You are correct. I just wanted to point out that if you are looking for accuracy (not something that just sounds better), you should be aware that "accuracy" changes over time. What was "accurate" one year ago, is not "accurate" today and most likely will not be "accurate" one year after. If there is such a big deviation between the same type of headphones, maybe they should not offer averaged calibration profiles for them at all?
It would be interesting to send them a set of headphones for calibration, then wait for some time and send them again. I would be curious to see how accurately both of those calibration profiles match.
you're looking at the wrong cause here.
-first of all, there is no absolute of accurate headphone signature. part of the signature needs to compensate for your own head and body(parts that are usually interacting with the sound but not when wearing a headphone), so while there absolutely exists a flat frequency response for you, there is no evidence that it will sound neutral to me. now while this may seem desperate, we're still all humans, we tend to have 2 ears more or less at the same place and very few people have a pyramid instead of a head. the differences will amount to a few dB in some places, but in general the big lines will fit most people. I'm just saying you need to be realistic in your expectations of what a neutral curve really offers.
-secondly, if you took 5 pairs of that headphone and measured them, you would get several dBs of variations somewhere. so of course as they get more pairs to measure, they average the "average response". even left and right drivers will almost always have some small variations of a few dBs unless the manufacturer makes some efforts to limit that specifically(and some do).
-third, pads with time will inevitably alter the frequency response, this is the one sure and most significant(as in audible) aspect of headphones changing over time. I don't know if they make their average curve based only on new pairs, or if they include the pairs sent by customers, but whatever they do, changes do occur over time (until you replace the pads ^_^). but that's only part of the fun stuff changing a signature a little. how you place the headphone on your head for example. if you have a lot of hair, glasses, a big head creating a stronger clamp, and of course what signature comes out of your very own pair. while all those can amount to very little sometimes, they are variables Sonarwork can't control when they give you an average curve. and at best they can measure your headphone, match the drivers, and remove part of the uncertainty. which is pretty significant given how wrong most headphones can be when it comes to FR.
the way I see it, when you use the average curve, you trust the brand. if you get an old Audeze headphone(apparently they have improved stability on the new models), from one pair to the next you never knew what would happen. but if you get a hd800, from one pair to the next there will be very little variation because Sennheiser is being more German than a German engineering caricature on those pairs. so using an average for that headphone will be a good deal more reliable when it comes to achieving Sonarwork ideal target response. else there is custom measurement.