Troubleshooting

Traceability and Lot Genealogy: Costly Mistakes and How to Catch Them

Troubleshooting guide for traceability and lot genealogy programs: the errors that inflate recall exposure and break genealogy, with the symptom, root cause, and numeric fix for each.

Symptom: your Traceability Coverage Score reads 98 percent but recalls still pull three times the units you expected. Root cause: coverage is counted at the case level while genealogy breaks at the unit level, so a single unscanned repack step orphans thousands of serials. Fix: measure coverage at the smallest sellable unit, not the pallet. If you serialize 40,000 units per shift but only verify 500 case scans, your real per-unit coverage may be 82 percent, not 98. Rerun Traceability Coverage with unit-level denominators and compare the two numbers. A gap over 8 points between case and unit coverage almost always means an untracked aggregation break.

Symptom: Scan Compliance Rate looks healthy at 96 percent but auditors find missing reads at receiving. Root cause: operators re-scan the same good label to clear a soft error, inflating the numerator with duplicate reads. A line running 120 scans per minute can hide 4 to 6 percent phantom duplicates. Fix: dedupe by unique serial before computing the rate in Scan Compliance Rate, and reconcile scan count against expected unit count from the work order. If you produced 10,000 units and logged 10,700 scans, that 7 percent surplus is not throughput, it is rework masking a failing verifier.

Symptom: Lot Genealogy Completeness passes at 100 percent per record, yet a trace-back for one raw lot takes six hours instead of minutes. Root cause: completeness checks that each record has a parent field populated but never validates that the parent actually exists in the prior lot table. Orphan foreign keys score as complete. Fix: validate referential integrity, not field presence. Run Lot Genealogy Completeness with a join against the parent lot master, and count any unmatched parent ID as incomplete. In a typical bill of materials with 12 components, even a 2 percent orphan rate across 5,000 lots leaves 100 broken links that surface only during a recall.

Symptom: Recall Exposure Radius returns a tidy number, but legal wants the worst case and yours is optimistic. Root cause: the radius is computed from the average lot size instead of the maximum commingled lot. If one raw lot feeds 8 finished lots through a shared blend tank, exposure fans out multiplicatively, not additively. Fix: model exposure from the largest single ingredient lot and its full downstream fan-out. A 2,000 kg flour lot split across 40,000 packages does not expose 2,000 units, it exposes every finished lot that touched the tank, often 6 to 10 times the naive estimate. Always run Recall Exposure Radius on the max, then report the average separately.

Symptom: Serialization Workload estimates 2 seconds per unit but the line runs 40 percent slower than planned. Root cause: the estimate counts print and apply time but omits verify, reject, and reconciliation cycles. A print-and-apply head rated at 60 units per minute drops to 36 once you add a vision verify at 300 ms and a 3 percent reject-and-reprint loop. Fix: build Serialization Workload from measured cycle time including verify and reject handling, not nameplate speed. If you sized labor from 60 units per minute and reality is 36, a three-line plant is short roughly 1.5 full-time operators per shift, which is where overtime quietly appears.

Symptom: Barcode Label Workload underruns and you keep running out of media mid-shift. Root cause: the label count ignores reprints, alignment waste, and end-of-roll scrap. A 2,000 label roll rarely yields 2,000 usable labels; expect 3 to 5 percent loss to threading and misfeeds. Fix: gross up Barcode Label Workload by measured scrap and reprint rate. If you need 38,000 good labels and your net yield is 95 percent, order for 40,000. On RFID lines the same logic applies to encode failures; a 1.5 percent tag kill rate means Barcode Label Workload and RFID Tag Cost both need the failure factor baked in, or your consumable budget runs 4 to 6 percent short every month.

Symptom: Traceability Gap Score improves on paper after a fix, but real trace-back time in Genealogy Lookup Time does not move. Root cause: teams close gaps that are easy to close, not the ones on the critical path. Reducing 50 low-frequency gaps feels productive while one high-volume junction still forces a manual reconciliation. Fix: weight Traceability Gap Score by transaction volume, then attack the gap with the highest volume times lookup time product first. A single junction handling 30 percent of units at 45 minutes of manual lookup outweighs 200 gaps that touch 0.1 percent each. Re-measure Genealogy Lookup Time after each fix to confirm the number actually fell.

Symptom: Lot Record Completeness reports 99 percent, but the missing 1 percent is always the same critical fields. Root cause: completeness is scored as an average across all fields, so 40 well-filled fields hide two empty ones that matter, expiry date and supplier lot. Fix: score critical fields separately with a hard pass or fail, not a blended percentage. In Lot Record Completeness, tag expiry, supplier lot, and quantity as mandatory, and let any single miss drop the record to incomplete. A blended 99 percent can conceal a 6 percent miss rate on the three fields a regulator will actually ask for during an audit.

Published 2026-07-01.