From Doubt to Devotion: Trials and Learning-Based Pricing

In the digital marketplace, a prominent feature of software products and digital services is the implementation of dynamic pricing mechanisms that are fine-tuned by consumer preference data. To fix ideas, consider a streaming service provider offering access to a library of entertainment content. Consumers are uncertain about whether the streaming library contains movies or shows tailored to their preferences but can become convinced of the service's value from finding content that resonates with their tastes. On the other hand, the pervasive collection of consumer data enables the streaming platform to predict whether the content matches a consumer and forecast the consumer's private experience. What pricing strategy the seller would take to leverage the buyer's ability to learn and data on buyer taste, and what are its welfare implications?

We address these questions by studying a dynamic informed principal problem where an informed seller designs a dynamic mechanism to sell an experience good. The seller has partial information about the product match, which affects the buyer's private consumption experience. As the first paper exploring the implication of an informed principal on dynamic mechanism design, we find a novel interaction between price discrimination against the buyer's learning and the belief gap about the learning process between both parties. In contrast to a static environment, having consumer data can reduce the seller's revenue in equilibrium as they optimize the dynamic mechanism using data forecasting the buyer's learning process.

In our framework, the ex-ante revenue maximizing mechanism always provides full access to the service and charges a price at the beginning; this prevents the buyer from gaining rent through information learned from their private consumption experience. However, this mechanism might not be supported in equilibrium due to the belief gap between the informed seller and the uninformed buyer. The seller with a high match value anticipates a greater likelihood of the buyer having a good experience compared to the buyer's own expectation. If the belief gap is large enough to offset the rent loss from the buyer's learning, it becomes profitable for the high-type seller to deviate to mechanisms that provide the skeptical buyer with limited access to the product and an option to upgrade if the buyer is swayed by a good experience. Depending on the seller's screening technology, this takes the form of free/discounted trials or tiered pricing, pricing mechanisms which are prevalent in digital markets. In equilibrium, the seller proposes the same such mechanisms regardless of their private information, leading to lower ex-ante revenue than selling the whole service ex-ante. The seller's signaling incentives further reduce their revenue as the high type tends to use dynamic mechanisms that maximize price discrimination to signal their type. Consequently, standard equilibrium refinements select equilibrium mechanisms with maximum price discrimination and lowest social efficiency. Thus, sellers may benefit from restraining themselves from using consumer data.

The full paper is available at https://arxiv.org/pdf/2311.00846.