JD Tech’s Event‑Tracking Data Governance and One‑Stop Platform: Practices and Innovations

As online traffic growth plateaus, companies increasingly treat data as a core asset. Event‑tracking data, one of the two most important internal data sources, drives product optimization, recommendation, and advertising. However, evolving business requirements cause stale or inaccurate tracking data, leading to quality, cost, efficiency, security, and standardization problems.

Why governance is needed – Data quality issues such as timeliness, accuracy, and consistency affect data‑warehouse reliability and metric logic. Rapid data growth inflates infrastructure costs. Manual, blind efforts reduce development efficiency, and inconsistent standards across teams hinder data integration.

Governance methodology – The team established a full‑link standard covering definition, collection, verification, metric definition, and lifecycle management, documented in the "Tracking Process Specification". A governance chain was built, and a dedicated "Pangu" project and data‑management committee were created to oversee each stage.

Platform construction – The "Qidian" one‑stop tracking platform provides unified metadata management, simple query interfaces, one‑click validation, and visual data‑quality monitoring. It supports code, visual, and no‑code tracking methods, and combines multiple tracking techniques (frontend code, server‑side, visual) for flexible scenario use.

Validation tool – Previously, testers needed database access and manual SQL queries. The new tool allows QR‑code scanning for real‑time data, one‑click compliance checks, and uses Lua scripts for high‑concurrency processing. Data is stored in Elasticsearch rather than MySQL, improving capacity and query speed.

Monitoring – Automated checks track upload success rate, cache rate, loss rate, retention, and storage rates. Threshold breaches trigger email alerts to developers, reducing data loss and manual effort.

Visualization – A visual inspection tool loads a plugin via the Qidian JS SDK only when the visualization UI is opened, showing element‑level tracking IDs, real‑time logs, and aggregated metrics.

H5‑native data unification – The platform aligns page visits, visit sequences, source events, first‑visit events, previous‑page URLs/CTPs, and device identifiers across H5 and native app pages, enabling continuous user journey analysis and accurate attribution.

Page‑ID scheme – To replace unstable URLs, a pageId is generated based on configurable dynamic‑segment rules. The client matches the current URL to the best pageId before reporting, improving metric correctness. Fallbacks include retry mechanisms, local caching, and monitoring of configuration updates.

Future direction – Continue refining visual tracking components, integrate tracking lifecycle tools into a unified workflow, and add intelligent analysis and prediction to the real‑time dashboard, further supporting JD’s digital‑operating capabilities.