Rudolf Groetz
Let's begin with some background information to understand our challenge. Let's start with the product we selected for our transformation journey. We used a Jira platform to validate our low code approach. 80% of our agile product teams are using Atlassian Jira, where each country has their own cloud instance. A dedicated platform team of six people located in the Vienna headquarter takes care that the platform is operated. The test strategy is that after an upgrade of the Jira platform, manual tests will be executed by the platform team and some end users. Additionally, each JIRA project administrator checks if the most important features in workflows are working correctly. These tests are very time consuming and we have based on the content of the upgrade, tests cycles with a length of up to 10 days and more. And of course, we have a test automation strategy. The test automation strategy is already defined, which should be implemented by one of the JIRA DevOps engineers who was upskilled in Playwright. Unfortunately, the implementation hasn't started due to other more important task. I'll show you a number. Per year, we spend around 400 hours in manual testing of the Jira platform. Keep this in mind. Let me introduce the people who are the protagonists in this journey. On the one hand, we have Philipp. Philipp is the product owner of the Jira platform, and he's very upset with the current test strategy. He wants business people can also automate tests to shorten their manual test cycles. And then, Rudolf, that's me. I'm not a fan of low code platforms, since I have never seen a successful implementation of this approach before. And for me, it is a fact that test automation is software development. For me, low code test automation is only marketing and sales jamming. And then we have Thomas. Thomas who is an Agile engineering coach and the test automation chapter lead. He's highly convinced that low code test automation cannot scale in large enterprises, and therefore, it is doomed to fail. And last but not least, Tim, the holiday intern biomedical student. He has no coding knowhow, no idea how this automation works, no idea about software development. And the internship was not his idea. Now, we come to the question why it was necessary to evaluate the low code approach for test automation. At first, we lost the person who should automate the tests. The person who was educated in test automation with Playwright. Secondly, the person who was responsible for the Jira test strategy left RBI. And last but not least, a new business continuity process was introduced that made automated tests mandatory for business-critical services. Jira was classified as business critical, which means the product owner had a problem with test automation. With the implementation of the business continuity process. Philipp was promoted to the new service manager and must implement all mandatory process requirements. That means Philipp must solve the test automation challenge, and he found the solution in adapting our test automation paradigm. And this was Philipp's approach. People without coding skills should be able to automate tests with low code platforms, meaning get rid of the engineering bottleneck. Since low code approaches has already been used in the development of business applications for two years, the idea was not that far-fetched, and our management also liked this idea. In the old test automation paradigm, test automation was largely in the hands of developers and test automation engineers. The tool of choice where open-source frameworks require a definitive skill set for effective implementation. Resources for test automation was always a bottleneck. What would be the new approach? Now, picture a landscape where everyone in the team, including our business counterparts, actively contributes to test automation? Coding skills are not long a prerequisite. This democratization of test automation empowers every team member to engage in the test automation process. This paradigm shift acknowledges that effective testing isn't just about code, it is about collaboration, diverse perspectives, and empowering every team member to contribute. Then we organized one day Testaton which product owners and other stakeholders with no coding skills. The goal was to find a low code test automation platform that we could use in a proof of concept in the next phase. During the Testaton, the participants evaluated a pre-selected set of low code test automation platforms if they supported their following requirements. Give our owners financial institution prioritizing top tier security standards is imperative. The key elements of these requirements are data hosting in the European Union, all data generated stored and processed by the platform must reside within EU based data centers. The platform also must be hosted in the European Union. And then it was about the license model. A critical requirement is the adaption of a license model based on the numbers of test execution as opposed to traditional license models. The selected low code test automation tool should offer a license model that is directly tied to the number of test executions. The license model should provide flexibility to scale our test automation efforts without incurring exorbitant upfront costs. Then, the AI support. We aim to cater both user without coding skills and experienced Developers looking to expedite the process of writing tests. The requirement is to provide an AI assistant that allows testers unfamiliar with code to express complex scenarios. This functionality will empower both testing and development teams, ensuring a seamless and accessible process for expressing and implementing diverse testing scenarios. And then we talk about test automation pattern support. This functionality is crucial for streamlining maintenance effort, especially when tests are organized into multiple test suites. And then natural language was also a feature. This feature aims to facilitate a more intuitive and accessible test script creation process. This should enabling users to express test scenarios, actions and expectations in a human-readable form. The tool should provide a syntax that is closely to spoken or written language. And then last but not least, we have the live authoring. This feature aims to provide real time feedback, eliminating the need for their traditional write, run, debug, repeat cycle. Test outdoors should be imbalanced to create tests more effectively by observing a both validating each step in real time on a headless browser. And we found one platform that supported all of our requirements. The name of the tool as I say it, this is not a product pitch. 4 weeks after the Testaton we did a detailed proof of concept with the winner from the test on, we got a dry license from the vendor for three months. The goal was for a person with no coding skills and no technical background to automate the set of tests for the Jira platform, which was normally manually executed by the JIRA product owner, and the person who did all the heavy lifting was our holiday intern. The goal of the POC was to create test, validate the 25 most used JIRA features to validate their most used JIRA workflows, and to automate crud operations via their Jira APIs. Additionally, our objective was to find out if each test suite can be executed in a data driven approach for each JIRA project in a build pipeline, and the time budget was defined with 20 person days. And here are some impressions of the implementation. We were really amazed at the steep learning curve with this tool, and how easy it was to implement the tests, and the birds got around while the implementation was still underway two other product teams decided to give the low code approach a chance. One was our on-premise GitHub platform and another customized software as a service application. That means, one is within three weeks, the low code testing automation approach turned our test automation paradigm upside down. And now let's take a look at the result of the proof of concept. The whole proof of concept was organized in four one week scrum sprints. Five days earlier explained, we closed the POC and the holiday intern presented the results and the outcome to the main stakeholders. In this presentation, the proof of concepts results left stakeholders very impressed, even initially skeptical product managers. We have fully convinced and enthusiastic about running experiments in their teams. The question was not whether to adopt this approach or not. Rather, there was the question when we can start the success of the Testaton sparked immediate interest and set the stage for a dynamic shift in our testing automation practices. The future of our test automation endeavors was not the question. It was a journey that began then.