The rhizosphere represents a dynamic and complex interface between plant hosts and the microbial community found in the surrounding soil. While it is recognized that manipulating the rhizosphere has the potential to improve plant fitness and health, engineering the rhizosphere microbiome through inoculation has often proved challenging. This is in large part due to the competitive microbial ecosystem in which the added microbes must survive, and lack of adaptation of these added microbes to the specific metabolic and environmental pressures of the rhizosphere. Here, we constructed an inoculation formula using a defined synthetic community (dSynCom) approach that we hypothesized would improve engraftment efficiency and potentially the relationship with the host plant, Sorghum bicolor. The dSynCom was assembled from bacterial isolates that were either: 1) identified to potentially play a role in community cohesion through network analysis, or 2) identified to benefit from host-specific exudate compounds. Growth of the dSynCom was first evaluated in vitro on solid media, secondly in planta under gnotobiotic laboratory conditions, and finally using sorghum plants grown in the field. We demonstrate that the dSynCom performs best in terms of maintaining diversity when grown in the presence of the plant host in lab conditions, and that many lineages are lost from the community when grown either in vitro or in a native field setting. Finally, we demonstrate that the dSynCom is able to promote growth of above- and below-ground plant phenotypes compared to uninoculated controls, both in the lab and when applied to plants grown in the field. These results demonstrate the potential utility of SynComs for supporting crop performance even in the absence of persistence, and the need for a deeper mechanistic understanding of community control of host fitness in agricultural contexts.