We joined genes and aligned them with different algorithms (ClustalW, Mafft, Muscle, Geneious translation-based). We also aligned non-annotated genomes to assess the effect of data partitioning, observing that the alignment of non-annotated genomes results in different genes aligning together, generating several false hypotheses of homology. We assessed the effect of increasing sampling, adding outgroup, partitioning the dataset, and analyzing the data employing different optimality criteria for phylogenetic analysis including maximum likelihood, Bayesian inference, and parsimony. Some clades, such as the clade encompassing all tick-borne viruses and the clade of Mammalian tick-borne viruses, are dependent on the annotation and presence of outgroup sequences. Increasing sampling allowed us to visualize missing links within the phylogeny of Flavivirus and recognize that previously non-monophyletic groups such as ZIKV + DENV form a clade. There is a higher degree of convergence among different optimality criteria than between annotated and non-annotated datasets or between datasets with and without outgroup sequences, indicating that annotation is critical in the phylogenomics of Flaviviridae. Furthermore, our results show that the annotation of Flaviviridae genomes is attainable with reasonable time and reduced resources, minimizing spurious homology statements, and leading to significant changes in the tree topology.