When designing scientific experiments, the focus is mostly on data acquisition rather than on online analysis of data. However, immediate analysis enables active control or instant redesign of the experiment. In this article, we elaborate on the opportunities of creating self-improving experimental designs following the Self-Improving System Integration (SISSY) concept. Here, we propose several research questions and assess their importance by focusing on one use case concentrating on SISSY systems in general as well as two specific use cases taken from physics and materials engineering, respectively.