Abstract : Research in HCI has produced many novel interaction techniques aimed at improving the usability of graphical applications. Yet very few make it into industrial products. This may be due to the difficulty of assessing the actual value of a technique and the required efforts to implement it. This thesis introduces three tools that can be used in synergy to ease the adoption of new interaction techniques by designers and developers. The first tool, Complexity of Interaction Sequences (CIS), is a model to describe an interaction technique and predict its performance in a given context of use. CIS abstract level is high enough to try different alternatives and evaluate them to make relevant preliminary design choices. The second tool, TouchStone, is a platform to help researchers to conduct controlled experiments. TouchStone allows to design an experiment in an exploratory fashion and to reuse components of previous experiments to ease the process of conducting an experiment. TouchStone also acts as a repository for empirical results so it is also useful for designers to assess the performance of techniques that have already been tested. Finally, SwingStates is a toolkit that introduces a structured graphical model and control structures to implement interaction techniques that differ from standard widgets. SwingStates is an extension of Java Swing, a toolkit that is widely used by industrial developers, so it allows developers to try new interactions easily while staying in a familiar programming environment.