Best Before It Goes Stale: Optimal Update Policies for Keeping Data Fresh

In present-day data markets, sellers increasingly monetize data whose economic value erodes as the underlying information becomes stale. In this paper, we study how a monopolist data seller should jointly manage update frequency, update effort, and the price of access to a dataset over a finite selling horizon. The model captures three salient features of data monetization: data quality decays between updates, update effort stochastically improves quality, and buyers' willingness to pay depends on realized quality. To reflect heterogeneity across data markets and update technologies, we study three canonical regimes in which returns to effort are increasing, decreasing, or S-shaped&nbsp;<span>–</span><span></span>&nbsp;that is, initially increasing and then decreasing.&nbsp; <div> <br> </div> <div> Our analysis characterizes the optimal update policy across these regimes. We show that the structure of optimal effort depends critically on how effort translates into quality improvement. When returns to effort are increasing, optimal effort is bang-bang (all-or-nothing); when returns are decreasing, optimal effort is state-dependent; and when returns are S-shaped, the seller updates only in diminishing-returns region. We then contrast the dynamic update policy with a static benchmark that fixes effort across periods. We show that dynamic adjustment increases profit by up to 20%, especially when setup costs are low, marginal effort costs are high, update outcomes are noisy, and the selling horizon is short. By mapping the model primitives to practical data-selling settings, our study shows that profitable data monetization requires coordinating prices and update decisions with the cost, uncertainty, and returns of restoring data quality. </div>