A Cheras sort that handles ordinary Tuesdays with room to spare will not offer you a complete tracking history in the second week of November. Bags sit, handhelds queue, and the exception cage becomes a second hub. Anyone who promises to read every consignment that week is promising a kind of work that does not exist.
Sampling is not a shortcut we invented for our own convenience. Dispatch already samples. A supervisor pulls the angry calls, the returns, and the “where is my parcel” chats, and those rows become the week’s truth. The risk is that the angry rows are not the broken path — they are only the loud one.
When we draw a sample for a peak-season check, we ask for three slices: a run of ordinary delivered stops, a run of failed attempts, and a run of anything that sat in exception longer than one shift. The ordinary delivered stops matter. They show whether inbound still happened, whether the notice still left, and whether proof of delivery still meant a photograph at the door or a tap at the van.
Size the sample to the question, not to a round number. If the question is “does inbound still get scanned after 10 p.m.,” you need night-sort rows, not more daytime proof-of-delivery. If the question is “do shop-lot notices still go out at lunch,” you need those streets.
Write the rule down before you open the extract. Otherwise the sample quietly becomes the rows that are easiest to explain, and the note will flatter the hub. A dated rule — every twentieth inbound bag between 8 p.m. and midnight, plus every exception older than six hours — is dull to read and honest to defend at the briefing.