The problem with this argument is that most of these are really just one example, because they are already known to be descended from a common source. In order to determine a true relationship, you have to go back to the earliest form of a word known, and you have to demonstrate that the relationship is not a coincidence. Namely, that you can see other examples of the same sound change in the two languages, and that neither language borrowed the word from the other or a third source.
The problem with going back earlier than proto indo-european is that PIE is a reconstructed language based on the earliest attestations we have from various Indo-European languages, like Ancient Greek, Sanskrit, Hittite, and Latin. This means that any PIE word you are using to build a relation to another linguistic group is a reconstruction which 1) we may not know how to pronounce correctly in the first place and 2) may not have been reconstructed correctly.