How Leading Tech Companies Design Scalable Quality Assurance Systems

ZTO: Designing a Quality Assurance System for a Thousand‑Person Tech Team

Senior architect Gong Linhua from Zhongtong Technology explains how their thousand‑plus R&D team built a comprehensive QA system. The talk covers four main areas:

Design thinking for quality assurance in large‑scale development teams.

Testing standards across multiple business scenarios.

Innovations in test automation and specialized testing practices.

Methods and best practices for test organization growth and cultivating a quality‑centric culture.

The presenter emphasizes systematic test design, automation, and cultural adoption to sustain quality as the organization scales.

Meituan: Event‑Tracking Quality Assurance Practice

Cheng Xiao shares Meituan’s approach to integrating event‑tracking data quality into the testing workflow. By establishing a systematic implementation process, the team improves data reliability and coordination efficiency, turning what is usually a product‑driven verification task into a formal QA activity.

ByteDance: Advertising System Quality Assurance

Speakers Zhang Linling and Chen Jun describe how ByteDance ensures reliability in its massive ad platform, which spans client, data flow, algorithm, and model layers. Key points include:

Front‑end (ad delivery) QA using large‑scale automated tests based on traffic cloning, recording/replay, and data coloring techniques.

Back‑end (conversion) QA employing AB testing, streaming test platforms, and diff testing to address challenges such as poor testability, complex case management, and high integration costs.

The session highlights reusable solutions that can be generalized to other server‑side services.

Kujiale: Multi‑Dimensional Online Inspection and Fault‑Injection Platform

Zou Haibin presents a framework for early detection and rapid localization of online issues across diverse business lines. The agenda includes:

Quickly perceiving online problems to detect issues before users notice.

Accurately pinpointing root causes to minimize impact.

Cost‑effective coverage of numerous business scenarios.

Custom inspection mechanisms for tightly coupled business logic.

The talk also outlines the evolution of Kujiale’s fault‑injection platform in three stages and discusses future directions for stability engineering.

Editorial Insights

Commentary notes that each session offers practical, well‑structured case studies suitable for traditional businesses seeking to strengthen their testing and reliability practices.