Abstract
Digital-skills training is increasingly used to advance agricultural digitalisation; however, participation and learning do not necessarily translate into application within production and business routines. This exploratory study investigates the training-to-application gap using 156 questionnaire records from agricultural actors in China recruited via convenience sampling through WeChat-based networks. The outcome is a broad, self-reported, four-category indicator of post-training digital application based on retrospective assessments of training experiences over the preceding two years, rather than a technology-specific measure of adoption, sustained use, or performance. Grounded in training-transfer theory, the analysis examines whether perceived content relevance, an inverse scenario/practice-barrier index, and course-linked post-training support are associated with higher reported application. Cumulative-logit estimates yielded no statistically detectable associations for the three indicators in the baseline model (likelihood-ratio test p = .995; McFadden pseudo-R² = .001). The conclusions concerning H1–H3 remained unchanged across alternative predictor coding, separate modelling of the two barrier components, and expanded covariate adjustment. An exploratory secondary finding indicated higher reported application among production-oriented respondents than among market- or service-oriented respondents; however, the broad outcome may not be fully comparable across business contexts. The study's principal contribution lies in conceptually distinguishing training participation, learning, initial trial, continued use, routinisation, and operational value. On this basis, the article advances a literature-informed Conditional Conversion Framework for future testing, in which technical feasibility, economic viability, platform and market fit, and ecosystem implementation support may enable or constrain progression across application stages.