Establishing the Foundation Layer

Based on individual customer data, the Foundation Layer is built to land source system data into the data mart. It extracts source data from the big data platform and relevant data marts, reorganizes it by thematic domains, and performs extraction, cleansing, consolidation, and aggregation, making it ready for downstream analysis.

Consolidating Individual Customer Information

The system consolidates information on individual customers' agreements/assets, products, transactions, channel agreements, and marketing activities, providing strong data support for customer segmentation, customer scoring, marketing analysis, sales analysis, and performance evaluation.

Application Layer Aligned with Analytical Needs

The Application Layer is closely aligned with the analytical requirements of each application system. It retrieves data from the Foundation or Aggregation Layers, defines application-specific metrics and unique indicators, and extracts common data access and statistical needs such as customer tags, customer scoring, customer 360-degree views, and wide tables for modeling.

Built on the big data platform, the Customer Management Data Mart System follows a customer-centric approach with a three-layer architecture comprising the Foundation Layer, Aggregation Layer, and Application Layer. It fully integrates individual customer information, transitioning the bank’s data architecture from account- or transaction-centered to customer-centered.

To enhance the timeliness of processing massive customer data volumes, the project adopts a hybrid Hadoop + Spark architecture. It integrates data from 39 source business systems across the bank, constructs seven major thematic domains—Customer, Agreement, Marketing, Product, Transaction, Public, and others—and establishes 133 models encompassing over 3,500 data items.