ABC-XYZ Inventory Segmentation for Trims and Fabric

Give each material a stock rule that fits

The thousands of materials held in a trims store can have quite different replenishment needs, even when they support the same production programme. Main fabric can tie up a large sum of money, while a printed label may serve just one collection. Basic thread may cost little per cone yet stop a sewing line when it runs out. One rule, such as keeping 30 days of every item, misses these differences.

ABC-XYZ inventory segmentation provides an initial classification that makes these differences easier to examine. ABC ranks materials by annual usage value, while XYZ describes how much their demand varies in relation to the mean. Combining the classifications produces nine groups that help planners choose suitable inventory policies for further assessment.

The classes do not set a safe buffer by themselves because lead time, shortage cost, order commitments and the chance of reuse still matter. A cheap label with no approved substitute may deserve close attention, even when it falls in class C.

Use segmentation alongside the EOQ and periodic review guide. Segmentation helps decide which approach suits an item; the chosen approach then needs its own inputs and checks.

Define the stock item and the data first

Each inventory item must represent a pool of usable material that can satisfy the same approved demand. Two fabrics of the same weight may have different shades, finishes or buyer approvals, so their quantities cannot form one usable pool unless the order allows that substitution. The same problem applies to thread colour, label artwork and zipper length.

Choose a consistent reporting period, recording fabric demand in metres or kilograms on a stated basis and trim demand in the appropriate pieces, sets or cones. Convert purchase packs before adding their quantities to individual units, since a carton of labels cannot be treated as one label.

For ABC, decide which value the analysis will rank. This guide uses annual usage value because it answers a different planning question from the value sitting on the shelf today. A slow, costly fabric could have high stock value but low annual usage. Oracle's ABC analysis guidance describes several ranking criteria, including historical usage value.

For XYZ, measure demand in equal periods such as calendar months, retaining known zero-demand periods as numeric zeros in the series. A missing record is different from a month with no demand, so flag missing data before calculating the class.

Also inspect stockouts, since issues from the store can understate demand when no stock was available to issue. Where possible, use valid material requirements or unmet requests to explain the gap, rather than interpreting a shortage month as evidence that the material was no longer needed.

Calculate ABC from annual usage value

For a stable unit cost, the basic calculation is:

Annual usage value = Annual quantity used × Unit cost

When prices change, add the value of the individual issues on a consistent cost basis. To avoid distorting the ranking by mixing old costs with new quotations, state the currency, valuation method and time window that make the calculation repeatable.

Sort the items from highest to lowest annual value, then calculate each cumulative share:

Cumulative share (%) = Running total of annual usage value ÷ Total annual usage value × 100

Suppose a small, complete inventory contains the six items below, with illustrative annual usage values that total $100,000.

This example puts the first 80% of value in A, the next 15% in B and the final 5% in C. The cutoffs are management choices that do not have to produce a fixed share of the total SKU count. MIT's inventory lecture also presents ABC boundaries as choices for inventory control.

Since real items rarely land exactly on a cutoff, decide what happens when one item crosses a class boundary. For example, a policy may keep the crossing item in the higher class, then begin the next class with the following item. Apply that rule consistently and show the actual value shares produced.

Avoid assigning a class from the material description before examining its actual contribution to annual usage value. A rarely used accent fabric may fall in C and a large thread programme may rank in A because classification follows the chosen values rather than the material name.

Calculate XYZ from demand variation

The coefficient of variation, or CV, compares the spread of demand with its mean:

CV = Standard deviation of demand ÷ Mean demand

Because the standard deviation and mean share the same demand unit, their ratio is a dimensionless measure of relative variation. Although a CV of 0.30 means that the standard deviation equals 30% of the mean, it does not mean that demand will stay within that range.

The NIST definition of coefficient of variation explains this ratio and its sensitivity when the mean is near zero. If mean demand is zero, place the item in a separate review group because its undefined CV cannot be replaced with a meaningful zero or extreme number.

The following bands provide an initial classification scheme whose usefulness should be assessed against the store's planning needs:

A different store may need limits chosen to reflect its own demand patterns and planning decisions. SAP's XYZ analysis guidance describes segmentation by demand variation. The limits here are examples, not a universal apparel standard.

A small worked example

Consider three demand records covering six equal periods, using this short window to demonstrate the arithmetic rather than to prescribe a history length for production planning.

Although all three records have the same total demand and mean, their different timing explains why a rule based only on average demand may fail.

For the steady record, the sample standard deviation is found by dividing the total squared deviations of 1,000 by five, then taking the square root:

Sample standard deviation = √(1,000 ÷ 5) = 14.14

CV = 14.14 ÷ 100 = 0.1414

In Excel, a sample-based calculation can use STDEV.S(range)/AVERAGE(range). Microsoft's STDEV.S documentation specifies the sample method. Since the population standard deviation uses a different denominator and can change the class of a borderline item, select a method suited to the purpose and use it consistently.

Check source cells as well as the formula because ignored blank cells can make a series look steadier when zero-demand months were left blank. Correct known zero months, but investigate genuinely missing records before filling them.

Look beyond the CV number

A high CV does not prove that demand is random, since a planned seasonal launch can create large swings that are easy to explain. Plot demand through time before choosing a policy because a steady rise in orders can increase CV even when the trend is clear.

For an item with many zero periods, inspect both the size of each request and the gaps between requests. Buying plans should reflect the distinction between items with similar CVs that serve different purposes, such as a predictable annual order or a series of irregular repairs.

Forecasting: Principles and Practice discusses methods for intermittent demand, including approaches that separate request size from the time between requests. These methods need evaluation against the actual planning problem because a Z label alone cannot select a suitable forecast model.

