Social networks play a fundamental role in the diffusion of innovation through peers’ influence on adoption. Thus, network position including a wide range of network centrality measures has been used to describe individuals’ affinity to adopt an innovation and their ability to propagate diffusion. Yet, social networks are assortative in terms of susceptibility and influence and in terms of network centralities as well. This makes the identification of influencers difficult especially since susceptibility and centrality do not always go hand in hand. Here, we propose the Top Candidate algorithm, an expert recommendation method, to rank individuals based on their perceived expertise, which resonates well with the assortative mixing of innovators and early adopters in networks. Leveraging adoption data from two online social networks that are assortative in terms of adoption but represent different levels of assortativity of network centralities, we demonstrate that the Top Candidate ranking is more efficient in capturing innovators and early adopters than other widely used indices. Top Candidate nodes adopt earlier and have higher reach among innovators, early adopters and early majority than nodes highlighted by other methods. These results suggest that the Top Candidate method can identify good seeds for influence maximization campaigns on social networks.