Understanding of structural change is compromised because scholars do not clearly articulate the limits of the classification infrastructure (NAICS or GICS) that shapes empirical analysis. These limits are particularly salient in the study of innovation, an activity that by its nature challenges existing categories. Because innovative industries are often not part of the classification infrastructure, they are invisible in empirical analyses and in government statistics. This paper examines the classification of a population of highly innovative, often small, firms working in gaming devices, packaging, filtration, photonics, imaging, biomedical research and fabless semiconductor design. I find examples of knowledge integration, vertical disintegration and emerging industries that challenge both NAICS and GICS exposing their strengths and weaknesses.