Data‑Driven Process Improvement: Development Self‑Testing and Code Review Practices at ZhaiZhai

In the era of rapid internet development, agile development and MVP (Minimum Viable Product) are widely used. ZhaiZhai adopts MVP and short‑cycle delivery to enable fast trial‑and‑error while continuously optimizing process standards, improving team collaboration and project speed.

The article demonstrates how data‑driven analysis of project quality and efficiency metrics, combined with the PDCA (Plan‑Do‑Check‑Act) principle, guides process improvement from pain points identified through team observation, frontline communication, and project data.

Development Self‑Testing enhances developers' quality awareness, reduces reliance on QA, cuts communication costs, and saves testing effort. By analyzing data, ZhaiZhai identified that bugs concentrate in large requirements while small tasks yield few bugs, prompting a pilot program with fast‑iteration teams. Standards were refined iteratively, achieving a self‑testing rate above 95% and extending the practice company‑wide, saving roughly 50% of QA effort.

Code Review as a Team Habit improves code quality, early bug detection, cost reduction, and knowledge sharing. ZhaiZhai promotes code review by analyzing abnormal project data, selecting pilot teams, and customizing review processes for underlying systems and business‑type systems. The Beetle platform supports configurable review rules per project dimension.

Continuous effect tracking involves regular analysis of review frequency, comment count, and trends, focusing on anomalies where reviews occur without comments. This feedback loop drives ongoing optimization of the review process.

Conclusion Process improvement is a long‑term, continuous effort. This article presented two cases—development self‑testing and code review—demonstrating how data‑driven practices can accelerate and stabilize team performance, with future case studies expected to further enrich the methodology.