Experimental research has shown that ordinary people often perform remarkably well in solving coordination games that involve no conflicts of interest. While most experiments in the past studied such coordination games among socially distant anonymous players, here we study behaviour in a set of two player coordination games and compare the outcomes depending on whether the players are socially close or socially distant. We find that social closeness influences prospects for coordination, but whether it helps, harms, or has no impact on coordination probabilities, depends on the structure of the game.