Of course it's sketchy - the reconstruction is only as good as the evidence available to support it. As such, the more analogies the better.That sounds very sketchy though.
I assume the more analogies there are, the more can be reconstructed, & the vice versa.
The concept of language trees also helps. If we know that dead language A was closely related to living language B and we have infomation about how living language B has evolved, linguists can draw respectable conclusions about dead language A too.