There have been many attempts to address the actuation problem using interspeaker variation (Baker et al. 2011), articulatory biases (Sóskuthy 2013), and informativity (Cohen Priva 2017). In this work we study another factor, functional load, by examining ongoing mergers in English and Dutch, trying to isolate the conditions under which functional load has a detectable effect in preventing a merger from the conditions under which no effect is visible. The effect is clear in cases of unconditioned mergers, like devoicing of Dutch fricatives, but unclear in cases of conditioned mergers, like the loss of a contrast in rhotic environments in English. We discuss the implications of these findings for the actuation problem and for phonological acquisition.