An emergent stream of research in management employs configurational and holistic approaches to understanding macro and micro phenomena. In this study, we introduce mixture models—a related class of models—to organizational research and show how they can be applied to nonexperimental data. Specifically, we reexamine the long-standing research question concerning the CEO pay–firm performance relationship using a novel empirical approach, treating individual pay elements as components of a mixture, and demonstrate its utility for other research questions involving mixtures or proportions. Through this, we provide a step-by-step guide for other researchers interested in compositional modeling. Our results highlight that a more nuanced approach to understanding the influence of executive compensation on firm performance brings new insights to this research stream, showcasing the potential of compositional models for other literatures.