Blog · Data quality and reconciliation · Supply chain
ABC inventory cycle counting uses approved classes to organize count priorities and assess the results. Compare the items due with completed counts, then separate exact matches, net adjustments, and gross dollar discrepancies.
ABC counting uses approved inventory classes to prioritize physical counts and review stock-record discrepancies. Measure both whether the required items were counted and how far their final counts differed from book quantities. A completed count with an error and an overdue count are different control problems.
This is a review of count coverage and results, rather than a new ABC classification exercise. Use the class already approved for the reporting period. The ABC classification example explains cumulative-value cutoffs; ABC and XYZ analysis adds a separate demand-variability dimension when that distinction matters.
The due list establishes the denominator. Include the count task ID, SKU, site, bin or other counting unit, approved class, due date, and the policy version that made the task due. Join the completed-count export to that list, retaining tasks without a result. Starting from completed counts alone makes missing work disappear.
Do not assume a universal frequency for each letter. The GAO's inventory-count guide describes frequency as a management decision informed by factors including dollar value, operational criticality, turnover, and theft risk. Its control guidance also emphasizes responsibilities, segregation of duties, and investigation of differences.
Capture book quantity at the count cutoff, the physical count, unit of measure, unit cost used for the review, completion status, and approval reference. Use the approved final count for the adjustment view, and preserve earlier counts separately to study the counting process.
A movement between the book snapshot and the physical count needs a documented bridge. Otherwise a shipment can appear as a counting shortage. A closed investigation or approved recount should not overwrite the evidence that explains how the discrepancy was resolved.
For a first-pass comparison, use exact quantity matches. If management permits a tolerance, publish that rule and retain exact-match results alongside it. An undocumented tolerance makes apparent improvement impossible to assess.
A US industrial distributor has the following due items for its October review. Each row is one SKU at one counting location, all quantities use the item's stocking unit, and all costs are in USD. A3 is still incomplete; its blank count is not zero.
| SKU | Approved class | Book quantity | Final count | Unit cost (USD) | Complete |
|---|---|---|---|---|---|
| A1 | A | 100 | 98 | 50 | Yes |
| A2 | A | 80 | 81 | 100 | Yes |
| A3 | A | 60 | 25 | No | |
| B1 | B | 200 | 196 | 10 | Yes |
| B2 | B | 120 | 120 | 20 | Yes |
| C1 | C | 300 | 290 | 2 | Yes |
Put these columns in A:F and the data in rows 2:7. Calculate quantity difference in G2, signed dollar difference in H2, absolute dollar difference in I2, and exact-match indicator in J2. Fill down:
=IF(F2="Yes",D2-C2,"")
=IF(F2="Yes",G2*E2,"")
=IF(F2="Yes",ABS(H2),"")
=IF(F2="Yes",IF(G2=0,1,0),"")
The indicators apply only to completed counts. Microsoft's COUNTIFS documentation confirms that a cell is counted only when all specified criteria are met. Use that behavior to separate completion status from class. Covirage is not affiliated with Microsoft.
List A, B, and C in L2:L4. The due-item count in M2 and completed-item count in N2 are:
=COUNTIF($B$2:$B$7,L2)
=COUNTIFS($B$2:$B$7,L2,$F$2:$F$7,"Yes")
Coverage in O2 is:
=IF(M2=0,"",N2/M2)
The class roll-up follows directly from the scheduled-item table. Percentages are rounded to one decimal place; dollar calculations use unrounded inputs.
| Class | Due items | Completed | Coverage | Exact matches | Net difference (USD) | Absolute difference (USD) |
|---|---|---|---|---|---|---|
| A | 3 | 2 | 66.7% | 0 | 0 | 200 |
| B | 2 | 2 | 100.0% | 1 | -40 | 40 |
| C | 1 | 1 | 100.0% | 0 | -20 | 20 |
| Total | 6 | 5 | 83.3% | 1 | -60 | 260 |
A1 is short by 2 units at $50, producing -$100. A2 is over by 1 unit at $100, producing $100. The A-class net is zero, but its absolute discrepancy is $200. Netting those differences would conceal the two stock-record errors.
B1 contributes -4 units at $10, or -$40; B2 matches exactly. C1 contributes -10 units at $2, or -$20. The signed total is -100 + 100 - 40 + 0 - 20 = -$60. The absolute total is 100 + 100 + 40 + 0 + 20 = $260.
Overall coverage is 5 / 6 = 83.3%. Exact-match accuracy among completed counts is 1 / 5 = 20.0%. Neither rate is the average of the displayed class percentages: compute the company rates from their underlying counts.
Tie completed and incomplete tasks back to the original due list. Then tie approved stock adjustments to the signed count differences, allowing for documented recounts, movements at cutoff, and adjustment approvals. Keep unresolved differences visible until the bridge is complete.
The immediate actions differ by class. A needs completion of A3 and investigation of the offsetting errors. B needs a review of B1's shortage, even though its coverage is complete. C needs its discrepancy investigated under the approved escalation rule.
These results do not measure inventory turnover, which uses stock and cost of sales, or inventory aging, which addresses time in stock. A count can confirm that an item exists while leaving its demand or valuation risk unresolved.
Bring your due list, completed-count export, and stock values to Covirage; it maps the columns, and its tools compute coverage and discrepancies by the supplied ABC classes and check the totals. The external AI model explains the results with every figure traced to its rows; the warehouse team performs the physical counts. Create a free Covirage account and bring a sample of your own count records.
See Covirage for supply chain teams for the supporting business context.
Management should approve a cadence based on value, operational criticality, error history, and available counting capacity. A items often warrant closer attention, but a low-value critical spare may also need frequent counts. The class alone does not establish a universal counting schedule.
Coverage asks whether the required items received a completed count. Accuracy asks how those completed counts compared with the recorded quantities. A team can finish every scheduled count while finding many discrepancies, so report both measures rather than treating completion as evidence of accuracy.
A recount of the same due item does not add another covered item. Preserve the first count and investigation history, then select the approved final result for adjustment analysis. Use a unique due-item identifier so recounts cannot inflate the completion numerator.