Guide · Classification
ABC XYZ analysis: how to read and use the 9-box matrix
Two quick classifications tell you where your inventory money sits and which items are hard to predict, so you can treat them differently.
Updated October 15, 2026
Most item lists follow the same pattern: a small share of items carries most of the value, and some items sell steadily while others are erratic. ABC sorts items by value; XYZ sorts them by demand stability. Crossing the two gives a 9-box matrix that's a practical starting point for setting different policies instead of one rule for everything.
ABC: where the money is
- For each item, compute annual usage value = annual usage in units × unit cost. With less than 12 months of history, scale the months you have to a year.
- Sort items from highest to lowest value and compute each item's cumulative share of the total, including the item itself.
- Classify: A up to 80% cumulative share, B up to 95%, C the rest. The top-ranked item is always A, even if it alone exceeds 80%. These thresholds are conventions; adjust them to your business.
Annual usage × unit cost measures how much value flows through each item in a year (consumption value). It isn't the same as the money tied up in stock today, which is on-hand quantity × unit cost. Some teams rank by revenue or margin instead; pick the basis that matches the decision you're making.
XYZ: how predictable the demand is
XYZ uses the coefficient of variation (CV) of monthly usage: the standard deviation (population formula, as in the tool) divided by the mean, classified before rounding. A common split is X for CV up to 0.5 (steady), Y up to 1.0 (variable) and Z above 1.0 (erratic). Items with no usage at all are kept separate, because their CV is undefined.
A six-item example
Six made-up items with six months of usage each. Annual value is the six-month total scaled to twelve months, times unit cost:
ABC/XYZ results for the example file · Swipe sideways to see all columns.
| Item | Unit cost | Annual value | Cumulative share | ABC | CV | XYZ |
|---|---|---|---|---|---|---|
| PUMP-100 | $250.00 | $30,000 | 38.1% | A | 0.13 | X |
| SEAL-KIT | $4.00 | $24,080 | 68.7% | A | 0.02 | X |
| FILTER-A | $15.00 | $16,800 | 90.0% | B | 1.01 | Z |
| BRKT-7 | $2.00 | $7,120 | 99.0% | C | 0.03 | X |
| GASKET-3 | $1.50 | $765 | 100.0% | C | 0.72 | Y |
| VALVE-9 | $90.00 | $0 | 100.0% | C | — | No usage |
Two items, a third of the list, carry 69% of the value. FILTER-A is a B item with erratic demand (it sold 0 in two of six months), and VALVE-9 hasn't moved at all, which is a question for whoever owns that stock rather than a planning parameter.
Reading the 9-box
Common starting policies by box (adapt to your business)
| Box | Typical profile | Starting policy |
|---|---|---|
| AX | High value, steady | Frequent, smaller replenishments; tight reorder points; review often |
| AY | High value, variable | Forecast-driven planning with regular review of safety stock |
| AZ | High value, erratic | Close attention; consider make or buy to order, or hold protection centrally |
| BX / BY | Middle value | Standard reorder point or buffer rules; periodic review |
| BZ | Middle value, erratic | Check whether demand is truly erratic or driven by a few large orders |
| CX | Low value, steady | Larger, less frequent orders; simple rules; low effort |
| CY / CZ | Low value, unpredictable | Candidates for reduced stocking, make-to-order or consolidation |
| No usage | No recent demand | Review for obsolescence before buying more |
The matrix also helps set service levels by class. High service levels are expensive at the top end (see the safety stock formulas guide), so many teams reserve them for items where a stockout hurts most, rather than applying one level to everything.
Pitfalls to watch
- Too little history. The tool accepts three months, but scaling a short window to a year assumes it's representative. Use enough history to see seasonal patterns, ideally more than one year when seasonality matters; a seasonal item can look erratic (Z) even when its pattern is predictable.
- Intermittent demand. Items with many zero months get high CVs; look at how often they sell, not only CV.
- New and discontinued items. Short or ending histories distort both classes; flag them separately.
- Mixing value bases. Use one basis (cost, revenue or margin) consistently across all items.
- Classes go stale. Reclassify on a schedule, such as quarterly.
The ABC/XYZ tool runs this classification on a CSV in your browser, with editable thresholds and a 9-box summary. The example above is its downloadable template.