The use of functional traits has increased exponentially in ecology, particularly in attempting to explain ecosystem functioning. This popularity has led to many proposed definitions of functional traits, which in turn has informed recommendations about how to gather, summarize, and analyze trait data. In this paper, we revisit the definition of the functional trait from the perspective of physiological, community and ecosystem ecologists, and reason towards a broad, unrestrictive, and applicable definition. We then demonstrate how a popular technique, ordination, which is used to collapse multivariate trait data into orthogonal axes, undermines this definition and the primary benefit of functional traits. We outline the conceptual pitfalls associated with ordination, and make specific suggestions about alternative methods that progress functional traits as generalizable proxies for how organisms affect ecosystem processes. We hope these suggestions will improve our ability to move towards an ecological synthesis using a trait-based approach.