We consider the problem of optimal pricing of a common-valueproduct in the presence of social learning effects. A new productreaches the market and agents obtain private signals that partiallyinform them about the value of this product. Agents decidesequentially whether to purchase this product. Before making their owndecisions, they also observe the purchasing decisions of agents whoacted previously and learn from those actions in a Bayesian rationalfashion. We address the problem of how a firm should price the productwhen taking social learning into account.

Our first result shows that firms do best asymptotically if the firmsselect prices that lead the customers to learn the true value of theproduct. We show how the firm can induce learning at a low cost byinducing a vanishing fraction of the agents to act according to theirprivate signals. We also show a lower bound on the agents' regret ofT^2/3 for a society of size T. We finally show a pricing policythat achieves this lower bound.