The problem with Dr Caldwell’s approach is:
he has a pre-conceived notion of snake evolution, demonstrated @28:00, when he put a rattlesnake head on a varanid body. Varanids are not snake ancestors (see below).
A better approach is to include as many extant and extinct candidates as possible, put them all into a matrix, then let the software recover the lineage in a resulting cladogram. This is a 2017 video, so the software was available back then.
For a competing analysis that recovers snake ancestors back to the origin of Cambrian chordates, see the large reptile tree: http://reptileevolution.com/reptile-tree.htm
Among extant taxa gekkos are closest to snakes. Early Cretaceous Chometokadman is the last common ancestor of both clades. Early Cretaceous Norellius and Late Cretaceous Tchingisaurus are the most basal member of Proserpentes, which currently includes 12 more taxa. The last common ancestors of the basal snakes Najash and Dinilysia are Middle Cretaceous Barlochersaurus (the size of a paper clip) and tiny Early Cretaceous Tetrapodophis. They had larger aquatic precursors, including Adriosaurus, Pontosaurus and Primitivus. At their genesis snakes split between medium-to-large terrestrial + aquatic snakes and ever smaller burrowing snakes.
Genomic tests too often diverge from phenomic (trait-based) tests and genomic tests do not include fossil taxa. Why the two test methods diverge is not settled, but everyone recognizes this issue. Why paleontologists ignore fossils while trusting genomic tests is another problem not yet settled.
Phylogenetic miniaturization at the genesis of major clades is common among chordates.
The Pterosaur Heresies 