Even with good simulation and engineering, balancing is still a key and tricky aspect. I've said before that I have a lot of respect for Toole and Olive. This is because their work mostly correlates exactly with KEFs historical findings and experience on what makes a good sounding loudspeaker. KEF were involved with a project headed by Søren Bech in the late 1980s that has similar conclusions and instigated the Uni-Q driver. Olive's preference score tells you that the responses in all directions need to be smooth but there's considerable "wiggle room" when you come down to the fine details. With a given set of drivers you can come up with a number of crossovers that score similarly (hopefully high if everything is working well!) but sound subtly different. And when you get down to fine margins, preference is quite a moving target as it depends on music selection, room, speaker position and listener taste. Brightness in particular seems to be an extremely personal judgement and varies a lot from person to person. Smoothness in the power response seems to be absolutely key, more so than other curves in the spin plots. Driver distortion and choice of crossover components and their associated distortion can quickly affect a perceived balance, and this can be hard to get a handle on because it won't show up in the frequency response measurements. Of course we're talking about small margins here, but they become very important for high-end products.
I think that the major thing is to try and keep a sense of perspective. "Neutral" for us also means that a speaker can be used by a wide range of people, in a wide range of rooms and on a wide range of music. We simply can't test a single product in enough scenarios to know if we're close to neutral. Which means that during balancing we rely a lot on comparisons with existing product, using a rooms that we're very familiar with, and feedback on existing products that have been in the market for some time.
With Reference META I wouldn't say there's a very big difference in the balancing target compared to Reference. If you look at the spins you can see that overall things are very similar. David Bosch, who did the crossover design work, spent a lot of time working out an approach for the crossover that complemented the natural characteristics as much as possible. He did a great job and was able to tidy up some areas significantly. This really comes across in the listening experience.
The major change in Blade META and Blade 2 META is the LF/MF crossover. The new Uni-Q allows a gentler crossover with the LF which improves the response at wide horizontal angles. This changes the characteristic in the upper bass a fair bit, making everything sound more immediate.