Unit variation and channel matching: the QC lottery

Mass-produced drivers vary. Two units of the same IEM commonly measure 1-3 dB apart in places, and budget models or particular production batches can vary far more. Pads and tips age, driver tolerances drift, and revisions ship silently under the same product name. A published measurement is one sample from that distribution, usually a healthy one, since reviewers often receive hand-picked units.

Channel matching is the version you can actually hear most easily: if the left and right earpieces differ by a couple of dB in a narrow band, imaging smears and center vocals pull sideways. Good manufacturers match channels within about 1 dB through the mids; the community's recurring QC complaints (channel imbalance being the classic) are why our graphs can plot left and right channels separately instead of only an average.

Practical defenses: buy from retailers with real return policies, listen for channel balance with a mono track as soon as a unit arrives, and treat small graph differences between two products (especially in the treble) as within the noise of unit variation. If a review unit measures unusually well relative to retail reports, that is a signal worth weighting.

This is also why qfiAudio aggregates community QC reports per product and shows them in the reliability card, rather than assuming the measured unit represents every unit sold.