Boosting Automation Test Substitution and Reuse: Building an Asset System from Mandatory Test Cases

Problem Insight

Accelerated digital transformation in banking has increased release frequency, leading to frequent functional regressions. Manual regression testing is time‑consuming, coverage‑limited, and prone to omissions, which can cause production defects and damage customer experience and bank reputation.

Root‑Cause Analysis

The bank identified two main reasons why automation failed to replace manual testing:

Selection of automation scope lacks standardized methods, relying heavily on subjective judgment, resulting in incomplete coverage and preference for easy‑to‑implement but low‑value features.

Automation scripts often miss critical checkpoints because test cases are ambiguous, script dimensions differ from manual case design, assertions are omitted, or environment/data stability limits verification.

Solution: Mandatory Test‑Case‑Driven Automation Asset System

The bank built a complete asset construction and application framework centered on "mandatory test cases"—the set of cases that must pass before a version goes live. The framework consists of three layers:

Standardized Input : Construct and govern mandatory test cases.

High‑Quality Transformation : Develop trustworthy scripts that faithfully implement the cases.

Diverse Application : Deploy assets across the entire testing lifecycle.

1. Standardized Input – Building Mandatory Test Cases

Mandatory test cases target critical business functions, core transaction flows, and high‑risk modules. Their creation follows a multi‑level funnel model:

Functional Tree Construction : A hierarchical tree maps subsystems → modules → transactions → business rules, providing a structured view of the system under test.

Mandatory Test‑Case Construction : Using a three‑stage funnel, cases are filtered by business impact, technical risk, and rule‑level impact (global, module, local) to obtain a concise, high‑value set.

Governance : Ensure clear, unambiguous steps with machine‑verifiable expected results, and make implicit checks explicit.

2. High‑Quality Transformation – Trustworthy Script Development

Each mandatory case undergoes feasibility analysis; non‑automatable cases are documented with reasons (e.g., mandatory human interaction, hardware dependencies, network isolation). Scripts follow a coding standard, combined with manual review and automated tooling, and are continuously validated through weekly full‑run checks. A barrier service simulates dependent systems to overcome environment constraints.

3. Diverse Application – Extending Automation Across the Pipeline

The assets are applied to smoke, integration, functional, and pre‑production verification stages, covering internal testing and version validation. Real‑time monitoring tracks case execution across projects, turning automation from a single‑point replacement into full‑process quality protection.

Practice Results

Implemented in over 40 projects, the framework achieved:

Automation transaction coverage >70% and 100% coverage of mandatory transactions.

More than 55% of systems reached >90% automation regression substitution; the remainder are steadily improving.

In a payment system, manual regression share dropped below 10% and round‑trip regression time shrank from ten person‑days to under one person‑day.

Script reuse per system increased roughly twofold, extending asset lifespan and ROI.

Future Outlook

Remaining challenges include low efficiency of manual script inspection and tight delivery schedules. The bank plans to:

Leverage large‑model AI to create a human‑machine collaborative inspection mechanism, automating semantic alignment, logic completeness, assertion validation, and style checks.

Build a low‑code script factory that packages common test actions, supports data‑driven templates, and integrates knowledge‑base rules to accelerate script generation while maintaining quality.

These innovations aim to further reduce development effort, improve asset freshness, and sustain high‑quality automated testing in banking operations.