Redesigning a Cycle-Count Program Under a One-Person Labor Cap
A distributor stocks 1,200 SKUs. An annual-dollar Pareto shows 60 A items carrying 78% of throughput value, 240 B items at 17%, and 900 C items at 5%. Last year's full physical inventory produced a $310,000 shrink adjustment, concentrated almost entirely in A items; the A-item discrepancy rate from routine counts is 4.1% versus 0.4% for C items. One stock clerk does all counting and can devote 3 hours per day to it across 250 working days. A full count takes 20 minutes for an A item, 10 minutes for a B item, and 5 minutes for a C item, including reconciliation. After the shrink result, plant management wants every SKU counted weekly. The clerk is the only person trained on the handheld system, and there is no budget for another headcount this year.
1. Test management's weekly-everything proposal against available labor, then propose a feasible count-frequency policy with the arithmetic shown.
Read the worked answer
Available capacity
3 hours/day x 250 days = 750 hours = 45,000 count-minutes per year.
Management's proposal (every SKU weekly)
| Class | Counts/yr | Minutes each | Total minutes |
|---|---|---|---|
| A | 60 x 52 = 3,120 | 20 | 62,400 |
| B | 240 x 52 = 12,480 | 10 | 124,800 |
| C | 900 x 52 = 46,800 | 5 | 234,000 |
| Total | 421,200 |
421,200 minutes = 7,020 hours, or 9.4 times the available 750 hours. It would require roughly 9.4 full-time counters. Reject it.
A feasible policy: A monthly, B quarterly, C annually
| Class | Counts/yr | Minutes each | Total minutes |
|---|---|---|---|
| A | 60 x 12 = 720 | 20 | 14,400 |
| B | 240 x 4 = 960 | 10 | 9,600 |
| C | 900 x 1 = 900 | 5 | 4,500 |
| Total | 28,500 |
28,500 minutes = 475 hours, about 63% of capacity, leaving roughly 275 hours a year for discrepancy research, root-cause work, and audits. That slack is the point: counts only detect error; investigation and process change are what reduce it.
Do not push to the ceiling. Counting A items twice monthly adds 14,400 minutes (240 hours), taking utilization to about 95%. At that level any investigation backlog stalls the count schedule and the program degrades to box-ticking. Hold A at monthly until A-item record accuracy has been inside tolerance for two consecutive quarters, then revisit frequency as a reward for improved accuracy, not as a substitute for it.
2. The A-item discrepancy rate is 4.1% while C items run 0.4%. What does that pattern tell you, and what actions follow?
Read the worked answer
Interpretation. A tenfold gap that tracks dollar value rather than count frequency points to transactional root causes, not to counting frequency. High-value, high-velocity items pass through more receipts, put-aways, picks, issues, and unit-of-measure conversions per year, so each transaction is an opportunity for a record error. Typical causes: mis-picks among look-alike part numbers, put-away to an adjacent location, unrecorded scrap or engineering samples, unreconciled partial receipts, and accessible stock that more than one shift can touch.
Counting A items more often would only surface the same errors sooner. The cost of the error is already incurred at the transaction.
Actions, in order
- Take the last 50 A-item discrepancies and classify each by cause (receipt, put-away, pick, issue, scrap, count error). Two or three causes will normally explain most of the value.
- Fix the dominant causes with controls: segregated receiving and put-away verification, scan-verified picking, locked storage for the highest-value A items, count-on-receipt for A items.
- Publish inventory record accuracy (counts inside tolerance / counts performed) as the KPI, with 0% tolerance on A items and 2% on C items. The annual shrink number is an output of the process, not a measure of it.
- Re-baseline the physical inventory after the process fixes, so the corrective adjustment is not simply recreated over the following year.
- Use the 275 hours of released capacity for step 1, not for extra counting.