Exploratory approaches to materials discovery have been used extensively for the formation of new phases with unanticipated compositions, structures and properties.  Exploratory reactions are generally designed to articulate formation principles of the target compounds and can often require many iterative experiments to optimize conditions.  The use of exploratory reactions in the formation of new templated metal oxides will be discussed in two ways.  First, a hierarchy of influences that control the formation of such phases has been identified.  These influences, which include reactant concentrations, charge density matching, symmetry, hydrogen-bonding and sterics, are controlled by reactant structure and the experimental conditions employed.    Second, in an effort to increase the rate by which new compounds are formed, machine learning techniques have been used to more fully develop our chemical intuition and focus initial reactions.  Efforts include database construction, the use of cheminformatics techniques to derive reactant and reaction properties and testing and evaluation of a machine-learning derived model for crystal growth.