Many policy systems and education systems have grown more complex in the last three decades. Power has moved away from central governments in different directions: upwards towards international organisations, sideways towards private institutions and non‐governmental organisations and downwards towards local governments and public enterprises such as schools. Where once we had central government, we now have governance, which can be defined as the processes of establishing priorities, formulating and implementing policies, and being accountable in complex networks with many different actors. Steering in such complex education systems emerges from the activities, tasks and responsibilities of state and non‐state actors, operating at different levels and from different positions and often has un‐deliberate, un‐intentional and un‐foreseen consequences. There are many conceptual models that encapsulate this complexity, but this article suggests that there is a real need for empirical research. Without empirical research it remains unknown whether and how steering in complex networks works out in practice, what are its effects and for whom. Moreover, it is only through empirical research that we can find out whether central government has become less dominant, or rather whether its appearance has changed and it has become less visible, but not necessarily less influential. Foucault's governmentality perspective is a useful notion on which to build such a framework for empirical research which allows for a careful study of the interactions that signify steering. Inspired by Foucault, this article develops a trilogy of assumed conditions for steering to take effect in modern societies. Following this reasoning, ‘something' first needs to be made thinkable, calculable and practicable by different actors for steering to occur. This trilogy is a promising starting point for empirical research into very specific phenomena which can help us to understand how steering in complex education systems works.