The number of boutique consumer speaker manufacturers using measurement as part of a strict incoming parts QC policy to secure zero variance in all parts is extremely low- perhaps non-existent.
If you do measure, as the engineering driven companies tend to do, OEM parts typically vary quite a bit unit to unit. SO you may do batch testing (test a few from each batch that arrives). If you are super strict, you are throwing a lot of parts away. Achieving perfect parts that perform exactly the same all the time and are perfectly consistent over hundreds or thousands is not a realistic target.
A 1/2 dB variance in sensitivity applied to the entire midrange or the entire tweeter response is very audible to a listener, even if that listener cannot identify a 1/2 boost or cut somewhere along the response curve. 1/2 across the entire band is different which is why parts sensitivity is such a big deal. Combine that with the idea that a passive crossover cannot be precision adjusted to account for this part variance. Then one more problem, I don’t think ANYONE is making parts plus or minus 1/2 dB in sensitivity specs over the life of the part. The only way to deliver that is two choices: through anything outside a super tight QC window away or develop precision manufacturing so good you get perfect matching part to part. I don't think that level of precision in part manufacturing exists. So variance from part to part and then complete speaker to speaker is a part of the business. That’s why they have a overall plus or minus spec so they can absorb theses variances and still meet an overall claimed spec.
How many in the field, dealers or end users, ever A/B the same speaker? I would say that almost never happens. It might surprise you if you did.
Brad

