A considerable amount of e-learning content is being delivered via virtual or managed learning environments. These platforms keep track of learners' activities including content viewed, time spent and quiz results. This monitoring trawl provides appropriate data to enable personalised e-learning experiences through the application of existing data mining and knowledge discovery techniques.The reasons for not providing this type of bespoke teaching is, in addition to technical and financial constraints, largely due to the plethora of educational and pedagogical issues which have to be overcome. This paper presents these obstacles and suggests solutions thus challenging the community to focus on a new research area which concentrates on facilitating the specification and application of pedagogical domain knowledge for incorporation into existing data mining and knowledge discovery frameworks. This includes educational thresholds, constraints, taxonomies and previously discovered knowledge as well as pedagogical interestingness and metrics.