The paper introduces a method for quantitatively analyzing the expressive power and semantic capacity of metamodels in the context of conceptual data modeling. The method leverages a formal semantic representation of metamodels, based on the Conceptual Layer of Metamodels (CLoM) and expressed using an ontological system of concepts. By extracting semantic constructs, such as concepts, semantic atoms, and semantic particles, the approach allows for a structured evaluation of metamodels’ abilities to express semantic complexity. The proposed method is applied to selected conceptual metamodels, including AOM, EER, ORM, and a reference graph metamodel, providing insights into their semantic capabilities. Results demonstrate alignment with expert intuition and reveal differences in semantic richness and complexity.