Generative AI Affordance Actualization, Digital Literacy, and Business Performance in China’s County-Level Rural E-Commerce: A Configurational Systems Approach

Generative artificial intelligence (GenAI) tools are cheap and easy to obtain, yet small online sellers gain very different amounts from them. We ask why, using a survey of 401 GenAI-using agri-related e-commerce merchants in Shuyang County, Jiangsu, a Chinese county-level rural e-commerce economy. Drawing on affordance actualization theory and a socio-technical systems view, we test whether wider and deeper use of GenAI improves performance by strengthening three capabilities: operational execution, content marketing, and customer responsiveness. A path model (partial least squares structural equation modeling) shows that more extensive use goes with stronger capabilities, and stronger capabilities with better performance. Digital literacy plays a double role: it is linked to stronger capabilities in its own right, and it is linked to a larger payoff from GenAI use, which is small for merchants with weak skills. A configurational analysis (fuzzy-set qualitative comparative analysis) looks for combinations of conditions. Under the pre-set thresholds, the only combination consistently linked to high performance joins broad use, deep use, digital literacy, and structured training; a less strict threshold suggests that village/township and other merchants may reach it through different combinations, both including digital literacy. Support programs for such sellers should combine tool use with digital-skills training.