Test data generation is the way to predictable tests

In projects where test data is selected from the database, there often arises a situation where the number of unstable tests increases, complicating maintenance efforts. As a result, a failed test run cannot reliably indicate a problem, and the results need to be validated manually.

I will explain why 'We need to fix this urgently!' is quite a poor solution. It is much more effective to move from data selection to test data generation.

I will compare different approaches to generating test data and evaluate their pros and cons. Using our project as an example, I will explain the challenges you may encounter during implementation. And of course, I will share the results.