I measure mine at the listener position in the room. This is the only way to measure large dipoles in a useful way - so the measurements reflect what you actually hear. Actually I used 6 different positions (3 points along the couch, each high and low) and found the differences from dead center position were insignificant. I measure with REW, change stuff (arrangement, room treatment, EQ, whatever), measure again, repeat until satisfied.
Subjectively, I found that the early changes, which corrected big problems, made big easily audible improvements. Once the biggest problems were solved (mostly with room arrangement & treatment with tube traps, bass traps, resonators, foam), the later differences, tuning the last few dB, the differences weren't obvious, it would require more careful controlled listening to detect.
I never encountered a situation where making a change that improved the measurements, made it sound worse. However, there were corner cases. For example: At one point I was trying to fix a 70 Hz null. Applying EQ took +9 dB which made the FR flatter, but it sounded terrible: bloated, muddy bass response. So you might say that it measured better (flatter FR) but sounded worse. But that isn't true. It didn't measure better overall. Sure, the EQ improved FR but it made CSD and distortion much worse. I killed that null using room treatments instead of EQ, which flattened the FR while also improving distortion & CSD at the same time. A single measurement like FR can be misleading, unless you get the full picture and understand them.