Complex systems often exhibit sudden and potentially dangerous regime transitions. Anticipating these changes can be crucial for implementing adaptation measures. Several data-based diagnostics have been proposed, but their ability to capture changes in the dynamics of a system depends on the characteristics of the system and on the observed data. In this talk, I will discuss the performance of classical and new indicators, using real-world observed data (vegetation images to identify desertification transitions), as well as experimental data and simulated data generated with controlled variation of a bifurcation control parameter.