Rotational grazing (RG) has attracted much attention as a cornerstone of multifunctional agriculture (MFA) in animal systems, potentially capable of producing a range of goods and services of value to diverse stakeholders in agricultural landscapes and rural communities, as well as broader societal benefits. Despite these benefits, global adoption of MFA has been uneven, with some places seeing active participation, while others have seen limited growth. Recent conceptual models of MFA emphasize the potential for bottom-up processes and linkages among social and environmental systems to promote multifunctionality. Social networks are critical to these explanations but how and why these networks matter is unclear. We investigated fifty-three farms in three states in the United States (New York, Wisconsin, Pennsylvania) and developed a stylized model of social networks and systemic change in the dairy farming system. We found that social networks are important to RG adoption but their impact is contingent on social and spatial factors. Effects of networks on farmer decision making differ according to whether they comprise weak-tie relationships, which bridge across disparate people and organizations, or strong-tie relationships, which are shared by groups in which members are well known to one another. RG adoption is also dependent on features of the social landscape including the number of dairy households, the probability of neighboring farmers sharing strong ties, and the role of space in how networks are formed. The model replicates features of real-world adoption of RG practices in the Eastern US and illustrates pathways toward greater multifunctionality in the dairy landscape. Such models are likely to be of heuristic value in network-focused strategies for agricultural development.