The current article questions how experimentation in policy design plays out in practice. In particular, it is interested in understanding how the content and process of policy-design experiments affect their outcomes. The article does so by building on an original study into 31 real-world examples of experimentation in policy design in the building sector in Australia, the Netherlands, and the United States. All examples aim to improve the environmental sustainability of the building sector. The article finds that these 31 examples have attracted moderate to substantial numbers of participants (policy outcome HO.i), but have not achieved substantial numbers of buildings built or retrofitted with high levels of sustainability (policy outcome HO.ii). By carefully unpacking these policy designs into a number of key characteristics, it finds that this mismatch between the two outcomes may partly be explained by flawed policy-design processes. The article concludes with the main lessons learnt and provides some suggestions on how to improve experimentation in policy design.