Why ABC alone is not enough
ABC classification ranks items by their share of sales value or consumption volume: a small group of A items usually accounts for most of the value, while a long tail of C items accounts for very little. That tells you where the money is, but nothing about how predictable each item's demand is.
Two A items can need opposite policies. One may sell a steady quantity every month; the other may sell in unpredictable spikes. Treating them the same either leaves the volatile item short of stock or ties up excess capital in the stable one. XYZ classification adds the missing dimension by measuring demand volatility with the coefficient of variation:
A low CV means demand is steady and forecastable; a high CV means demand is lumpy and hard to predict. Crossing ABC with XYZ produces nine segments, each with its own stocking policy.
ABC thresholds: share of value
Items are sorted by sales value from highest to lowest, then assigned by cumulative share of the total:
| Class | Cumulative share of value | Meaning |
|---|---|---|
| A | Top 80% | The vital few — a small number of SKUs driving most of the value. |
| B | Next 15% | The middle band — important, but not critical. |
| C | Bottom 5% | The trivial many — a long tail of low-value items. |
These 80/15/5 cut-offs are the classic defaults, not fixed laws. The StockLogic tool lets you adjust them to match your portfolio.
XYZ thresholds: demand variability
Items are classified by the coefficient of variation of their monthly demand:
| Class | Coefficient of variation (CV) | Demand pattern |
|---|---|---|
| X | CV < 0.5 | Steady demand — reliable to forecast. |
| Y | CV 0.5–1.0 | Seasonal or fluctuating demand — forecastable with care. |
| Z | CV > 1.0 | Erratic, lumpy demand — forecasts are unreliable. |
Published XYZ cut-offs vary between textbooks and industries; 0.5 and 1.0 are common reference values. The StockLogic tool defaults to tighter limits (X ≤ 25%, Y ≤ 50%) and lets you set your own. Whichever limits you choose, apply them consistently across the portfolio.
The 9-box ABC-XYZ matrix
Each cell combines a value class with a variability class and suggests a stocking policy:
| Segment | X — steady | Y — fluctuating | Z — erratic |
|---|---|---|---|
| A — high value | AXHigh value, steady demand. Automate replenishment, run just-in-time with low safety stock, and review frequently. | AYHigh value, fluctuating demand. Keep a moderate buffer, watch seasonality, and review forecast settings often. | AZHigh value, erratic demand. Buffer carefully or move to make-to-order; a stockout here is expensive, but so is overstock. |
| B — medium value | BXMedium value, steady demand. Standard reorder points with modest safety stock work well. | BYMedium value, fluctuating demand. Combine reorder points with periodic review of the demand pattern. | BZMedium value, erratic demand. Hold limited stock, favour shorter lead times, and consider order-on-demand. |
| C — low value | CXLow value, steady demand. Bulk-buy infrequently to cut ordering cost; simple two-bin control is enough. | CYLow value, fluctuating demand. Keep small buffers and reorder in batches; do not over-invest in forecasting. | CZLow value, erratic demand. Drop the item or order strictly on demand to prevent dead stock. |
The pattern to remember: the further right and down the matrix you go, the less stock you should hold and the more you should rely on order-on-demand. The top-left AX corner rewards automation and lean just-in-time replenishment; the bottom-right CZ corner is where dead stock accumulates if you let it.
How the StockLogic tool builds the matrix
The ABC-XYZ tool classifies every SKU in your CSV by cumulative share of sales value and by the coefficient of variation of its monthly demand, then renders the full 9-box matrix with counts and value per segment. It calculates all items locally, up to 1,000 SKUs per upload.
You choose the ABC cut-offs, the XYZ CV limits and the tier names. Items without enough monthly history get a flat demand split, so read their XYZ class as unknown rather than steady. For the full calculation reference, see Methodology & Formulas.