Full protein sequence or only conserved domains

Question

[Vasiliy Krestov]

I have a big group of proteins and I want to infer phylogenetic relationships between them. All proteins of this family have two highly conserved domains. Some of them have additional domains which are widespread across the family but these domains tend to be diverse and many proteins don't have them at all.

Is it ok to take as a rule to use only highly conserved domains when building phylogenetic trees for distantly related proteins (because unconserved domains can accumulate mutations very fast and because of the amount of "noise") and full sequences for closely related proteins?

Thanks in advance!