Data Governance Practices and Logical Closed‑Loop at KuaiKan

Speaker : Qu Shichao, Head of Data Development at KuaiKan.

Introduction : KuaiKan, founded in 2014, evolved from a comic platform to a Z‑generation interest community with over 340 million users and 50 million monthly active users, leading to rapid data volume expansion and governance challenges.

Governance Background : The fast‑growing business lines caused data construction pressure, scattered data sources, low data quality, and inefficient development due to lack of standardized modeling.

Three‑Step Governance Path :

Step 1 – Break through a single core business line with the most severe data quality issues.

Step 2 – Consolidate governance strategies from the first line and migrate them across other lines.

Step 3 – Replicate the MVP solution to other businesses, reducing redundant effort.

Logical Closed‑Loop : The framework consists of three parts – business scope management, data asset governance, and application feedback – forming a continuous improvement cycle.

Business Scope Management : Track business feature changes, manage indicator priorities, and maintain a knowledge‑base of process and data source changes to keep data teams aligned.

Data Asset Management : Establish governance standards, develop platform tools (e.g., metadata, lineage, quality monitoring), and enforce end‑to‑end data pipeline norms from source to warehouse.

Collaboration Techniques : Align data product analysts, developers, and business owners, start with high‑impact core services, define MVP processes, and set unified output standards with regular retrospectives.

Summary & Outlook :

Effectiveness is measured by reduced indicator duplication, higher warehouse data reuse, and shorter development cycles.

Current limitations include slow tool iteration due to manpower and incomplete cross‑domain governance.

Future plans focus on incremental improvements, expanding lineage management, and strengthening data quality at the source.

Q&A : The data lineage is stored in MySQL, handling a few thousand daily tasks, with no immediate performance issues.

Thank you for attending the session.