A reasonable question. As
@46AND2 says, it's the patterns of similarities and differences that are very hard to explain except by common descent. Picking up their example, when we see a particular gene that has been disabled by an identical mutation in several species, the odds are overwhelming that the species will cluster together on the phylogenetic tree, because the mutation happened only once and was inherited by closely related species. Ditto for the insertion of a virus into the DNA -- the same insertion found in multiple species will be found in ones that could have inherited it from a common ancestor, and will not be found in more distantly related species.
We can also look at genetic differences that have no effect on the organism, like "synonymous mutations", that is, changes to the DNA that don't affect the protein produced by a gene. These typically have minimal effect on the organism. We find that closely related species have very few synonymous differences, while more distant species have a lot. There is no functional reason for this patter. If we look at differences overall, we can also see that they look exactly the way we would expect if they had been caused by lots of mutations. I've written up this piece of evidence in more detail
here.
