Judge QC photos with a score, not a feeling

Warehouse photos are cheap to request and expensive to ignore. They are also easy to skim: the item looks broadly right, so it gets approved, and the detail that matters is discovered at home.

The discipline is to score the same dimensions every time, note what could not be seen, and write a verdict that names the weakest points. That record is what makes a later dispute resolvable.

Verdicts here separate three outcomes rather than two. Approval is warranted when every scored dimension meets the chart and no dimension has been left unassessed. A reshoot is warranted when a dimension that decides fit cannot be judged from the images supplied. Rejection is warranted when a scored dimension fails against the chart, or when the photographed item differs from the listing image in kind rather than in lighting. Naming the outcome before writing the verdict prevents the impression-led approval that is hardest to defend later. An approval given without a stated reason cannot afterwards be examined, defended or improved, which is why the outcome is named first of all.

The economic weight of a verdict tracks the price of the item. Across the catalogued set, 44 rows fall below $20 while 15 sit at or above $80, against a median of $36.18. On a low-band item a domestic return can consume more than the goods are worth, so the practical remedy is a recorded note rather than a case. On an upper-band item the same defect justifies the handling time, and a dated verdict converts an impression into evidence. Scoring effort should be roughly proportional to that asymmetry. Where an item sits within that spread should decide how much effort a verdict on it can reasonably justify in practice.

Approving on an overall impression is expensive for a specific reason: it destroys the record needed to prove a defect. A claim supported only by a memory of how the photographs looked is a negotiation, while a claim supported by a dated scorecard naming the failing dimension and the chart value is a comparison. Requesting a second set of the same angles is the milder form of the same error, since it spends a handling day without producing the measurement that was missing. The measurement, not the number of images, is what settles the question. The same discipline applies to a rejection, which also needs its own named dimension.

Recorded verdicts are the input that step 7 cannot reconstruct. Whether to rehearse a pack, discard a shoebox or keep corner guards depends on what the photographs showed about the structure and condition of the item; a defect accepted knowingly changes which protection is worth paying for. The verdict, the photographs and their date also travel forward to step 9, where delivered contents are compared against what was approved, and where a discrepancy without a baseline cannot be resolved. An approval that exists only as a click leaves nothing to compare against. What survives the order is the written document rather than the decision that the same document recorded.