Currently, the use of mobile devices, mainly smartphones, has been gradually increasing due to the increasing development of mobile technologies. However, it is often the case that web applications are not tested on mobile devices which has led to, among others, usability errors. Therefore, the importance of usability testing in mobile devices has been increasing, but it is costly and time-consuming; the use of these devices in many cases cannot be replicated solely in a laboratory, but it should be replicated in the field which is more complex to implement. In addition to these usability testing and to reduce the time and complexity of testing in these devices, we proposed an automatic approach to identify indicators of usability problems (Usability Smells). To do so, we implemented a process of pattern matching in the tool UseSkill and proposed Usability Smells and their corresponding patterns for this context. This tool uses the data from the interactions of the user with a Mobile Web Application, discovering Usability Smells. This new process was experimented in two study cases with two applications where it collected data of participants who used it on a mobile device (smartphone). The results obtained were promising in identifying usability smells in relation to usability problems. © 2020 ACM.