Abstract
For digital health interventions, the "gold standard" of evaluating effectiveness is the randomized control trial (RCT). Yet, RCT methodology presents issues such as precluding changes to the technology during the study period as well as the use of study settings that do not reflect "real world" contexts. In this paper, we draw on empirical material from our ethnographic research on an app-based program called HIVSmart!, which is a digital strategy designed to support people in the process of HIV self-testing. We explore how this digital health approach was made to work in different contexts, and what this means for monitoring and evaluation of digital health interventions. Our analysis reveals that making technology which is "ease to use" depends on various user practices developed over time and the multiple, shifting users who define this ease of use. Furthermore, it shows that asking whether an app is "working" provides different answers at different moments, as the technology, as well as the human and physical supports around it, change and adapt over time. Therefore, questions around the feasibility of the technology, as well as ongoing monitoring efforts, need to be developed in a way that capture this fluctuation in the technology and its surroundings. Based on our insights, we contribute to implementation science approaches in the field of digital health by putting forward recommendations regarding how we can assess the effectiveness of digital health tools.