Dynamic Pricing and Product Quality with Review-Driven Learning

Online product reviews play a pivotal role in empowering consumers by reducing uncertainty about product attributes, a phenomenon widely studied from the demand perspective. However, the supply-side implications remain less understood—particularly how firms can leverage consumer reviews to develop an integrated pricing–quality strategy. To address this gap, we examine a firm’s dynamic pricing and product-quality decisions over two selling periods, where consumer reviews play a central role in shaping the market’s perception of a new experience good. Both the firm and its consumers are initially uncertain about the product’s perceived (market-based) quality and rely on early reviews to update their beliefs. Our analysis identifies two critical review metrics— volume and valence —as jointly shaping the firm’s optimal strategy. Review volume reflects initial sales, while valence measures the average rating. Together, these metrics generate what we call the learning precision effect , whereby review information enhances the accuracy of quality inference and, in turn, guides the firm’s dynamic pricing and quality decisions. We find that, without the option to refine (adjust) quality after launch, the firm prefers a higher initial product quality and a lower introductory price to boost review volume and improve learning. When post-launch quality refinement is feasible, the learning precision effect intensifies, as the firm further increases initial quality to enhance learning from reviews. However, the resulting optimal pricing strategy in the first period can depart from conventional intuition. Depending on market conditions, the firm may either raise or lower the introductory price—relative to the case without quality refinement—in order to enhance review generation. A notable outcome is that the firm’s optimal strategy not only increases its overall profit but also improves consumer surplus for both early and late buyers. Although quality refinement is often expected to favor the firm at consumers’ expense, our results show that learning from reviews can generate a genuine win–win outcome. Finally, we extend our model to incorporate under-reporting bias, uninformed consumers, nonzero marginal quality costs, and alternative distributions of perceived quality, and find that our main insights remain robust across these variations.