Previous research suggests that a characteristic of the construction industry is a lack of technological innovation. Since this is seen as a problem, much theoretical development within construction management focuses on explaining the lack of innovation. Less effort has been expended on using such explanatory theories for investigating those rare exceptions in which construction firms succeed in the unlikely: successfully developing a new technology. This article makes use of the recently suggested framework by Dubois and Gadde. They describe the construction industry as a 'loosely coupled system' with four types of couplings, discuss why the particular mix of couplings in the construction system leads to a lack of innovation, and suggest types of couplings that construction firms should experiment with and change in order to boost innovation. A case study of a contractor developing a new technology is presented in terms of Dubois and Gadde's concepts and implications. The findings partly support and partly contradict their hypotheses. It seems that innovation is possible even if only a few of the existing couplings are changed. The most important changes relate to the tightening of intrafirm sequential inter-project couplings enabling learning from project to project, and contract-related couplings especially the design-construction interface.