In this paper we argue that open government data as such do not lead to more transparent and accountable government and corporate actions but rather that different stakeholders engage in different, at times contesting, data practices by mobilising and repurposing different kinds of data. In so doing, they reconfigure transparency and redistribute accountability. For this purpose we analyze the data practices of a diverse set of stakeholders (ranging from NGOs, Brazil’s Public Prosecutors Office to JBS, the world’s largest meat packer) around the Beef Moratoria, signed in 2009 between Brazil’s Public Prosecutors Office and the four largest meat packers from the state of Parà. The aim of this Moratoria was to ensure deforestation-free beef supply chains. In contrast to the linear and “successful” narrative suggested by some NGOs and private companies, the emergence of this transparency regime has been conflictual and tentative. NGOs have found new and creative ways to produce transparency by combining and repurposing different datasets, while companies directly or indirectly have endeavored to create opacity by breaking key reference points (e.g. fiscal ID number of farmers) that are necessary to render visible transactions within the supply chain. Therefore, while NGOs strived to detect deforestation within the supply chains via different datasets, the companies (aided by some parts of the government) strived to erase the possibility of traceability while adhering to the rhetoric of transparency by making available increasing amounts of raw data. From this, the paper argues that data play a central role in producing both “clean” and “dirty” beef supply chain, and as such the strategic availability and (re)interpretation of certain datasets became a key place of contention.