The European Common Fisheries Policy recognizes the importance of accounting for heterogeneity in fishing practices, and métier-based sampling is now at the core of the EU Data Collection Framework. The implementation of such an approach would require Member States to agree on the standard regional métier definitions and on practical rules to categorize logbook records into métiers. Several alternative approaches have been used in the past to categorize landings profiles, but no consensus has yet emerged. A generic open-source workflow is developed to test and compare a selection of methods, including principal components analysis (PCA), hierarchical agglomerative clustering (HAC), K-means, and Clustering LARge Applications (CLARA), and to provide simple allocation rules. This workflow is applied to a unique regional dataset consisting of bottom-trawl logbooks of five North Sea countries. No method proved to be infallible, but combining PCA with either CLARA or HAC performed best. For 2008, a hierarchical classification with 14 species assemblages is proposed. Discriminant analysis proved more robust than simple ordination methods for allocating a new logbook record into an existing métier. The whole approach is directly operational and could contribute to defining more objective and consistent métiers across European fisheries.