Abstract: This study challenges the idea that categories have little relevance for understanding how innovations are valued. Categories scholars have had difficulty accounting for such contexts because valuation is often based on forward influence rather than hierarchical rankings and consumption decisions. In comparison, innovation studies have focused on the link between an innovation’s attributes and its value, while largely overlooking the effects of broader socio-cognitive structures such as categories. We address this gap by combining insights about categories and recombinant search to predict when an innovation is likely to be noticed and built upon by others within the same versus different categories. We then consider how a category’s position within a classification system affects the likelihood of the innovations therein being noticed in cross-category search, and how this relates to overall influence. We test our predictions with a quasi-experiment that uses data from 8,823 patent families to compare how the influence of the exact same innovation varies based on its categorization in different patent systems. This approach isolates the role of categories in valuation and shows that considerable variance is explained by an innovation’s position within a given category and by that category’s position relative to others in a classification system.