Can UI designers agree to test projects

The perfect test concept - with ease!

Now it is important not to adopt this model 1: 1 in your own test concept, but to ask yourself based on the criteria: Is this really important for our project? If so, can we check it specifically? Do we have the competence to do this?

This shows whether certain non-functional test perspectives are still missing, for which you should work with the help of third parties (central experts in a test center, external companies, recognized experts in other projects).

Step 3, Doing the Right Thing: Setting Priorities and Developing Strategies

Now you are mostly at the point where you almost freeze in front of the burden that apparently comes from testing needs. In reality, not everything is equally important and relevant. To give an example: A web portal that has to allocate registrations (for sports courses, funding programs) on the principle of first-come, first-served, tends to struggle with security or load / performance. The functionality is extremely simple. Accordingly, you have to show the courage to take gaps, limit the unimportant to the core functions and possibly reduce the rest to a few sample tests. This is called a risk-based test strategy.

But there are also other test strategies that you should also think about. So z. B. the strategy of testing as early as possible. The earlier errors are found, the (significantly) lower the costs for rectification. For example, an agile team can commission in-house experts to carry out complex performance tests towards the end of a release. But that could be too late to react adequately. Why not make assumptions about the neuralgic weaknesses in advance and measure them with self-written (cheaper) JMeter tests almost daily? In this way, improvement trends can be made visible or the alarm can be raised immediately in the event of sudden deterioration. The above Complex performance tests should still be carried out for confirmation.

On this occasion, one can rather come back to the regression test and ask oneself what strategy is used to cope with this task? To be on the safe side, test everything over and over again because working software could suddenly no longer work after a change or extension? Nobody does that! But testing only the new features is grossly negligent in most cases. Anyone who relies on test automation early on, if possible on the unit level, but at least on the service level (see Mike Cohn's concept of the test pyramid [6]), can certainly afford a larger test portfolio than someone who relies entirely on manual tests is. Dr. Rößler recently presented a supplementary approach towards AI, which could also be a very interesting option for you [7].

Step 4, test environment and test data: Apply for the infrastructure in good time

And now write one thing very firmly behind your ears: in most cases, test data and test environment require particularly long-term preparation and a lot of input from employees outside the project team, who are certainly not willing to wait for orders or requests for administrative assistance from you.

Therefore, the question of how productive the environment needs to be should be clarified early on. Can one fall back on cloud solutions if necessary [8]? In the case of non-functional tests, such as safety and load / performance, this is a very uncomfortable topic. Coordinate with experts here. Perhaps the test environment can be cleverly reduced to the weak points to be expected and thus save a second productive environment.

It also gets tricky with the test data. In many cases, the use of productive data is even prohibited by data protection law. And according to the EU General Data Protection Regulation, it will be even stricter in the EU countries from 2018 [9]. As long as only one department acts as a tester, it's easier here. But for all other testers, synthetically generated test data or anonymized test data from production are a must. Using the tools for test automation, there are now good methods for generating high-quality test data yourself.