Abstract Transaction data from digital payment systems can be used to study economic processes in such detail that was not possible previously. Here, data from the Sarafu token network, a Community Inclusion Currency in Kenya, are analysed. During the Coronavirus Disease 2019 (COVID-19) emergency, Sarafu was distributed as part of a humanitarian aid project. In this work, the transactions are analysed using network science. A topological categorization is defined to identify cyclic and acyclic components. Furthermore, temporal aspects of the circulation that takes place within these components are considered. The significant presence of different types of strongly connected components compared to randomized null models shows the importance of cycles in this economic network. Especially, indicating their key role in currency recirculation. In some acyclic components, the most significant triad suggests the presence of a group of users collecting currency from accounts that are active only once, hinting at a possible misuse of the system. In some other acyclic components, small isolated groups of users were active only once, suggesting the presence of users only interested in trying the system out. The methods used in this article can answer specific questions related to user activities, currency design, and assessment of monetary interventions. The methodology provides a general quantitative tool to analyse the behaviour of users in a currency network.