Error correcting output codes (ECOC) are commonly used to reduce multiclass classification tasks to multiple binary classification subproblems. In ECOC, classes are represented by the rows of a binary matrix, corresponding to codewords in a codebook. Commonly, given a codebook, codewords are implicitly assigned to classes arbitrarily. In this paper, we show that the traditionally-overlooked codeword-to-class assignments play a major role in the performance of ECOC. We demonstrate that assigning similar codewords to similar classes, induces easier subproblems while preserving other favorable properties of a codebook. Our experiments highlight the importance of our observation on ECOC applied in different regimes. Notably, we find that similarity-preserving assignments improve performance significantly for existing codebooks designed especially for large scale tasks (extreme classification). We conclude that carefully assigning codewords to classes should be an inherent step in ECOC schemes.