Also (going only on common sense and limited experience – I’m not an expert at this stuff), 2% is absolutely naff-all in terms of measurement precision for something like this. I am very sure that lots of factors like recent sleep, recent exercise, recent eating, time of day, mood, caffeine intake, recent challenges to your immune system would all combine to induce much more than 2% test-retest variability on a measure like this. Some of these can be controlled, some not so much, of course.
In practice we’re not going to do 5 FTP tests over a couple of weeks and take an average, so I think the whole “listen to your body and tweak over subsequent sessions” is the best answer. The only difficulty is knowing how hard it’s supposed to feel. I’ve done a few lately where I’ve thought “Well, the instructions seem to imply that I ought to be struggling to finish this. I’m definitely working, but could definitely be working harder.”, but I wasn’t really sure whether the instructions were taking that tone just to make weaklings feel better! 🙂