The Hidden Geography of Housing Demand Exposes Policy Failure: Evidence from Billions of Housing Searches

House prices are widely used across the social and environmental sciences to inform crucial business and policy making decisions meant to reduce spatial economic inequalities and future-proof locations at risk from environmental shocks. Yet prices are biased by supply and only observed for the selected properties that transact. This paper is the first to introduce a direct measure of latent location demand, using billions of housing searches in Great Britain between 2019 and 2024. Across three causal applications spanning public health, environmental science and economics, we show that searches reveal demand that transaction data either obscure or cannot measure. The COVID-19 pandemic triggered a "race for space", but demand for private greenspace returned to pre-pandemic levels before new supply could respond, leaving planners and developers chasing yester day's preferences. Local flooding generates temporary dips in demand, suggesting people are myopic. New housing supply developments do not induce increases in location demand, implying new town development policies are misguided. In all three cases, transaction data miss the fundamental behavioural response, leading to suboptimal policy design. Finally, we use the search data to construct a measure of excess housing demand for 235,243 micro neighbourhoods and release it through the WhereToBuild mapping tool.