While the use of numbers in governance has a long history, the kinds of numbers we now produce enable a range of new possibilities for monitoring, regulation and policy decision-making. Global policy actors are now calling for a steep increase in investment in education data. The growing trust in numbers has been critiqued by education policy scholars and social scientists, who have pointed to the reductionist nature of numbers and the dangerous decontextualisation of information which are leading to detrimental policies. The entry of big data poses even more complex epistemological and ontological challenges, many of which we do not fully understand as yet. This paper acknowledges these challenges, and at the same time speculates that big data might afford unique possibilities for new relational theories that may lead to better policy decision-making.