Disruptions of the gut microbiome have been correlated with a range of diseases, leading to the assumption that we may be able to improve people’s heath by manipulating their gut community. One major challenge lies in detangling the dense network of community-wide metabolic interactions to identify the most promising targets for microbiome manipulation. Here we propose that quantifying the functional redundancy of microbial metabolites through metagenome-scale metabolic modelling can help to narrow down these targets. We introduce the Metabolite Exchange Score – a metric designed to identify keystone metabolites and understand how microbiomes differ in functional redundancy. We applied our framework to over 1600 individuals and found that microbiome compositions associated with six diseases were marked by a significantly lower functional redundancy, while microbiome compositions associated with five other diseases showed no difference when compared to healthy controls. When applied to a Crohn’s Disease (CD) case-control cohort, our analyses identified hydrogen sulphide as one of the key metabolites for which functional redundancy was lost in diseased individuals. Intriguingly, microbiomes of individuals with CD were marked by a loss of hydrogen consumers, rather than a gain in producers. This work sets the stage to leverage gut ecology for the development of microbiome-targeted therapies.