In the spring of 2020, the public authorities of more than 40 countries around the world placed their hopes in digitalized Covid-19 contact tracing via mobile application solutions – a digital technology development promising to alleviate manual contact tracing, encourage individual contacts-cases to get tested and confined, and thereby curb the contagion to facilitate a return to the normal. The realization of such promises was based on a script in which different actors, human and non-human had a role to play. Considering that adoption by a critical mass of users was essential for digital contact tracing to be effective, two major issues were promptly identified by the different design teams: the reluctance from citizens worried about their privacy and the technical opacity of private mobile telephone operators, whose systems could "kill the apps”. We examine in this paper how different countries have implemented their respective solutions with respect to these potential failures (Vinck, 2017). We draw insights from comparing cases: France, Switzerland, Japan, and Colorado in the US. In this light, we depict how the promises to enroll actors were problematized, focusing on the consequences of valuing resistance and non-adoption, as well as design and implementation misalignments. We unfold how the perspective of these two obstacles has determined the design strategies in technical (centralized/decentralized systems – monitoring and maintenance competencies), organizational (public/private - local/national service provider), and functional (services offered) terms, and how finally, this led to losing sight the promise of epidemic efficiency. We associate the different responses in this "fears-of-failure" driven design to deep differences in the digital culture of the health systems, to digital sovereignty stakes and to territorial/organizational lack of coordination. This paper thus contributes to research on the digitization of public action, especially those underlying organizational (Peliza, 2016) and governance (Al Dahdah, 2022; Wesselink, and al., 2015) issues.