The reporting interval affects the result because daily demand often looks more variable than monthly demand after aggregation smooths some of the changes. Compare CVs only when the period definitions are consistent, and make sure those periods remain useful for the lead times being managed.

Use at least a full relevant seasonal cycle where suitable records exist. Although twelve months can be a useful starting window for monthly demand, neither that window nor a longer history will always represent a changed buyer programme.

Turn the two classes into nine policy choices

Combine each item's ABC letter with its XYZ letter to create a starting point for policy review, with the resulting group supporting rather than replacing the replenishment decision.

Vendor-managed inventory may help an AX fabric programme, but the label alone does not make that arrangement work. Both parties need an agreed view of usable stock, demand and replenishment, together with compatible ownership terms, minimum quantities and buyer approvals.

Likewise, buying AZ materials only against firm orders may reduce leftover stock but expose the factory to a long lead time. When a late order release leaves insufficient time to obtain fabric, the practical solution may involve an earlier commitment, a shared base fabric or a qualified substitute. Each option needs approval before its potential saving can be included in the stock plan.

For C items, compare the effort of frequent replenishment with the shortage cost and the expense of holding additional usable stock. Low purchase value still leaves a risk of waste, since a larger buffer may be economical for reusable tape but leave boxes of outdated care labels.

Apply the method to three apparel materials

The following examples assume that ABC classes have already been calculated from a larger store inventory. Their A, B and C positions depend on that larger ranking and cannot be established from these three values alone.

When assessing weekly review and vendor-managed supply for the AX fabric, check approved shades and usable width before assuming that all recorded metres can cover the next order. The usable fabric width guide explains why nominal width can overstate what cutting can use.

For the BY label, a pre-build six weeks before a season may be worth testing if the launch is confirmed. The six-week example interval must cover the actual approval and supply process, with artwork that remains valid when the order is released. A seasonal pattern does not protect stock from a cancelled collection.

Although a simple reorder point may reduce routine planning work for the CX thread, the proposed 45-day buffer remains a trial choice until service risk and holding cost have been checked. Colour, ticket size and buyer approval still define what counts as a substitute.

Set the actual stock level separately

A safety-stock calculation requires information beyond the ABC-XYZ classification, including demand during the replenishment period, lead-time variation and the level of service being sought.

Suppose a reusable trim has mean demand of 20 units per working day and a replenishment lead time of 10 working days. Expected demand during lead time is:

20 × 10 = 200 units

If a separate service analysis supports a 60-unit buffer, a simple reorder point is:

Reorder point = 200 + 60 = 260 units

The 60 units are an assumed result of a separate service analysis, rather than a quantity derived from the item's ABC or XYZ class. A 45-day quantity would be 900 units at the same mean rate, which shows why a broad days-of-stock rule needs scrutiny.

Trigger the order from the appropriate inventory position, with usable stock, valid inbound supply and committed demand treated consistently. Fabric awaiting a quality-release decision cannot provide the same immediate service protection as an approved roll. Refer to the roll acceptance guide when defining which fabric quantities are ready for use.

Handle exceptions before they become shortages

Apply a production-critical override when the absence of a material can stop an order and no timely substitute exists, even if its annual usage value is low. Keep that flag alongside its ABC-XYZ class, with a short reason that planners can check.

New items need a separate route because the limited demand history of a first collection can give a precise CV without a reliable basis. Use the confirmed programme, technical approvals and supplier commitments until enough relevant observations exist.

Review returns and cancellations before treating negative quantities as demand, since a late return may concern an earlier issue while a cancellation removes a future requirement. Simply replacing negative values with their absolute values can invent demand that never occurred.

Finally, distinguish a price change from a demand change. If an item moves from B to A because its price rose while usage stayed steady, its greater cash exposure does not by itself justify a larger physical buffer.

Check supplier conditions alongside demand

Two X items with similarly steady demand may still need different protection when one supplier delivers within a few days and the other uses an uncertain import route. Record actual replenishment time from the point used in the stock rule, including any approval or release delay that belongs in that interval.

Minimum order quantities can also change the result when a large minimum for a low-use printed trim exceeds the remaining needs of its collection. Before accepting that quantity, check whether the supplier can split deliveries or hold approved stock under clear terms. Splitting deliveries reduces the quantity on site, but it does not automatically remove a financial commitment.

Check package size and shelf life before choosing a large C-item buffer. A full carton may support a convenient replenishment rule for a reusable item, yet offer poor value when part will expire or lose approval before use.

Keep the classes useful over time

For a stable store, review ABC at least twice a year as a starting schedule, recalculating sooner after major price changes, a new buyer programme or the end of a repeat style. Review XYZ when enough fresh data show a different pattern, while avoiding frequent class changes caused by tiny movements around a cutoff.

Because items at CV 0.49 and 0.51 have very similar measured variation, they should not receive sharply different policies without a practical reason. A defined review band near the boundary can prompt a planner's check instead of an automatic change.

After changing a stock rule, compare the outcome with its intended purpose by reviewing usable stock, shortages, emergency purchases and obsolete balances over a relevant period. A lower inventory balance is less useful if missed materials now delay cutting or sewing.

ABC identifies the materials that account for most annual usage value, while XYZ adds a separate view of their relative demand variation. The most useful stock plan brings those two views together with lead time, reuse and the cost of being short. Keep those decisions visible so a class remains a guide to action, rather than just a coloured cell in a spreadsheet.

References

1. MIT OpenCourseWare: Inventory Management lecture.

2. Oracle Inventory User's Guide: ABC Analysis.

3. NIST: Coefficient of Variation.

4. SAP: XYZ Analysis.

5. Microsoft: STDEV.S Function.

6. Hyndman and Athanasopoulos: Forecasting: Principles and Practice, practical issues.