There are two major constraints for experimental methods in the high added value products industry: the complexity of the products, which has as a consequence an incomplete knowledge, and the high cost, which induces a minimum number of runs. An approach that takes into account these two constraints is a progressive strategy. Several strategies of this type exist: one factor-at-a-time designs, group screening designs, and sequential bifurcation. We propose a “dichotomic designs” approach based on sequential bifurcation that presents many advantages. First, it allows one to find the effective factors with a few runs. Second, all runs can be reused to constitute an orthogonal fractional design if an amount of uncertainty remains. Last, this strategy can even be applied without drawing up the precise list of all factors to be studied.