We analyze a coordination game characterised by varying degrees of conflict of interest, incentive to coordinate and information asymmetry. The primary objective is to question whether endogenous leadership better enables coordination. A secondary objective is to question whether preference and information asymmetries cue who should lead. Both experimental and theoretical results are provided. We find that in theory leadership should allow coordination, whether or not preferences are common knowledge. In practice we found that leadership did enable coordination but information about others preferences also helped. This was explained as due to some participants being too eager to lead. Which may be surprising given that we find, both in theory and in practice, leaders get relatively low payoffs, particularly when preferences are private information.