Xianyu User Growth Task System Architecture and Implementation

In recent years, user growth has become a key focus for internet companies. Xianyu’s growth team abstracts diverse growth playbooks into a unified "user task" model, separating user quantity and quality concerns.

The system defines tasks with five dimensions: who (target audience), how (conditions and timing), where (trigger events), what (rewards), and extend (additional capabilities). By encapsulating common logic into a task lifecycle, developers can customize only the nodes specific to each playbook.

The task lifecycle includes management, delivery, tracking, settlement, and reward modules. Management handles metadata and instance state; delivery selects target users; tracking listens to business events and updates progress; settlement monitors completion and notifies reward nodes; reward validates inventory and issues benefits.

For the "Browse to Earn" playbook, tasks are cyclical (7‑day period) with daily browsing goals and special rewards on days 2, 5, and 7. The architecture adds three domain services: period domain (configurable cycles and special nodes, using MVEL expressions), benefit domain (stock‑checked reward issuance), and schedule domain (playbook timing).

Implementation follows a layered design with strategy patterns for period types (rolling, sliding) and benefit types (points, coupons). The task flow follows a template pattern with nodes such as start, join, process, complete, close, and query, allowing playbook‑specific extensions while keeping core logic stable.

PS：基础产品层为闲鱼业务系统，底层依赖层为阿里内部封装中间件服务，在阿里云或开源基本都有对应产品。

The overall system enables rapid iteration of growth experiments, improves development efficiency, and provides a reusable framework for future user‑engagement features.