As public policy issues are frequently considered as "wicked problems" that require the setting up of co-production devices with citizens and users, design-based approaches are increasingly mobilized for the design of public services and policies. One reason is that design is seen as particularly appropriate considering the complexity and uncertainty characterizing contemporary contexts. These few diagnostic elements seem to justify the development of collaborative innovation by design methods within the public sector, and the creation, all over the world, of “policy labs”, defined as structures with an organizational identity and relative autonomy, while being connected to public organizations. However, beyond this generic term, the various organizational forms and practices that are implemented in order to institutionalize these innovation processes, i.e. to make them sustainable and legitimate, have been largely neglected by academic research so far. In this article, we explore the institutionalization process of Innovation-Oriented Teams (IOTs), located at different territorial levels and relying on the use of methods partly inspired by design. This analysis first aims at better understanding the organizational construction of “policy labs”: How do these teams emerge and perpetuate? Do all IOTs eventually become “labs”? Moreover, we investigate the link between the organizational construction processes and the ways of appropriation of design practices: How is the conception of design IOTs manipulate a means or an obstacle to their institutional anchoring and development? Our analysis is based on the empirical study of three IOTs at three different administrative scales in France (National state, Region, Department). These IOTs were first chosen because they all use design-oriented approaches. Secondly, they have the advantage of having several-year existence, which enables us to study an institutionalization process deployed over time. Finally, it seemed interesting to locate our analysis at different territorial levels, the responsibilities of the three administrative host organizations being quite different, which should a priori impact the nature of their activities. We exploit different qualitative sources: interviews with IOT members, administrative officers involved in their birth and projects, with external stakeholders collaborating with these teams; documents (internal or public) relating to their birth, doctrine and the projects carried out; observation of key events that punctuated the trajectories of these IOTs. Beyond similar characteristics of emergence, we show that the anchoring, development and legitimacy of IOTs vary, and largely depend on organizational tactics. Above all, their institutionalization relies on a process of “metabolization” of design methods that is contingent to each IOT and that allows it to enrole key actors such as elected representatives and managers. We conclude by highlighting the pragmatic and careful aspects of their strategy – and their plastic (or instrumental) approach of design – that enables them to move forward, survive and follow an institutionalization trajectory. Contrary to policy labs previously studied by academics, these IOTs have chosen to dissolve in the administrative organization to better disseminate the effects of the doctrines of innovation they carry. To that extend, they would thus work more to hybridize bureaucracies than to replace them.