Suppose a purchase agreement requires each submersible pump to receive a defined wet test. The factory submits a passing spreadsheet with the shipment quantity, but the buyer cannot find the original result for a pump taken from a carton. Match the physical unit, the test event and the packing identity before deciding whether it passed. Equal row counts and shipment counts do not prove coverage. Retests, repeated scans, relabeling and mixed cartons can hide a unit that never received its required check.
Keep one unit identifiable at three points
Record the pump serial or a controlled temporary identity at the test bench. If the permanent nameplate is fitted later, retain a unique link from the temporary ID to the final one, with the reason for any relabeling. The wet-test record should identify the time, model configuration, procedure and decision. Flow, head or other readings should be recorded where the agreement calls for them. At packing, compare the nameplate inside the carton with the packing identity. A lot number may cover many units and cannot substitute for individual identification where the purchase calls for unit-level evidence.
Some stations overwrite a first result with a later pass, or publish only the final disposition. For service and release, retain both the original events and the current decision. If rework changes a part relevant to acceptance, apply the agreed retest rule and identify which run cleared the unit. Match market versions and packed accessories too. A test result for one build is not automatic evidence for every pump that shares the same housing.
Check the record in both directions
Start from the actual outbound pallet. Choose a few units without first choosing rows on the test sheet. Open the boxes, identify the physical serial, version and kit, and trace each back to its test. Then start from a few test records and locate the units in the shipment, rework area or sample hold. Both directions matter. One can reveal a boxed unit with no pass; the other can reveal a tested unit that was never part of this shipment. When identities do not match, hold the defined affected scope under the agreed procedure and distinguish missing records, mislabeled units and tests that did not happen.
For illustration, suppose eight cartons contain eight pumps and the sheet has eight passing rows. One serial appears twice after rework, while another never appears. Retain the two events as history for one pump and resolve the missing serial before release. The total did not change, but the number of unique covered units did. Counting completed rows and releasing identified equipment are different jobs.
Specify what the supplier must be able to produce
Agree whether the buyer receives original data or a signed traceability export, which fields are needed and how long the results remain accessible. Unrelated customer information is not needed to verify this order. Record actual inspections and authorization for any deviation affecting safety or performance. A video can illustrate a method, but cannot usually replace agreed unit-level evidence. When the same serial returns later, service staff should be able to see what was tested, whether it was reworked and how it was packed.
Include a rule for an offline station. A controlled temporary record can be reconciled to a serial while retaining the actual test time and the later entry time. If the link cannot be established, treat the evidence as missing. Do not backfill an identical pass for every unmatched unit.
Frequently asked questions
Does a common lot code prove every pump had a wet test?
No. It helps identify the population. Unit-level evidence still needs a link from each pump to its test result.
Can the result be reused after relabeling?
Check the controlled link between old and new IDs and whether the pump configuration changed before accepting the original evidence.
Why do we need records if the supplier sent a test video?
The video can show how a test is performed, but generally cannot link each packed serial to its final accepted result.
