Building a Unified Credit Card Metric System at ICBC: Design, Implementation, and Impact

1. Current Situation and Issues

The credit‑card data‑intelligence scenarios at ICBC have grown explosively, creating strong demand for flexible, self‑service data usage. Existing metric construction suffers from three main problems: (1) difficulty reusing metrics because processing logic is scattered across business scenarios, leading to tight coupling and maintenance challenges; (2) high human cost for building derived metrics, with manual development causing long delivery cycles and slow validation; (3) incomplete metric management across systems, manual identifier generation, and lack of a unified view, which raises operational difficulty and hampers asset lineage research.

2. Solution

Guided by a business‑centric architecture and the goals of sharing, reuse, and innovation, the team built a three‑layer data model (source layer, aggregation layer, extraction layer) on the data‑mid‑platform and created a unified credit‑card metric system. The solution includes:

Metric Standardization : Tagging and metric‑izing source‑granularity data to form a unified metric foundation, solving reuse problems.

Low‑Code Self‑Service Tool : A configurable tool that lets business users define derived metrics without coding. Compared with manual development, the response time shrank from a half‑month per version to three days, dramatically improving development efficiency and reducing labor.

Comprehensive Asset Management : Integration with a big‑data asset management system that provides full lifecycle management, unified access, detailed asset records, and lineage, lowering the usage barrier for business users.

The aggregation layer is “thickened” by pre‑placing common data‑processing logic, ensuring high cohesion of metric definitions and low coupling with business scenarios. This produces reusable aggregation tables that serve as templates for future metric generation.

3. Application Effects

After deployment, the system was first applied to risk‑control scenarios. By aggregating core indicators such as issuance, credit‑limit adjustments, and installment business, the team reduced the risk‑alert calculation cycle from T+7 to T+0 and cut the number of risky staff by nearly 50%. In the MOVA profit‑sharing scenario, the new credit‑card profit‑sharing metrics enabled automatic distribution of commission revenue to 16,000 outlets, improving accuracy and timeliness. The same approach was extended to issuance, installment, and limit‑adjustment domains, achieving “one‑time asset construction, multiple‑time reuse.” By September 2024, more than 30 aggregation tables covering over 3,000 credit‑card metrics had been rebuilt.

4. Future Outlook

The unified metric system now has a solid data foundation, but further expansion is planned to include more business entities, deepen content, and maintain reliable asset records. Continued focus on business pain points will empower credit‑card marketing, risk‑control, and product operations, supporting ICBC’s broader digital transformation.