The efficiency with which flying animals convert metabolic power to mechanical power dictates an individual's flight behaviour and energy requirements. Despite the significance of this parameter, we lack empirical data on conversion efficiency for most species as in vivo measurements are notoriously difficult to obtain. Furthermore, conversion efficiency is often assumed to be constant across flight speeds, even though the components driving flight power are speed-dependent. We show, through direct measurements of metabolic and aerodynamic power, that conversion efficiency in the migratory bat (Pipistrellus nathusii) increases from 7.0 to 10.4% with flight speed. Our findings suggest that peak conversion efficiency in this species occurs near maximum range speed, where the cost of transport is minimized. A meta-analysis of 16 bird and 8 bat species revealed a positive scaling relationship between estimated conversion efficiency and body mass, with no discernible differences between bats and birds. This has profound consequences for modelling flight behaviour as estimates assuming 23% efficiency underestimate metabolic costs for P. nathusii by almost 50% on average (36-62%). Our findings suggest that conversion efficiency may vary around an ecologically relevant optimum speed and provide a crucial baseline for investigating whether this drives variation in conversion efficiency between species.