I argue that the Knowledge-Action Principle (KAP) helps to explain social knowledge in terms of social action. According to KAP, it is rational to act on the premise that p only if you know that p (Fantl and McGrath 2002; Williamson 2005; Stanley 2005). KAP links knowledge and action, since the success of action depends on knowledge. However, KAP has mainly been applied to individual actions (Williamson 2000) whereas social action is a central notion in social sciences (Parsons 1937; Skinner 2002) and social ontology (Gilbert 1989; Searle 1990; Bratman 1992; Tuomela 2000). My aim is to show that there is no reason not to apply KAP to social actions insofar as it applies to individual actions. In other words, it should be rational for a group G to Φ if G knows that p. The challenge is to clarify who is supposed to know that or act on p within a group. The inflationist thesis (Bird 2010) argues that no member of the group needs to know that p for the group to know that p. The summativist thesis (Goldman 2014) argues, on the contrary, that a group knows that p if its members know that p. Lackey (2020) argues an intermediate position according to which a group knows that p only if its operative members know that p. I will show that neither of these positions is satisfactory. I shall argue for an alternative position based on Weber (1922) and Boudon (2007)'s methodological rationalism (MR) and knowledge-first epistemology (KF). According to MR, social action is the action of the ideal member of a group. According to KF, knowledge is the mental state of an ideal rational agent. Thus, according to RM and KF, asking whether a group knows that p is equivalent to asking whether the ideal-type member of the group knows that p. I shall call this position Idealized Social Knowledge (ISK). ISK both avoids the shortcomings of the inflationist and summativist thesis and offers a unified epistemology based on knowledge and action rather than belief and justification.