Very often fossils are not in the order evolution would predict. Rather it takes non plausible scenarios to justify many inconsistencies. You know many decendants have been found with their ancestors. Much of the evidence is based on a few chards of bone or a single bone that have been reconstructed into a complete organism with a whole life story and this is used as evidence.
Once one goes back past the family rank what you have is a variety of kinds, idealised from chards and single bones mostly, all thrown into one basket by some similarity. With homology found in non related species this is erraneous reasoning at its classification base.
Let's take tiktaalik for example, the great and irrefutable evidence of transition to land. Tetarapod footprints have been found that predate tiktaalik by 18my.
Discovery pushes back date of first four-legged animal : Nature News
To me these look like paw prints, similar to a bear. Paws belong to mammals.
Here is a link to a natural bear footprint. One can easily see the similarity to the 380my footprints.
BEAR FOOT PRINT
The fossil evidence is what is found. The interpretation of the evidence is subjective, not objective.
A 380 million year old mammal of course would totally falsify evolution as proposed. Therefore evolutionary researchers would never even entertain such a proposal.
Here is a link that discusses subjective taxonomic classifications.
The Fossil Record