How to Price Fresh Data with Strategic Users

The interests in obtaining fresh data in real-time applications have facilitated fresh data markets. However, existing works on designing fresh data markets have ignored strategic users. Being strategic means that users can optimally time their data purchases, which affects markets' profit. In this paper, we study a fresh data market, where strategic users with heterogeneous data valuations stochastically arrive over time. The strategic users decide the time of data purchase, considering the evolution of data freshness and prices, while the platform decides the data pricing policy over time to maximize its profit. We first consider a dynamic pricing policy, where the platform offers a price menu to each arrival user. The analysis is technically challenging, as heterogeneous users face different integer programming problems in optimizing their data purchase time, making direct optimization of data prices infeasible. To tackle the challenge, we adopt a mechanism design approach. We show that the direct mechanism design problem relax the original problem and obtain the optimal pricing policy analytically. Next, to reduce the implementation complexity, we study a single pricing policy, where the price is fixed over time. We derive the optimal single price analytically in a two-period refreshing model. Perhaps surprisingly, although strategic users have more purchasing options than non-strategic users, users who behave strategically may be worse off. Simulation results show that, although a platform refreshes the data less frequently in the presence of strategic users than facing myopic users, it can earn up to 5 times higher profit.