Mercer Test Automation
Increasing Test Coverage Effiency
For more than 70 years, Mercer have been providing trusted advice and solutions to build healthier and more sustainable futures for their clients, colleagues and communities in the areas of Superannuation. Investments and wealth and Asset management.
What was required
In 2019, Mercer Australia identified automation as one of the key enablers of their strategy to provide cost effective, tailored superannuation management services to their clients. Mercer’s existing software testing process was primarily manual and seen as an impediment to being able to support increased change velocity.
The systems in scope for testing were Legacy and Java based registry systems, with Web-based client and member facing information systems. Integration between these systems and to external client and industry systems was achieved through the use of a range of messaging and B2B systems.
Accuteque’s task was to review existing testing being undertaken by Product Delivery and identify areas where test automation could be beneficial and propose a strategy and framework for automation in testing.
The preferred automated testing framework was to be developed to MVP level and validated through use on a small testing project before handing over to broader use by their off-shore automation resources.
What we did
Over 8 months, our team provided a strategy that included automated testing as a key driver for increasing test coverage while reducing manual effort. We provided a framework for automation that:
- Included the use of existing manual testers in the activity of writing and executing automated tests.
- Supported execution of end-to-end tests across all systems.
- Maximised reuse of tests across 40+fund-based variants of their member facing applications.
- Prepare Mercer’s off-shore automation engineers to support and maintain the framework..
We evaluated existing testing activities, test tools and planned project of work to identify a strategy for achieving a reduction in test effort of 20%. We identified tool standards for automation of testing across Web, Java, Terminal and API based technologies. And we designed and delivered an automation framework for use by the entire testing team.
As part of delivery, we also provided the following training and mentoring services:
- Trained offshore automation engineers on the use of advanced Frameworks and best-practice automation approaches.
- Trained automation engineers and test analysts on the use of the Semantic Automation framework.
- Provided handover to offshore automation engineers of the Java code libraries used in the framework.
- Performed code reviews for automation in use in existing projects and advised on areas for improvement.
- Advised Project Managers and Test Managers on the opportunities and approaches for automation of testing in their projects.
The initial Strategy identified savings of $1m in the first 9 months, with increased year-on-year savings of over $1.5m.
By getting Mercer staff involved with the definition of the Semantic Test language, they were more engaged and prepared to adopt automation as a key part of their testing approach and were actively writing tests using the automated framework.
The use of an AI approach to UI automation proved effective in allowing tests written for one client fund to be used with minimal changes for other funds, despite their individually tailored UI designs. This delivered an 80% reduction in testing effort for testing of ICOP changes to 3 client funds when compared to the manual test effort recorded for testing the same changes across other client funds. The success of this allowed Mercer to bring forward testing of a project to transition 27 funds to the latest Web platform, with expected savings of greater than 80% for testing.
Automation libraries were written in the Framework to support their Digital Platform, registry database queries and registry Web Services interface. As well as providing support for functional testing of integrated systems, it allowed test data to be dynamically accessed by the test cases, significantly reducing the effort around Test Data Management.