The emergence of social media increases the need for the recognization of social influence mainly motivated by online advertising, political and health campaigns, recommendation systems, epidemiological study, etc. In spreading processes, it is possible to define the most central or influential vertices according to the network topology and dynamic. On the other hand, the least influential spreaders have been disregarded. This paper aims to maximize the mean of information propagation on the network by recognizing the non influential individuals by making them better spreader. Experimental results confirm that selecting 0.5% of least influential spreaders in three social networks (google+, hamsterster and advogato) and rewiring one connection to some important vertex, increase the propagation over the entire network.