The diffusion of national standardized testing, large-scale survey assessments and the promotion of policies of self-evaluation are making large amounts of data on education systems available and transforming schools into collecting units for a notable range of educational, institutional and socioeconomic indicators. The datafication and related digital technologies for collecting, analysing, retrieving and displaying data activate, at least in principle, new spaces of visibility and forms of school data-based managerialism (Williamson, 2017). While the policy of transparency is oriented to the development and consolidation of data-based school governance (Selwyn in "European Educational Research Journal," 15, 54-68, 2016), its implementation in practice remains an open question. It solicits the analysis of the enactment of school data infrastructures to understand their mobilization in the governance of schooling. Schools can align with digital technologies and data, or they can resist these in many ways. By drawing on a multi-sited ethnography on the development and consolidation of the digital governance of education in Italy (Landri, 2018), I will display how schools can align, imitate, and fabricate their data, use them partially and instrumentally, "gaming," or opting-out from the current regime of accountability. These findings complexify a typology of resistance to the digitalization proposed by Souto-Otero and Beneito-Montagut ("European Educational Research Journal," 15, 14-33, 2016). They trouble the either/or logic that presents 'alignment' and 'resistance' as they were different alternatives to underline the subtleties of the policy enactment of the data-based school governance. The investigation illustrates that the space of the school agency is not entirely lost: the destiny of the digital governance of education; in other words, is not inevitable. It draws attention to the singularity of the schools concerning the policies of digital accountabilities. The singularity is a capacity to react that depends ultimately on the sedimented circuits of knowledge. These latter ones orient in different ways how noticing, interpreting and drawing conclusions from data.