Crossings are important yet vulnerable parts of a railway network. This warrants that, in the Netherlands (a country with high traffic loads), crossing geometry is measured twice per year by dedicated measurement vehicles. This produces so much (point cloud) data that prioritizing and predicting is not possible by hand. It can be done by automating the assessments, by an automated process of four steps: cleaning the data, generating relevant performance indicators (features) per measurement, drawing conclusions from all features and visualizing/communicating the results. This project encompasses the first three steps for the most common type of crossing (the 1:9 fixed UIC54 common crossing). The results show that the features yield useful information and insights for both prioritizing and predicting.