This study investigates the impact of digital transformation on visitor satisfaction, engagement, and recommendation intentions at World Heritage Sites in Chinese coastal cities. A survey-based quantitative research design was employed, collecting data from four hundred and two respondents across eight cities using systematic random sampling. structural equation modeling (SEM) was applied to analyze relationships among digital participation, perceived authenticity, visitor satisfaction, and recommendation behaviors. Results highlight that technologies such as VR and AR enhance satisfaction, engagement, and authenticity, driving recommendation behaviors. Extending Cultural Authenticity Theory and satisfaction–loyalty frameworks, the study emphasizes the dynamic interplay of digital tools and cultural narratives. Practical recommendations focus on implementing culturally sensitive, interactive digital strategies to strengthen heritage tourism’s sustainability. Future research is encouraged to explore emerging technologies like AI and the metaverse.

Digital interpretation at heritage sites is usually measured as a single experience construct. This study treats it as emitting two classes of evidence, procedural evidence about how the system behaves and substantive evidence about what it asserts, which visitors weight separately when forming satisfaction. The model is tested on 1000 visitors at World Heritage attractions in China and 989 in the United States, using multigroup structural equation modeling after establishing configural, metric and scalar invariance. The two samples differ moderately in the size of the indirect effect of digital provision on satisfaction and on recommendation intention, and almost completely in its composition: procedural evidence carries 75% of the effect on satisfaction in China and 9% in the United States, where authenticity judgements dominate, and satisfaction converts into recommendation intention about twice as strongly among Chinese visitors. Monte Carlo intervals show that the difference in composition is estimated with precision and that the totals are not statistically equivalent. The composition of an effect, read alongside its size, emerges as the more informative object of comparison for heritage managers deciding whether to invest in the performance of a digital system or in the credibility of its content; the allocation implications are advanced as propositions rather than as tested rules.