Everyone who has ever lost their keys in a busy room knows that they cannot move at full speed and hope to find them ; one must slow down enough as not to miss them. This compromise between speed of moving and success of detection is not specific to humans, of course, and in fact is commonly encountered in foraging animals, and even in cell biology. This opposition between relocation and detection can even lead to an intermittent behavior, i.e. with different phases during the search. We model such search processes by memoryless explorations on graphs, i.e. random walks, where you can decide at each step to query a node or not. The goal is to balance the time spent walking and the time spent querying.