This article studies the rise of market research and opinion surveys in Sweden from the 1930s and onward. By focusing on the promotion of empirical knowledge by actors between the academic and the commercial world, such as the economist Gerhard Törnqvist, the article shows how new practices of classifying consumers into social classes were established among marketers and advertisers. These approaches were passed on to the Swedish Gallup, which produced opinion surveys from the early 1940s. The final section of the article charts Swedish newspapers’ preoccupation with classifying practices of the Swedish population into classes. The article investigates market research and opinion surveys through following “the social life of methods”, a theoretical perspective that sees methods of knowledge as political. I analyze how a class taxonomy constructed by the Swedish statistical bureau in 1911 migrated and became productive in the commercial sector starting in the 1930s. These taxonomies could be called “difference machines” in that they repeatedly produced statistical differences as new knowledge.