Transition metal-catalysed reactions are often strongly dependent on ligand and selection, among other factors, which makes discovering the ideal metal/ligand/solvent combination demanding. Furthermore, the effect of ligand and solvent choice is often subtle and unpredictable. This perspective describes how two statistical techniques, design of experiments (DoE) and principal component analysis (PCA), can be combined to guide the decision-making process. The general approach to using these techniques is described, and illustrated with a brief worked example on challenging 'borrowing hydrogen' chemistry. The unique combination of DoE and PCA is a useful decision-making tool to support the selection of ligands and solvents for challenging catalytic reactions. (c) 2014 Society of Chemical Industry