XYZ Analysis

The coefficient of variation is the standard deviation of demand per period divided by mean demand per period. Common cut‑offs are below 0.5 for X, 0.5 to 1.0 for Y and above 1.0 for Z, but the thresholds are a convention rather than a standard and are set to fit the assortment. The class then drives policy: X items suit automated ordering with thin safety stock, Y items need seasonal profiles and an event calendar, Z items need either a different forecasting method or a different way of sourcing them. XYZ is rarely used alone – combined with ABC analysis it forms a nine‑box matrix in which value and predictability together decide how much planner attention an item deserves.

• CV = σ / µ, both taken from demand per period (day, week, month)
• X – CV below 0.5: stable demand, safe to automate
• Y – CV 0.5 to 1.0: variation with a recognisable pattern (season, trend, promotion calendar)
• Z – CV above 1.0: irregular demand, handled by exception

XYZ analysis in practice

The classification is only as good as the history it runs on, and raw sales carry two artefacts that push items into Z for reasons unrelated to real demand variability. A promotional week inflates the standard deviation with a spike that was planned months ahead, not random. A stockout does the same from the other side: sales fall to zero because the shelf was empty, and the gap reads as volatility. The standard response to a Z classification – raise the buffer – then answers a data problem with stock. Clean the history of stockouts, one‑off extremes and promotional effects before computing the coefficient of variation, and part of the Z class moves back to X or Y.

The second trap is the length of the period. An item selling a few units a week has more zero days than non‑zero days, so its daily CV comes out high whatever the underlying pattern, while the same item measured in weekly or monthly buckets can land in Y. Slow movers of this kind belong in the intermittent demand toolkit rather than in a larger buffer. Classes are also worth recomputing on a rolling window: an item with a strong season sits in Z when measured across the whole year and in X when measured inside the season, so a once‑a‑year classification describes the calendar rather than the item.

Veritico STOCK cleans the sales history of extremes, stockouts and promotional effects before the forecast is computed, and predicts real demand instead of censored sales. Portfolio segmentation, ABC by Pareto and a target availability set per segment then decide where safety stock is worth holding. In the Dr.Max deployment across 490 pharmacies, automated forecasting and ordering lifted availability by 4 % and revenue by 5 % while saving pharmacists two hours a day. See demand forecasting and inventory optimization and replenishment and allocation management.

Related terms

• ABC Analysis
• Intermittent Demand
• Safety Stock
• Inventory Optimization
• Dead Stock