Goal modelling is one the most important early activities in requirements engineering. Here, we describe a vision for a conceptual basis for the systematic identification and treatment of uncertainty in goal modelling. We aim to characterize the wide variety of uncertainty in goal modelling and to provide a theoretical framework for systematic uncertainty analysis. We thus adopt Walker's taxonomy which distinguishes among three dimensions of uncertainty: location, level, and nature. In addition, we propose to adapt Walker's uncertainty matrix as a heuristic tool to categorize various dimensions of uncertainty in goal modelling to serve as a conceptual framework for improving comprehension and communication of uncertainty between modellers and stakeholders and among modellers themselves. Understanding the various dimensions of uncertainty is a vital step towards the sufficient recognition and treatment of uncertainty in goal modelling activities. This in turn will help identify and prioritize critical uncertainties, which affect the goal modelling process in its entirety. We thus propose a long-term research agenda and urge community contributions in this research direction.