Tag System Construction Practice at 58: Pain Points, Solutions, Architecture, and Management Platform

Introduction – The article introduces a comprehensive tag‑system practice for 58, outlining four main parts: pain points and solutions, the tag architecture, the tag management platform, and a summary.

When to Build a Tag System – Tags become necessary during the growth and mature stages of a business when coarse‑grained operations no longer suffice and precise user segmentation is required for ROI improvement.

1. Pain Points and Solutions – Three stages present distinct issues: (a) start‑up stage suffers from data‑quality vs. application conflicts; (b) growth stage faces tension between permission control and operational flexibility; (c) mature stage deals with tag value versus cost, requiring evaluation and lifecycle control.

2. Tag‑System Architecture – A three‑layer design is proposed: Organizational Collaboration (big‑data team, business data team, operations), Platform Construction (tag modeling, metadata, permission, evaluation, recommendation), and Operations (use‑case case library). Emphasis shifts from efficiency in early stages to value‑growth and cost control in mature stages.

3. Tag‑System Construction Plan – The plan follows a "structure + scenario" approach, integrating data collection (events, logs, third‑party), storage & computation (ODS/DWD/DWS/ADS layers, user‑/customer‑/enterprise‑ID modeling), and platform capabilities (full lifecycle management, evaluation, recommendation).

4. Tag Lifecycle Management – Defines five actions: add, view, use, evaluate, and retire tags, and builds four core capabilities: production, metadata, marketplace, and security, with metrics for quality and business impact.

5. Tag Evaluation – Evaluates tags on three dimensions: usage (analysis, crowd‑selection, API calls), attention (views, favorites, permission requests), and effect (CTR, push performance). Scores are weighted to produce an overall value rating.

6. Tag Management Platform – Provides standardized naming, definition, classification, and update frequency; includes approval workflow, monitoring, a tag map for quick retrieval, and recommendation based on tag scores.

Conclusion – Building a tag system should start from concrete business scenarios, ensuring scalability, ease of use, and measurable value through a combined structured and scenario‑driven methodology.