We show that regulators can successfully adapt to market conditions like demand and supply, even if those conditions are unpredictable and evolve constantly. In our model of adaptive monopoly regulation, the regulator receives market data that it can use to revise policies over time. Building on the literature on learning in games, we develop new solution concepts for the firm that generalize Bayesian rationality but require no prior knowledge to satisfy. Our results culminate in a foundation for customer-first regulation, which uses taxes and data-driven subsidies to incentivize the firm to prioritize welfare over profits.