Speaker design is a number of compromises.
Floyd and Toole Toole and Olive are obviously generally right that on-axis performance important, but there are a number of reasons a less than perfect on-axis performance is the right trade-off.
As mentioned by me and others, speakers may be specifically designed to NOT have toe-in. That means the actual listening axis is not on-axis, but typically 10-15 degrees off-axis. Then
that is what should have a flat even response not on-axis. You may also have artifacts on-axis that evens out significantly (into an actual flat response) off-axis, that will end up being uneven off-axis if you correct the on-axis too much. So the combined response of 0-15-30 (at least) should be considered, and it's not necessarily a good thing if one axis is made perfect at the expense of others.
While the on-axis response is a strong indicator of a good speaker, you can't look at on-axis response in isolation and assume you know everything about the speaker.
Speaker design also can't be done by measurements alone. Overcorrected drivers doesn't necessarily sound right, so sometimes leaving good enough alone actually sounds better than paper perfect response.
With regards to "sounding bright", most speakers that sound bright has significant rise from 2-4khz. This speaker has hardly any rise before 8khz. The sound of moderately added level at 8khz and beyond sounds very different that the typical brightness or harshness, and is rarely offensive given a good driver with low distortion.
The measurement focus of ASR is great, but when suddenly the entire forum is a speaker design expert and can determine the quality of any speaker just by reading graphs they hardly understand, we're headed in the wrong direction (in my humble opinion).