We present an analysis of how the structure of a social network influences the diffusion of information in a viral marketing context. We performed diffusion simulations on a large number of real world and artificially generated network datasets. We analyze how the characteristics of a network and parameter settings like the selection of initial start nodes influences the diffusion. The results indicate that the network structure has a significant effect on the diffusion. Extreme cases show a difference in the diffusion of over 65%. Our investigation also proves that a viral marketing diffusion may be predicted without the knowledge of the whole network. We further provide useful recommendations for marketers which could be taken into consideration when marketing campaigns are conducted.