Online reviews: information content, drivers, and platform design

Online ratings emerge from a multi-stage process that can systematically distort their informational content. We develop a unified framework decomposing the rating process into distinct components: experienced quality (driven by intrinsic quality, seller effort, and price), expectations formed prior to consumption, contextual influences, strategic distortions, idiosyncratic tastes, and selection into reviewing. This decomposition organizes a growing theoretical and empirical literature and clarifies how seemingly disparate findings—from fake reviews to disappointment effects to selection biases—relate to distinct stages of the data-generating process. Our framework also provides a lens for evaluating platform design interventions: effective policies target specific components of the rating process, yet many distortions remain difficult to address without introducing new trade-offs. We highlight open questions where further research is most needed.