A new crew management system is being developed under intense time pressure. Requirements are flexible, acceptance must be spot on. Model-based testing creates a common language between the specialist department, development and quality assurance. The approach starts at high altitude, consolidates domain knowledge via interviews and reviews and refines models iteratively. Test cases are automatically generated from them, coverage and change impact become transparent in tools such as Jira, and regressions remain up-to-date. Clear technical models, good requirements and a focus on business value are crucial.
In this episode, I talk to Oliver Schuhmacher and Marvin Jakob about model-based testing and how it brings teams together. Using Lufthansa CityLine as an example: new crew management, high time pressure, clear acceptance required. Models serve as a common language. They start at a high altitude, interview specialist departments, testers and developers, refine them step by step and review them regularly. They generate test cases from the models using a tool, make coverage and impact visible in Jira and keep tests up to date more quickly in the event of changes. Important: good requirements and a focus on business value.
"You can only write good tests if the requirements are also good. And requirements and tests form a single unit." - Oliver Schuhmacher, Marvin Jakob
Oliver Schuhmacher is a consultant at cimt AG. He has been dealing with the topic of software quality for over 20 years and has supervised various projects in the areas of finance and logistics during this time. For more than 10 years he has been involved in model-based testing and the derivation of test cases from test models.
Marvin Jakob is a consultant at cimt AG. In the areas of architecture, IT operations and project management, he supports software throughout its entire life cycle.He thus looks at the topic of software quality from many perspectives and deals with innovative test methods that meet the unique requirements of his projects.
In the latest "Software Testing" podcast, Richie, Oliver and Marvin talk about their experiences with model-based testing. They talk about how this method was introduced at Lufthansa CityLine and the benefits it has brought to team collaboration. The conversation shows that it's not just about technology, but also about people, communication and trust.
Model-based testing means that test cases are no longer created on the basis of long text documents. Instead, the team uses graphical models to describe procedures, decisions and processes in the system. These models help to understand requirements more clearly and derive the right tests. All those involved - the specialist department, developers and testers - have a common language thanks to the same model.
Lufthansa CityLine faced a major challenge: a new airline had to be set up, with all its infrastructure and IT connections. The crew management system was particularly complex. The teams had to ensure that the IT solutions worked exactly as they had been ordered. This required not only good testing expertise, but also a strong common understanding of the processes - this is where models came into play.
Oliver describes how knowledge is usually spread across several groups. The business department knows the processes, the testers know how to test, and the developers implement it technically. The model helped to bring these groups together in workshops. They started at a high level of abstraction, gradually incorporated further details and visualized processes together. Questions such as "What happens now?" or "What comes next?" guided the participants through the requirements in a structured way.
Marvin emphasizes that models work like a simple language. Instead of endless textual deserts causing confusion, diagrams are created that everyone understands. This makes coordination much more efficient. In the workshop, everyone can have their say, point to something in the model, identify a gap or quickly understand where a mistake is.
The discussions no longer focus on minor details or technical details. They focus on the big picture and the important issues that have the greatest benefit for the company. This increases the quality of the results and saves time.
One of the biggest advantages is the simple derivation of test cases. As soon as the model is ready, the team can generate tests at the touch of a button. This speeds up the creation process massively and reduces errors. The test cases are traceable because they come directly from the agreed model.
The models also help if requirements change afterwards. Marvin describes how the visualization makes it clear which areas are affected. This allows quick reconciliation between new requirements and existing tests. Maintaining large quantities of tests, which otherwise takes a lot of time, becomes more manageable and targeted.
Oliver emphasizes yet another effect: transparency and trust grow. The model is understandable for everyone, stakeholders feel involved. Test tasks are no longer completed behind closed doors, but are communicated and reviewed openly. New team members can also familiarize themselves quickly.
Pitfalls and tips from experienceNot everything runs smoothly right from the start. Good tests need good requirements. This is where the model helps to identify gaps at an early stage. Testers should keep the models as simple as possible and regularly test them with new people to make sure they are understandable.
Marvin and Oliver advise simply getting started and trying things out. Mistakes are normal at first, but the learning effect is huge. It is important to stay in the team, check regularly, communicate openly and get help from the community if necessary.
Model-based testing is more than just a new method in the toolbox. It is an approach that brings teams together, provides clarity and measurably improves the quality of testing. Experience at Lufthansa CityLine shows: With models, the whole team speaks the same language - and that releases energy that really counts.