Ping An Life Insurance

Ping An Life Insurance, one of the world's largest retail financial services groups with 227 million retail customers and RMB 10.1 trillion in total assets, faced a structural technology crisis common to large insurers: hundreds of legacy systems supporting a sprawling set of front-end applications, each embedding its own redundant business logic. In the construction channel, micro-services lacked clear business boundaries, preventing meaningful integration and data sharing. Multiple platforms duplicated the same functions, creating user confusion and blocking data unification. This fragmentation slowed innovation, increased IT maintenance costs, and left the insurer unable to align its technology estate with evolving sales and distribution strategies.

In 2021, Ping An Life partnered with Accenture to introduce Domain-Driven Design (DDD) methodology into its construction channel. Accenture contributed domain modelling expertise and business process insight, enabling domain models to be systematically abstracted from real operational knowledge rather than inherited from legacy system boundaries. The channel's middle-layer platform was horizontally sub-divided into three layers: core agent capability domains, core insurance process domains, and supporting domains. A dedicated 'Strategy Center' generic domain consolidated AI capabilities — including NLP, NLU, and intelligent assistance — so that a single implementation could serve multiple scenarios such as Human-Machine Training and Online Meetings. Multi-layer reuse was achieved by designing aggregations, entities, and domain services within shared boundaries. Business and IT specialists were co-located under specific domain teams to accelerate alignment.

The DDD-based platform integration delivered measurable efficiency gains across development and operations:

• 30% reduction in development time — achieved by modularising functions into composable, Lego-like components that support rapid assembly without heavy custom coding
• 25% service re-use rate — generic domain capabilities, including NLP and planning functions, are now shared across multiple business scenarios rather than rebuilt per application

Beyond the headline numbers, redundant platform functions were consolidated, reducing IT maintenance overhead and eliminating user-facing duplication. Applications built under the DDD framework now directly support channel sales and management workflows, including digital recruitment and agent training, delivering end-to-end productivity gains across the sales organisation.

• Domain boundaries must reflect business reality, not system history — legacy micro-service boundaries that mirror old technology structures will reproduce the same fragmentation problems at a new level of abstraction.
• Centralising AI capabilities in a shared generic domain multiplies their ROI — NLP and NLU integrated once into a Strategy Center layer serve every downstream business scenario without redundant development cycles.
• Business–IT co-location under domain teams is an enabler, not an afterthought — without embedded business expertise, domain modelling produces technically coherent but operationally irrelevant boundaries.
• Modular, low-code assembly is the practical payoff of DDD — the 30% development time reduction came not from the architecture itself but from the composable components it enabled.