Andrew! 
I’ve missed you!
Possibly.
I have been in a similar situation and I ended up spinning my wheels a bit and doing things that just weren’t going to bring me any closer to my goal of fixing that IA. I was doing things for the sake of doing them and it was silly. Eventually my practical and pragmatic self showed up and took charge! I cut all the noise out and focused on doing what would solve the problem and deliver the most value in the shortest space of time. Sorry I can’t be more specific than that.
Question: are you designing a completely brand new IA or are you redesigning an existing one?
Sounds like it might be the latter. If that’s right, in the very first instance I would run a Treejack study on the current state to find out exactly what’s wrong with it and while that’s running its course, I would do a content audit of some description - it would depend on how big a job it was and how much time I had. Next I would take what I learned in the tree test and the content audit and update the IA and then run a tree test on that in parallel to testing the screen flows.
Another approach might be to tackle the IA in sections - a series of continuous iterative improvements over time. That could work if it gets too big to wrangle. For me it’s all about creating a strong foundation with that first tree test to benchmark the current state of the IA and then you build it up from there always moving forward and continually getting shit done.
PS I still drink that beer that I can never remember the name of and I am NOT ‘in a gutter somewhere’ in Brisbane 