Why AI Needs to “Speak with Data”: The Impact Mechanism of Digitalized Descriptions by Virtual eWOM Senders on eWOM Effectiveness

Based on the Persuasion Knowledge Model (PKM), this research investigates how virtual electronic word-of-mouth (eWOM) senders’ message framing—numerical versus experiential—influences eWOM effectiveness across three experiments. We find that: (1) numerical descriptions from virtual eWOM senders significantly enhance eWOM effectiveness compared to experiential descriptions, while this effect does not emerge for human senders; (2) perceived diagnosticity mediates the relationship between message framing and eWOM effectiveness; and (3) product type moderates this effect pathway, with numerical descriptions showing stronger positive effects for search products than for experience products. This research enriches theoretical understanding of eWOM communication in interactive marketing and provides practical guidance for e-commerce companies to optimize their content marketing strategies.