An Explore-Then-Commit strategy for revenue management via sequential estimation

We introduce and analyze a sequential Explore-Then-Commit (ETC) strategy for revenue management that balances the trade-off between learning about demand and maximizing revenue. Assuming the demand function belongs to a parametric family with unknown parameters, we derive a closed-form stopping rule based on observed Fisher information that allows the decision maker to adaptively determine when to stop price experimentation and commit to a price, eliminating the need to specify a fixed sample size in advance. We show that the proposed algorithm outperforms any fixed sample size ETC procedure in terms of regret and achieves asymptotically optimal min–max regret, a robustness guarantee that no predetermined sample size procedure can attain. We extend the algorithm to a parametric choice model with multiple products and to a network revenue management setting with nonparametric demand and finite resource capacities. Numerical experiments confirm the practical benefits of our approach across all settings.