For microbiome research to be translated into clinical applications, the field needs to move beyond correlational studies; we suggest adopting a multi-phase iterative approach that leverages in vivo and ex vivo strategies, as well as a combination of traditional microbiology, with big-data approaches.