Many common phylogenomic algorithms that were well-adapted to classify limited numbers of species have become increasingly intractable as large whole-genome sequencing datasets have emerged. Various novel approaches use characteristics of DNA sequences, including variations in codon usage biases, to establish the phylogenetic relatedness of species. Codon choice affects transcription and translational efficiencies, which can lead to differential protein expression and phenotypic variation that may be a target of selection. Several functional biases exist within genes, including the number of codons that are used, the position of the codons, and the overall nucleotide composition of the genome. Although recent algorithms capitalize on specific codon usage biases to improve phylogenetic tree inference, the phylogenies produced by these algorithms vary significantly and indicate different evolutionary histories. Therefore, we propose that gene-specific analyses of the phylogenetic signal of specific codon usage biases are required to best incorporate these biases in phylogenomic models.