In this study Mo-V-Te-Nb based multi-component catalysts were designed and tested using combinatorial and high-throughput methods. Based on the composition of the M1 matrix phase new compositions were designed containing additional promoters Mn, Ni, W and In. The following promoters were tested too in preliminary experiments and disregarded because of the detrimental effect on acrylic acid yields: Cu, Sb, Fe, Sm, Sn, Bi, Co, and Cr. In addition, citric acid in different ratios was used in the synthesis as a structure-directing agent. An optimization variable has been defined as the molar ratio of a given component to Mo in the synthesis mixture. Consequently, the experimental space had eight variables. The discrete levels of variables are established in such a way that the size of the multi-dimensional experimental space was in the range of 200 000 theoretical experiments. Five generations were designed using an optimization platform consisting of artificial neural networks and holographic optimization algorithm. Altogether 215 catalysts were prepared and tested. The elite list in each generation was created according to the yield of acrylic acid (AA). The yield of AA over the best catalyst after five generations was 59%. On the basis of holographic maps correlations between the composition of the synthesis mixture and yield of AA were visualized. Two catalysts families amongst the good performing catalysts have been distinguished that differ from each other in their low and high V content as referenced to molybdenum. (C) 2013 Elsevier B.V. All rights reserved,