How ICBC’s Software Development Center Digitized Test Execution: Strategies and Results

ICBC’s Software Development Center has long used an automated testing model covering interface and batch testing, driven by metrics such as case execution rate. As business expands and architecture transforms, the existing model shows limitations: high false‑positive rates, execution success sensitive to environment and data fluctuations, and costly test‑site retention that hampers traceability.

To address these issues, the center launched a test‑execution digitization initiative, focusing on test design and execution. By treating standardized cases, traffic cases, and coverage as digital assets and leveraging recording‑playback, AI, and precise testing techniques, they built a platform that delivers test standardization, visualization, and measurability.

Test Standardization – The team analyzed massive existing test scenarios and created a three‑stage framework (test preparation → test execution → data assertion) together with a four‑layer asset structure (element → module → case → scenario). This modular “building‑block” approach raised asset reuse by more than 40%, cut average case‑generation time from days to minutes, and tripled the number of generated cases.

Test Visualization – The center introduced a “no‑dependency” recording‑playback technique that isolates test execution from environment and data constraints. By mocking external services and data, testers can validate application functionality without rebuilding the full environment, achieving high stability, success rate, and coverage.

The digitized platform presents the full transaction chain of traffic cases, with each node carrying mock data. During replay, it automatically detects abnormal or error nodes, performs field‑level diff, and achieves 100% assertion coverage, cutting comparison time by 50%. Issue‑analysis time shrank from hours to minutes, and false‑positive rates approached zero, demonstrating the efficiency of visual testing.

Test Measurability – By extending industry coverage tools, the center built a precise testing platform that silently collects coverage data, generates function‑call graphs, and supports white‑box testing. It records branch coverage and links it to test cases, providing program‑level, method‑level, line‑level, and instruction‑level metrics with 100% accuracy, enabling objective quality quantification.

Applying the same digitization to overseas applications, the team faced massive regression workloads and coordination constraints with external partners. The digital solution eliminated external dependencies, allowing on‑demand testing of internal transactions, covering over a hundred service‑scenario transactions and generating tens of thousands of valid test cases. Visualization and measurability reduced business‑cutover risk and ensured smooth project progress.

In summary, the digitized test‑execution framework now supports nearly a hundred application testing scenarios, with tens of thousands of automated assets. These assets safeguard critical business transactions, enable 24/7 stability, and empower efficient R&D. Looking ahead, the center plans to expand intelligent, diverse test scenarios to further boost testing automation and support high‑quality fintech development.