By Elie Dufeu, CTO & Co-Founder, Metreecs. Published 10 September 2026.

What is ABC XYZ analysis? A practical inventory guide

ABC XYZ analysis classifies inventory along two dimensions at once: ABC ranks products by value (usually annual consumption value), and XYZ ranks them by demand variability. Combined, they produce nine categories, from AX (high value, stable demand) to CZ (low value, erratic demand), each pointing to a different way of managing safety stock and reorder frequency.

Most planners know ABC analysis. Fewer apply the XYZ layer, and fewer still keep either one current once the spreadsheet is built. That gap is where the real cost sits: a product classified as stable in January can turn erratic by June, and a static classification stops matching reality long before anyone notices.

This guide covers how ABC and XYZ are each calculated, how the combined matrix works, and where the classic method runs into trouble once you’re managing hundreds of products across multiple stores or channels.

Key Takeaways

• ABC XYZ analysis combines value-based ranking (ABC) with demand-variability ranking (XYZ) to produce nine inventory segments, each with a different service-level and safety-stock strategy.
• XYZ classification uses the coefficient of variation (standard deviation of demand divided by mean demand). A common threshold is CV below 0.5 for X (stable), 0.5 to 1.0 for Y (variable), and above 1.0 for Z (erratic).
• Early adopters of AI-enabled supply chain tools cut logistics costs by 15%, inventory levels by 35%, and improved service levels by 65% compared with slower-moving competitors, according to McKinsey.
• Top-performing supply chain organizations use AI to optimize planning processes at more than twice the rate of lower-performing peers, per Gartner’s February 2024 research.
• A classification built once a year misses the products that migrate between categories mid-season, which is usually where stockouts and overstock cluster.

What is ABC XYZ analysis?

ABC XYZ analysis is a two-axis inventory classification method. The ABC axis groups products by their financial contribution, typically annual consumption value (unit cost multiplied by volume sold). The XYZ axis groups the same products by how predictable their demand pattern is, measured through the coefficient of variation of historical sales.

Overlaying the two axes produces a 3x3 matrix: AX, AY, AZ, BX, BY, BZ, CX, CY, CZ. A product’s letter combination tells a planner not just how much it matters financially, but how hard it is to forecast, which changes how much safety stock it needs and how often it should be reviewed.

Getting this classification right feeds directly into safety stock and reorder policy. See this breakdown of AI-driven inventory optimization for how the two connect.

How ABC analysis ranks products by value

ABC analysis, adapted from the Pareto principle, sorts products by their contribution to total consumption value and splits them into three tiers.

A products typically represent around 70 to 80% of total value from roughly 10 to 20% of the catalog. B products sit in the middle, contributing moderate value from a moderate share of variants. C products make up the bulk of the catalog by count but contribute a small share of total value.

The calculation itself is simple. Multiply each product’s unit cost by its annual units sold to get annual consumption value, rank products from highest to lowest, then calculate cumulative percentage of total value. Products up to roughly 80% cumulative value become A, the next tier to roughly 95% becomes B, and the remainder is C.

Naomi runs replenishment for a mid-market home decor retailer with around 3,200 active products across ceramics, textiles, and lighting. When she first ran an ABC pass, 14% of the catalog (mostly statement lighting and large ceramics) accounted for 79% of annual consumption value. That reordering alone told her buying team exactly where to focus supplier negotiations and where daily stock checks mattered most.

How XYZ analysis ranks products by demand variability

XYZ analysis adds the dimension ABC alone misses: how consistent demand actually is. A product can carry high value and still be nearly impossible to plan for if its weekly demand swings wildly.

The standard metric is the coefficient of variation (CV), calculated as the standard deviation of demand divided by the mean demand over a defined period, usually 12 months.

CV = standard deviation of demand / mean demand

A commonly used threshold set classifies products as follows: X products have a CV below 0.5, meaning demand is stable and forecastable with a simple moving average. Y products fall between 0.5 and 1.0, showing moderate fluctuation, often tied to seasonality or promotions, that needs statistical forecasting rather than a flat average. Z products have a CV above 1.0, meaning demand is irregular or intermittent, and a standard forecast model will consistently miss.

