This paper aims to explain differences in structural change paths across countries using a measure of relatedness between economic activities. It uses World Input-Output Database data to construct a matrix measuring relatedness between 84 activities based on comparative advantage. Clustering and network analysis of this matrix reveal patterns of related activities. Econometric models test if relatedness predicts future comparative advantage gains, finding employment in highly related activities promotes structural upgrading. However, moving between low- and high-skill activities is difficult. Overall, the paper presents a novel approach using international input-output data to empirically study patterns of structural change.