What I do is, go to the problem case and find the
problem transistor(s) and see what their behavior
problem is and what their main parametric excursion
is. Straight debug, case by case.
Vdsat/Vds applies to only one thing, DC gain as
affected by Rout. But you don't know whether that
Rout is dominated by the gain device or its load
device, necessarily, and you don't know whether it
comes from VT being high, from bias current being
high, from short channel devices (deltaL driven to
its extreme) or even if it's really a gm rather than
Rout problem. But you can dope some things out,
like short channel will degrade Rout but raise gm,
etc.
This is probably the lesson that's trying to be taught.
But there's a million ways to die, not just one.