The facts and numbers are invented to teach a repeatable investigation method. Adapt the sequence to your operation and obtain competent legal, safety or engineering advice where required.
The operating problem
A 12,400-SKU parts warehouse reported 96.8% location accuracy, yet urgent orders still failed at pick. The percentage hid concentration: 41% of the value difference came from only 63 fast-moving items, and most errors appeared after replenishment. All numbers in this case are invented for teaching; the investigation method is reusable.
Evidence before opinion
The team froze a four-week extract containing item, location, system quantity, counted quantity, transaction time, user, movement type and order reference. It checked unit-of-measure conversions and removed recounts that lacked independent verification. Differences were grouped by activity and time, not merely by employee.
What the pattern showed
Replenishment confirmations were posted at pallet quantity although inner cartons were sometimes moved. A second pattern occurred when pickers used an adjacent overflow location but the system retained the primary bin. The visible symptom was a count variance; the controllable causes were unit configuration and unrecorded location movement.
Control change
The team corrected pack data for the affected items, required scan confirmation at source and destination for replenishment, labelled overflow locations, and cycle-counted the 63 high-impact items twice during the first month. It did not recount the entire warehouse, because that would have consumed time without addressing the transaction causes.
Verification
For each affected item, the owner tracked transactions since last good count, absolute unit variance and financial variance. Closure required four weeks without the same error mode plus independent spot checks. The lesson is to measure accuracy by risk and error cause, not only one blended percentage.
Reusable review questions
- What primary record proves the starting condition?
- Can another reviewer reproduce the calculation or event timeline?
- What control changed, who owns it and where is it documented?
- What evidence will demonstrate effectiveness after implementation?
- Which condition would require escalation or reopening?
Prepared and reviewed by Amit Mehta · Published 2026-10-16 · Editorial and corrections policy