The design and analysis of biodiversity and ecosystem function (BEF) studies have both evolved in recent decades to permit the addressing of increasingly complex research questions. Simple linear regression techniques can address how average patterns in ecosystem function are affected by species richness, while ANOVA approaches permit the comparison of specific communities contained in the design. The Diversity–Interactions modelling approach can address both of these research questions and, furthermore, it can assess the relative performance of individual species, assess if and how species interact, and can identify species diversity regions (characterised by species richness, composition and proportions) that lead to the best-performing mixtures. In this paper, we discuss experimental designs for varying research questions. We also show the added benefits from using the Diversity–Interactions modelling approach over traditional approaches using simulated BEF data and data from a productive grassland experiment. Being able to identify the best-performing mixtures can provide valuable knowledge for the management of both nutrient-rich productive grasslands where there is a degree of control over mixture design at sowing, and nutrient-poor semi-natural grasslands where management decisions can be taken to influence species dynamics.