Inflation expectations surveys are receiving increasing attention. There is no optimal approach and often limited discussion of key characteristics of individual surveys. We use a South African dataset to argue that survey design should be given far more attention as it may undermine our ability to use the data with confidence. Users of survey data need to understand existing differences in survey design and the extent to which survey data reflect decision-making shortcuts under uncertainty as opposed to a true belief about what the public more generally really thinks expected inflation will be.