The rapid expansion of social networks has brought with it questions on their economic sustainability. The most promising means of value extraction from such networks is their use for targeted advertising of goods and services. We consider goods that are consumed periodically, and would like to create a two-tier price structure for maximum revenue extraction through selective discounts. We show that users' consumption capacity can naturally be divided into two parts -- marked price and discounted -- and propose a scheme to ensure that users are discounted enough that the entire demand is realized, but no so much that profits are  decreased. Our scheme incentivizes forwarding of discount  coupons using a social network, and uses ideas of back-pressure routing to create an implicit learning algorithm that provably extracts the entire surplus in a distributed fashion.  We validate our analytical results using simulations.