The article compares the shift-left and test-first approaches in software testing, emphasizing the historical context and development of these methods. Test-driven development (TDD), a test-first approach popularized two decades ago, has been proven to reduce bugs significantly, despite increasing development time, as demonstrated in studies by Microsoft and IBM. The shift-left approach, which advocates for early testing, aligns with test-first methods like TDD and BDD but raises the question of how early testing should begin. The article discusses the challenges of implementing test-first methods in end-to-end (e2e) testing, particularly in test automation, and highlights the role of model-based testing (MBT) as a solution. It introduces a two-phase MBT process, featuring high-level and low-level models to facilitate test-first practices even before application implementation. The high-level model, which is human-readable and abstract, allows testers with domain knowledge to execute tests independently of the implementation, while the low-level model becomes machine-readable for test automation once the application is ready. The author concludes that although test-first approaches are more challenging, they are preferable to shift-left and can be largely implemented in test automation, leading to fewer bugs and cost-effectiveness.