuan operates a "platform + brand" dual model, generating data across the entire supply chain. Four data products—Product Data Operations Platform, Marketing Data Operations Platform, Mobile Data Workbench, and Supply Chain Data Operations Platform—drive business across the full chain, with the data middle platform serving as the essential engine supporting these products.

The company defines a data product as a decision‑oriented tool that not only visualizes data but also provides actionable recommendations, such as monitoring diagnostics that can directly trigger business actions like generating purchase orders.

Each business unit (Product, Marketing, Supply Chain) has a dedicated data product: the "Fuxi" marketing platform, the "Heluo" supply‑chain platform, the "Damai" product platform, and a mobile data workbench for managers, which is built as an H5 overlay within the Yanxuan app and offers KPI dashboards, traffic maps, and real‑time monitoring.

The supply‑chain platform focuses on demand‑driven supply, sharing sales and demand plans with suppliers to enable proactive production scheduling, while the mobile workbench is heavily used by the CEO, illustrating top‑down data adoption.

Yanxuan's data middle platform provides high‑efficiency, high‑quality data services that power the data products and business systems. Its architecture includes a data application layer (front‑end products and BI), a green data‑service layer, and a blue data‑warehouse layer built on NetEase YouShu.

Efficient analysis is achieved through the YouShu BI tool, which allows analysts to create reports without coding and enables rapid iteration; over 900 users generate more than 80,000 charts daily.

A unified query service (model‑as‑a‑service) abstracts data models behind a single API, ensuring consistent indicator definitions across all products and systems, with over 70 systems consuming the service.

Data governance is embedded throughout the lifecycle, from requirement review to model design, development, testing, and operation, with KPIs such as cross‑layer dependency rate, baseline completion rate, and average response time guiding continuous improvement.

In summary, NetEase YouShu's BI and big‑data development platform enable rapid construction of a data middle platform, providing unified data services that support both data products and business systems, while rigorous governance and performance metrics ensure reliability and scalability.