ABC-XYZ inventory classification
Which items carry most of the value, and which have predictable demand? Paste monthly consumption, split items into A-B-C classes by value and X-Y-Z classes by variability, and see the 3×3 matrix.
Data is processed in your browser; nothing is sent anywhere.
The delimiter is detected automatically (tab, ; , |). You can copy and paste from Excel. The first column is the item name; every following column is one period's consumption (month, week). Decimal comma and point are both accepted. An empty, unreadable or negative cell counts as 0 and is reported.
Settings
In percent. Default 80% / 95%: items are ranked by value; those making up the first 80% of value are A, the next 15% B and the rest C.
CV = standard deviation / mean. Default 0.5 and 1.0; this varies by organisation (0.25 / 0.5 is also common). Anything above is Z.
01
How to use it
A
Paste item names and period consumption or upload a CSV; if you have unit values, add them as the second column and tick the box.
B
Adjust the ABC and XYZ thresholds to your organisation; the defaults are a common starting point.
C
Read the matrix, the item table and the policy suggestions; download the table as CSV to load into your inventory system.
02
ABC: classification by value
ABC rests on the Pareto principle: a few items carry most of the value. An item's value is its total consumption (total consumption × unit value if you entered a unit value). Items are ranked from largest to smallest value and their shares are accumulated. If the cumulative share before an item is below the A threshold (default 80%) the item is A, below the B threshold (95%) it is B, otherwise C. The item that crosses the threshold stays in that class, so there is always at least one A item.
The thresholds are a convention, not a standard. Values such as 70%/90% or 80%/95% are used; with few items or values close together the class boundaries look sharp. That is why the tool lets you adjust them.
If you rank by quantity instead of value, items with very different unit values get mixed up; that is why adding the unit value column is recommended for ABC.
03
XYZ: classification by demand variability
XYZ measures how regular consumption is. The measure is the coefficient of variation: CV = standard deviation / mean. The standard deviation is the sample one (n − 1). The smaller the CV, the more stable the demand; because CV gives relative fluctuation against average demand, items of different size can be compared. The default thresholds are CV ≤ 0.5 for X and CV ≤ 1.0 for Y; above that is Z.
For an item with no consumption the mean is zero so CV is undefined; the tool marks these items Z and reports how many there are. With intermittent demand (zero in most periods) CV comes out high and the item is Z; that is correct, but such demand should also be treated separately with Croston-type methods.
CV also counts seasonality and trend as fluctuation: a strongly seasonal item can come out Y or Z even though its seasonal pattern is known. For seasonal items it is more meaningful to extract the pattern first (as with the seasonality decomposition tool on this site) and look at the remaining fluctuation.
04
How to use the matrix
The matrix combines two axes: value (A, B, C) and predictability (X, Y, Z). AX items are high-value and stable; continuous replenishment and low safety stock make sense. AZ items are both valuable and unpredictable; they need the most management attention. For CZ items the value of holding stock should be questioned.
The policy suggestions are a starting point. The real decision also depends on lead time, the cost of a stockout, shelf life and contracts. For safety stock and reorder point you can use the matching calculator on this site.
FAQ
- Is the data I enter sent anywhere?
- No. All calculations run in code inside your browser; the data is neither transmitted to a server nor stored.
- How many periods of data are needed?
- Two periods are enough for the calculation, but the coefficient of variation is unreliable on a small sample. 12 months of monthly data or 26–52 weeks of weekly data is a good start. The tool warns if there are very few periods.
- How should I choose the thresholds?
- 80%/95% for ABC and 0.5/1.0 for XYZ are a common starting point. Adjust them to your number of items, your value distribution and your management capacity; class A should cover a number of items you can monitor closely.
- Can ABC be done without unit values?
- Yes, but it then ranks by quantity, which misleads for items with very different unit prices. Add the unit value as the second column and tick the box.
- Why can I choose between column and row layout?
- Some systems give the item in a row, others in a column. In the second layout the first column is the period label and every following column is an item.
- How should I treat new or end-of-life items?
- Items with a short history or fast-falling demand can have a misleading CV. Assess them separately or remove the affected periods from the data.
Let's set your stock policy by class
We derive demand behaviour by item and prepare forecasting and replenishment policies suited to the ABC-XYZ classes. Let's talk about your data in a free discovery call.