Complex analysis for two columns
Complex ABC computes classes A, B and C for two metrics at the same time and gives an item a two-letter code: the first letter is its class in the first column, the second in the second. That makes nine cells, AA through CC. The point is the pairs that contradict each other: AC and CA call for opposite decisions even though a single metric makes them look alike.
An XLSX file with headers in the first row. You need two numeric columns — revenue and stock on hand, or sales and margin — plus a column with the item name. The 80 and 95 thresholds are set separately for each of the two columns: the metrics behave differently, and one shared boundary distorts the picture.
Important: merge duplicate rows before uploading. The service does not detect or combine duplicates — an item that appears in the file twice is counted twice, and the shares and classes come out wrong.
The table gains a share and a class for each of the two metrics plus the combined cell code — AA, AC, CA and so on. The matrix shows how many items and how much money sit in each of the nine cells, and the recommendations work through the difficult ones: frozen money, traffic drivers, the tail. The report downloads as XLSX.
From classic ABC — by the second metric: an item gets two letters instead of one. From XYZ+ABC — by taking both metrics at a single date rather than as a series of periods: complex ABC answers “how much”, XYZ+ABC also answers “how evenly over time”.
share = xᵢ / Σx × 100% (for each of the two columns separately)
Shares and classes are computed for each metric independently — the same way classic ABC does it, with the same 80 and 95 thresholds, set separately for each column.
The point of the analysis is the pairs that contradict each other. AC is an item that earns revenue while barely sitting in stock, CA is stock that earns nothing; a single metric cannot tell them apart.
| Item | Revenue | Revenue class | Stock | Stock class | Cell |
|---|---|---|---|---|---|
| SKU-101 | 412,000 | A | 60,000 | B | AB |
| SKU-102 | 265,000 | A | 310,000 | A | AA |
| SKU-103 | 158,000 | A | 120,000 | A | AA |
| SKU-104 | 96,000 | A | 6,000 | C | AC |
| SKU-105 | 74,000 | B | 210,000 | A | BA |
| SKU-106 | 51,000 | B | 18,000 | C | BC |
| SKU-107 | 33,000 | B | 95,000 | B | BB |
| SKU-108 | 21,000 | C | 9,000 | C | CC |
| SKU-109 | 12,000 | C | 150,000 | A | CA |
| SKU-110 | 8,000 | C | 5,000 | C | CC |
Classes are computed for revenue and for stock value separately; the last column is the cell code.
Twice, by the ordinary ABC rules: first on the first column, then on the second. An item gets two letters — the first from the first metric, the second from the second — and together they form a cell code from AA to CC.
Thresholds are set for each column separately: revenue and stock behave differently, and one shared boundary distorts the picture.
The first letter is the class in the first metric, the second the class in the second. AA is large in both, CC is small in both, and neither needs an analysis to spot.
The work starts with the contradictory pairs. AC earns revenue while barely sitting in stock — easy to lose to a stockout; CA is frozen money that revenue never returns. In the table above those are SKU-104 and SKU-109.
Any pair of numbers you want to see against each other: revenue and stock value, sales and margin, shipments and what is left on hand. What matters is that both columns are numeric and belong to the same item.
The file does not leave your browser while this runs — the computation is local; the details are on how the data flows.
The theory behind the method — in the article: Complex ABC analysis: what two metrics show at once
