Ahh yes you are right it is in dBm as reported by the AP management software.
These numbers are all from my desk directly under the ceiling mounted AP.
I think I’ve figured out what is going on here - indeed 20dB of gain would be insane, assuming your antenna is well matched to the transmitter. But these boards are broken in some way, and I expect there’s a horrible impedance mismatch between the transmitter and the antenna system, so rather than the energy from the transmitter being emitted as radio waves, it’s being reflected back to the transmitter and ultimately being lost as heat inside the board.
By adding the wire loop, we haven’t improved the antenna, we’ve probably just changed its impedance to be a bit closer to the rest of the system, so more of the energy is being converted to radio waves and a little less is being wasted as heat.
These broken boards have quite a different layout to the working ones with the red antenna - I wonder if the antenna matching system is actually set up for a PCB trace instead of a chip antenna.
This also might explain why this board gets a lot hotter than my other esp32-c3 boards - I’d put it down to having less board area, but if all the transmitter energy is being soaked into the board as well I bet that doesn’t help!