Any policy can be criticized on ethical or ideological grounds, but eventually it will need to be shown whether it is empirically creating worse outcomes than intended in order to argue against it. (And anecdotes are too easy to dismiss, so it needs to be statistical enough to measure.)

I'm especially thinking of indirect & unintended outcomes, like the number of women enrolling to join the armed forces (or else average length of service).

You can probably think of more. I think it's important to start finding data early, and beware that the data collection can change—especially when it's politically embarrassing! In Canada for example, we stopped reporting the sex on crime statistics, and started reporting gender identity instead... but guess what, it shows sudden jumps where you'd expect them, like "women" suddenly committing category 2 (violent) sexual crimes when previously there were approximately zero. That kind of stuff should speak for itself: hold on to it.