Mining maximal groups from spatio-temporal data of mobile users is a well-known problem. However, number of such groups mined can be very large, demanding further processing to come up with a readily usable set of groups. In this paper, the authors introduce the problem of mining a set of K maximal groups which covers maximum number of users. Such a set of groups can be useful for businesses which plan to distribute a set of K offers targeting groups of users such that a large number of users are covered.