Both free boundary problems and optimal design problems involve unknown or variable domains. The describe a direct approach extending the formulation of the problem to a larger fixed domain without modifying the differential operator. This argument is based on an approximate controllability-type property.