Blade tips play an important role in the efficiency of gas turbines in minimising the adverse effects of the tip leakage flows. Various tip shapes such as squealers and winglets have been proposed and investigated for improving the performance. The advance of adjoint based methods have allowed for more flexible tip shape to be parameterized with hundreds or thousands design variables. However, the flexibility also increases substantially the dimension of the design space so that the searching for the optimum becomes a lot more costly. In this paper, we present a parameter reduction method based on the sensitivity obtained from the adjoint solution so that the optimiser can narrow down on the key design parameters for efficient shape optimisation without the constraints of pre-specified topologies.