XYZ Analysis and ABC-XYZ Matrix for Inventory, with Example

What XYZ analysis is

XYZ analysis sorts inventory items by how predictable their demand is: X items have steady consumption, Y items vary moderately (trend or seasonality) and Z items are irregular and hard to forecast. The measure is the coefficient of variation (CV) of consumption.

On its own it says nothing about the money involved, so it is usually crossed with ABC analysis, which ranks items by value. The result is the ABC-XYZ matrix: nine classes, each with its own replenishment policy.

Visual map of the ABC-XYZ matrix

The map summarizes the CV calculation, the X, Y and Z bands, the 3×3 matrix and the 10-item example.

How to calculate CV and set X, Y and Z

Take the last 12 months of consumption (or weekly data for fast movers) and calculate the mean and the standard deviation. CV is the ratio of the two.

CV = standard deviation ÷ mean consumption

Common bands, which you should tune to your business:

• X: CV up to 0.5. Steady demand, reliable forecast.
• Y: CV above 0.5 up to 1.0. Moderate variation, with seasonality or trend.
• Z: CV above 1.0. Irregular demand, with months of zero usage and spikes.

Standard deviation is the concept covered in normal distribution and standard deviation. The calculations below use the population standard deviation of the 12 months.

Example: 10 items classified by ABC and XYZ

A fictional spare-parts store. Annual value is 12 months of consumption times unit price; the total is $520,330. The ABC class follows the cumulative percentage: up to 80% is A, up to 95% is B, above that is C.

Two cases show why XYZ matters. P02 is second in value, but its monthly usage is 300, 280, 0, 650, 310, 290, 0, 700, 320, 300, 0 and 600 units: mean 312.5, standard deviation 231.2 and CV 0.74. It is Y, so the class is AY. P04 has a mean of 71.7, a standard deviation of 112.1 and a CV of 1.56: it is Z, with usage concentrated in a few months, so the class is BZ.

The ABC-XYZ matrix

Crossing the two classifications, the 10 example items fall as follows:

Focusing on the cells with high value and low predictability (AY, AZ, BZ) pays off more than treating every item the same way.

Which policy for each cell

• AX: the best case. Frequent small-lot replenishment, kanban or scheduled deliveries, minimal safety stock.
• AY: forecast with moving average or exponential smoothing plus a seasonal adjustment, reviewed monthly. Safety stock is higher than for AX.
• AZ and BZ: avoid stocking to forecast. Make or buy to order when lead time allows, or combine a minimum stock with a lead-time agreement with the supplier.
• BX and CX: automatic replenishment with a reorder point and larger lots to cut ordering cost, as in inventory management.
• CY: simple periodic review with an economic order quantity.
• CZ: candidates to trim from the catalog, buy on demand or discontinue. Keep only what can stop production.

The safety stock calculation should use each item's real standard deviation, which makes XYZ a good filter for deciding where the effort is worth it.

Step by step to build your matrix

1. Pull 12 months of consumption per item (actual issues, not cancelled sales orders).
2. Calculate annual value, percentage and cumulative percentage, and classify as A, B and C.
3. Calculate mean, standard deviation and CV for each item and classify as X, Y and Z.
4. Build the 3×3 matrix and count the items and value in each cell.
5. Define the policy per cell and review the classification every quarter or half year.
6. Check outliers with cycle counting, counting class A items more often.

Common mistakes and limitations

• New items: with less than 12 months of history, the CV is unreliable. Classify them separately.
• Zero-demand months: they are real data and must enter the calculation. Dropping them makes a Z item look like X.
• One-off orders: a single large sale inflates the CV. Treat the outlier before classifying.
• Fixed bands: 0.5 and 1.0 are rules of thumb, not a standard. Adjust to your mix.
• Classifying only once: demand changes. Reclassify periodically.
• Ignoring criticality: a CZ item can stop the line. Value and predictability do not measure the impact of a stockout.

Frequently asked questions

What do X, Y and Z mean in inventory?

They are demand predictability classes: X steady, Y variable and Z irregular, set by the coefficient of variation of consumption.

What is the difference between ABC and XYZ?

ABC classifies by the value of annual consumption. XYZ classifies by how regular demand is. One tells you how much an item is worth, the other how predictable it is.

How do you calculate the coefficient of variation?

Divide the standard deviation of consumption by the mean. In the example, P02: 231.2 ÷ 312.5 = 0.74.

What CV limits define X, Y and Z?

The most common are up to 0.5 for X, up to 1.0 for Y and above 1.0 for Z. No standard fixes them; adjust to your business.

What is the best period for the analysis?

Twelve months cover seasonality. For fast movers, use weekly data from the last 6 to 12 months.

Should AZ items be stocked?

Generally only with a minimum stock and a lead-time agreement with the supplier. Stocking to forecast ties up cash in an item the forecast handles poorly.

About the author

Vagner Soares

Lean Manufacturing & Behavioral Management Specialist

Over 20 years in the automotive and metalworking industries (GM and Dana), Lean Manufacturing practitioner since 2006. SENAI instructor and mentor in Brazil’s Brasil Mais Produtivo program, delivering consulting, training and audits for 50+ companies, combining quality, productivity and people development.

LinkedInInstagramFull profile →