Hundreds of diabetes self-management apps are available for smart phones, typically using a diary or logging methodology. This paper investigates how well such approaches help participants to make sense of collected data. We found that, while such systems typically support data and trend review, they are ill suited to helping users understand complex correlations in the data. The cognitively demanding user interfaces (UI’s) of these apps are poorly adapted both to the restricted real estate of smartphone displays and to the daily needs of users. Many participants expressed the desire for intelligent, personalized and contextually aware near-term advice. By contrast, users did not see tools for reflection on prior data and behavior, seen as indispensable by many researchers, as a priority. We argue that while designers of future mobile health (mHealth) systems need to take advantage of automation through connected sensors, and the increasing subtlety of intelligent processing, it is also necessary to evolve current graphs and dashboards UI paradigms to assist users in long-term self-management health practices.