Slightly disappointed with their idea that even if the measurements look good, the sound can be bad. This is standard audiophile folklore.
If I were them, I would be thinking that if measurements and subjective listening impressions fail to correlate, I should be looking into the following:
- The listening tests are faulty - not enough time in the universe to properly test every permutation by listening to it, even if you believe that blind listening tests are scientific. And sighted listening tests really are going to lead you down some meaningless paths.
- The myriad *other* aspects of the system have not been set up optimally (woofer & mid levels, crossover frequencies & slopes, etc.). You can't assess a tweeter by listening to it alone.
- The measurements are incomplete or looking at the wrong things. Solution: improve their understanding of measurements.
Of course, I am probably being naive. If I were running a commercial speaker company, I would seek to appeal to all types of target customers, so I would have to talk about measurements *and* subjective listening - my company has artistic integrity, 'soul', etc.
I would also seek to imply that performance is not just about specs, because my speakers would not be better than everyone else's in that regard.
The making-your-own-drivers thing is a great angle.