Hail and severe convective storms rank among the most damaging weather hazards in Central Europe. Relevant processes span a range of scales: from hailstone growth at the microphysical level over storm dynamics and organization on the mesoscale to larger-scale synoptic environments and long-term climate change. The analysis of societal risk adds complexity, since impacts depend on exposure, vulnerability, and how extremes combine. This talk brings together several lines of research, linking processes to impacts across scales. Firstly, we investigate if we can learn about relevant processes from machine learning models that predict observed hail stone sizes. Secondly, on a regional scale, we explore whether emerging convective hotspots around the European Alps can be connected to large-scale patterns. Finally, we focus on impact relevance, presenting resampling and storyline approaches that allow for process-sensitive risk modeling to address both short-term and long-term impact assessments.