Self-assessment skills have long been identified as important graduate attributes. Educational interventions which support students with acquiring these skills are often included in higher education, which is usually the last phase of formal education. However, the literature on self-assessment in higher education still reports mixed results on its effects, particularly in terms of accuracy, but also regarding general academic performance. This indicates that how to foster self-assessment successfully and when it is effective are not yet fully understood. We propose that a better understanding of why and how self-assessment interventions work can be gained by applying a design-based research perspective. Conjecture mapping is a technique for design-based research which includes features of intervention designs, desired outcomes of the interventions, and mediating processes which are generated by the design features and produce the outcomes. When we look for concrete instances of these elements of self-assessment in the literature, then we find some variety of design features, but only a few desired outcomes related to self-assessment skills (mostly accuracy), and even less information on mediating processes. What is missing is an overview of all these elements. We therefore performed a rapid systematic literature review on self-assessment to identify elements that can help with understanding, and consequently foster an effective self-assessment of learning artifacts in higher education using conjecture mapping as analytical framework. Our review revealed 13 design features and six mediating processes, which can lead to seven desired outcomes specifically focused on self-assessment of learning artifacts. Together they form a model which describes self-assessment and can be used as construct scheme for self-assessment interventions and for research into the how and why self-assessment works.