Social media has become instrumental in mobilizing around political issues and social movements. While many studies have examined the linguistic trends in protester mobilization, there is limited analysis of the network dynamics that support sustained protest participation. This paper presents a cross-national multi-method analysis focusing on social network metrics and their relationship to protester attrition in two datasets: the #Nirbhaya protests in India and #BlackLivesMatter in the USA. Our findings reveal that network position metrics – particularly degree centrality, eigenvector centrality, and PageRank – are strong predictors of sustained participation, often outperforming emotion-based predictors. Users with higher degrees of connectivity and centrality within the protest network are significantly more likely to remain engaged throughout the movement. These results suggest that the formation of social ties within protest networks serves as a crucial mechanism for maintaining activist commitment beyond initial emotional reactions. We conclude with theoretical implications for understanding collective identity formation in digital activism, implications for protests in an AI-mediated, high-misinformation and high-polarization social media environment.