Climate change will likely reshuffle ecological communities, causing novel species interactions that could profoundly influence how populations and communities respond to changing conditions. Nonetheless, predicting the impacts of novel interactions is challenging, partly because many methods of inference are contingent on the current configuration of climatic variables and species distributions. Focusing on competition, we argue that experiments designed to quantify novel interactions in ways that can inform species distribution models are urgently needed, and suggest an empirical agenda to pursue this goal, illustrated using plants. An emerging convergence of ideas from macroecology and demographically focused competition theory offers opportunities to mechanistically incorporate competition into species distribution models, while forging closer ties between experimental ecology and macroecology.