The measurement of success of an IT project depends on the criteria of
the 'iron triangle' i.e time, cost and quality. Is the iron triangle
an inappropriate measurement for defining a success in software
projects today or is their a case for change.

Time and cost are relatively fixed variables, but I think you could
make a case for redefining success in software implementations when
you define "quality."

If a CRM implementation functions perfectly but the sales people don't
use it, is it a success?

I think you also have to ask if the implementation met the client's
objectives.