Functional diversity is increasingly used alongside taxonomic diversity to describe populations and communities in ecology. Indeed, functional diversity metrics allow researchers to summarize complex occupancy patterns in space and/or time (what is changing?) that lead to changes in communities and/or populations (the process; how is it changing?) in response to some stressors (the mechanism; why is it changing?). However, as the diversity of functional diversity metrics and methods increases, it is often not directly clear which metric is more readily appropriate for which question. We studied the ability of different functional diversity metrics to recover patterns and signals from different processes linked to common assembly mechanisms (environmental filtering, competitive exclusion, equalizing fitness, and facilitation) in community ecology. Using both simulated data and an empirical dataset affected by more complex and nuanced mechanisms, we tested the effectiveness of different space occupancy metrics to recover the simulated or empirical changes. We show that different metrics perform better for different tasks, emphasizing the importance of not using a one-size-fits-all metric. Instead, researchers should carefully consider and test whether a particular metric will be effective in capturing a pattern of interest.