Given a model of the environment's behaviour, a set of system goals, and a set of controllable actions, the controller synthesis problem is to automatically generate a controller that only restricts controllable actions. Qualitative controller synthesis techniques yield controllers that guarantee achieving a given goal in the presence of an adversarial environment. However, synthesis produces one out of many possible solutions and no support for expressing preference over them is provided. Quantitative synthesis techniques are a natural way of optimising control with respect to preferences, however, they require quantitative modelling, which in many cases is not available, possible, or desired, and incur in high computational complexity. In this work, we aim to develop a qualitative preference framework for control problems.