Notes
Newest first. Dates are when I wrote the note, not when I recorded.
A stairwell with a 240 ms tail
Six floors, painted concrete, a metal handrail that rings if you touch it. Clap tests put the decay somewhere near 240 ms, which is short for the volume and almost certainly the fault of a fire door propped open on the third landing. Closing it added roughly a third again.
Useful bit: the handrail is a better indicator than the clap. If it rings after you stop, the room will smear anything with transients.
Cheap mics, honest placement
I spent a season convinced the capsules were the problem. They were not. Moving the stand forty centimetres away from a plastered wall did more for the recording than any capsule swap I have tried since.
The rule I now follow: fix the geometry until the room stops being audible, and only then argue about equipment.
Wind, and other things you cannot fix later
A short list, kept because I forget it every spring: wind on an unprotected capsule, handling noise through a rigid mount, clipping on a transient you did not expect, and a refrigerator compressor that starts halfway through a take.
Everything else is negotiable in post. These four are not.