In an e-learning environment it is crucial that accurate web usage data can be collected and analysed to ensure that on-line learning material is optimally structured and designed. In this paper we investigate the relationship between cached and non-cached web log files as we identify and analyse user paths through an e-learning web site. We outline the development of a web data mining and data visualization tool DMASC (Direction Mapping and Sequence Chart). We also demonstrate, using a set of test data, how our mining and mapping methods can be used to reveal otherwise unseen web design flaws.