<p>This study investigates how emotional valence and emotional trajectory in donation-based crowdfunding messages influence actual donation behavior. Using a large dataset from two major crowdfunding platforms—GoFundMe and DonorSee—we apply automated sentiment analysis to examine two dimensions of emotional appeal: (1) mean emotional valence (positive or negative tone) and (2) emotional trajectory (the flow of valence over the course of a message). Results show that messages with overall negative valence and a positive emotional trajectory (i.e., shifting from negative to positive) were associated with greater donation amounts, supporting theories such as the expectancy-contrast model and the negative-state relief model. Further, we find that the interaction of these two dimensions significantly predicts donations on DonorSee, though not on GoFundMe, suggesting platform-level differences in emotional responsiveness. This study contributes to the literature by reconceptualizing emotional appeal as a dynamic construct and by offering new methodological tools to capture emotional progression in real-world narratives. It also provides practical implications for fundraisers seeking to craft more compelling campaign messages in increasingly saturated digital donation environments. By integrating emotional valence and emotional trajectory, this study advances a more comprehensive framework for understanding the mechanisms behind charitable giving in the era of peer-to-peer fundraising.</p>