Improving Test Data Quality and Automation Efficiency: A Ctrip Ticket BU Case Study

Drawing on Shannon's information theory, the author argues that reducing system uncertainty requires high‑quality information, and that test data quality directly impacts automation effectiveness.

Two core principles are proposed: acquire more high‑quality data to increase coverage, and improve engineering efficiency to speed up data processing.

A case study of a complex feature (Feature A) shows that limited test cases cannot cover the combinatorial explosion of possible scenarios; therefore, the team expands information sources, builds a test information model, and uses a crawler to collect usage data, filter unused combinations, and prioritize remaining ones.

Data persistence is addressed by creating data mirrors for both message and relational data, supporting SQL Server and MySQL, with automatic date shifting and one‑click restoration.

To handle growing data volume, a searchable data service platform is built, featuring scoring, sorting, locking, and keyword suggestions to help users retrieve relevant test data efficiently.

Engineering efficiency is improved through backend traffic‑replay testing (using mock servers, dual environments, automated comparison, and CI integration) and frontend image‑comparison testing (leveraging crawled data, dynamic mocking, schema‑driven navigation, image diff, and headless browsers via Puppeteer).

The SnapDiff platform incorporates report accuracy enhancements (OCR‑based diff, ignore rules, intelligent clustering) and execution speed gains (headless browsers, distributed concurrency, self‑service task creation and job launching).

Overall, the integrated system enables continuous testing across all stages, has already uncovered 60+ bugs in 30 days, and will continue to evolve to further reduce uncertainty in software delivery.