Biodiversity underpins ecosystem services, functions, and resilience and is now a central focus in restoration. There is, however, little consensus on how biodiversity should be measured or integrated into ecosystem accounting frameworks and biodiversity trading mechanisms that can be used to value restoration actions. We assess the performance and cost-effectiveness of multiple biodiversity metrics by evaluating their ability to detect responses of plant, bird, and fish communities to restoration. Using a network of restoring and reference wetlands within the Blue Heart Sunshine Coast, a 5000ha recovering seascape in eastern Australia, we contrast the potential utility of multiple ‘simple’ (e.g. alpha) and ‘complex’ (e.g. beta and functional) measures of diversity. Metrics that describe changes in species composition (e.g. Sørensen-Similarity), consistently outperform other biodiversity measures in terms of their sensitivity, reliability and cost-effectiveness. By contrast, simpler measures of species numbers and diversity (e.g. species richness) always perform poorly. Our findings highlight the need for greater emphasis on community composition metrics in ecosystem accounting frameworks and biodiversity markets. We outline the tests used to quantify these findings and discuss the advantages of incorporating alternative biodiversity metrics beyond species richness to enhance our ability to measure and value ecological and economic benefits from restoration.