IoT systems generate large volumes of sensor data from which invaluable insights can be extracted to gain insight into the processes that are performed in and with the IoT systems. These insights can then be used to monitor the process’ compliance with respect to the given constraints and to improve the processes themselves, at run time or design time. Process mining techniques can be leveraged for this aim, but a gap needs to be filled between the IoT-system data and the event logs required to apply process mining. In particular, IoT-system events might be too fine-grained to immediately match the concepts at the level of the human understanding: very likely, they need to be aggregated to higher-level concepts to obtain a suitable level of granularity to further apply process discovery techniques. This chapter starts by discussing the literature on event-log abstraction that enables altering the event granularity to the right level to gain meaningful insights. Then, it focuses on experiences on real-life IoT data and reports on the application of event-log abstraction, aiming to discover IoT process models that are readable and accurate. These models are a prerequisite to provide actionable insights for a subsequent process’ monitoring and optimization.