Worked example: a product sells 100, 110, 95, 105, 108, 112, 98, 102, 115, 120, 108, and 110 units across 12 months. The mean is 107.5 units and the standard deviation is roughly 7.5 units. CV = 7.5 / 107.5 = 0.07. That product lands firmly in the X category, well under the 0.5 threshold, and can be forecast with a basic model and modest safety stock.

The ABC XYZ matrix: what to do with each segment

Combining the two axes gives nine segments, and each one implies a different mix of safety stock, review frequency, and forecasting effort.

AX products get the most planning attention per unit sold because they combine high financial weight with predictability, which means small forecasting improvements pay off fast. AZ products deserve just as much attention but for a different reason: they carry real revenue at stake and still resist accurate forecasting, so they often need a wider safety stock band or an alternative sourcing strategy rather than a tighter forecast.

CZ products sit at the opposite end. Low value and erratic demand rarely justify sophisticated planning, and a retailer’s time is usually better spent elsewhere unless a CZ item has strategic value (a loss-leader or a product tied to a key account).

How to run an ABC XYZ analysis step by step

1. Pull 12 months of unit-level sales data for each product, ideally at the product and variant level rather than the category level, since category averages hide the exact variability the XYZ axis is trying to capture.
2. Calculate annual consumption value for each product (unit cost x annual units sold) and rank from highest to lowest.
3. Assign ABC tiers based on cumulative value percentage (roughly 80% for A, 95% for B, the rest C, adjusted to fit the specific catalog).
4. Calculate the coefficient of variation for each product’s monthly (or weekly) demand over the same period.
5. Assign XYZ tiers using the CV thresholds above, adjusted for the retailer’s own demand patterns if the standard cutoffs don’t fit cleanly.
6. Combine both letters to place each product in one of the nine segments.
7. Map each segment to a management strategy: safety stock policy, review cadence, and forecasting method.

Where the classic method runs into trouble

The static, once-a-year version of ABC XYZ analysis has three recurring weak points.

New products have no history to classify. A product launched two months ago cannot generate a meaningful 12-month coefficient of variation. Most teams default new products to a placeholder tier (commonly BY or AY, treated as moderate risk) until enough sales data accumulates, then reclassify at the next review.

Classification drifts faster than most review cycles. A product can move from X to Z within a season, whether from a competitor stockout, a viral moment, or a promotional cannibalization effect nobody modeled. A retailer running quarterly or annual ABC XYZ reviews is planning against a classification that’s already out of date for the products that moved the most. Tracking days of inventory on hand alongside the classification helps flag that drift sooner. See which KPIs matter most for networked inventory control for the full list.

The matrix rarely accounts for location. A product can be AX in a flagship store and CZ in a smaller outlet with a different customer mix. Company-wide classification hides this, which means safety stock policy ends up either too loose in the flagship or too tight in the outlet.

One mistake we repeatedly see is a planning team building a clean, correct ABC XYZ matrix once, treating it as settled, and not revisiting it until year-end, by which point a meaningful share of the catalog has quietly shifted categories. Across Metreecs’ work with multi-location retailers, the products causing the most stockouts and markdowns are rarely the ones the original matrix flagged as high-risk. They’re the ones that migrated segments after the matrix was built.

How AI-driven classification keeps pace with changing demand

The math behind ABC XYZ analysis hasn’t changed since it was first applied to inventory decades ago. What’s changed is how often it’s practical to recalculate it. Metreecs’ AI-powered product x location forecasting recomputes the coefficient of variation continuously as new sales data arrives, rather than on a fixed annual or quarterly schedule, which catches the products drifting into a riskier segment before that drift shows up as a stockout or a markdown.

Early adopters of AI-enabled supply chain tools cut logistics costs by 15%, reduced inventory levels by 35%, and improved service levels by 65% compared with slower-moving competitors, according to McKinsey’s research on AI in supply chain. Gartner’s February 2024 research similarly found that top-performing supply chain organizations use AI to optimize planning processes at more than twice the rate of lower-performing peers. Neither figure is specific to ABC XYZ classification, but both point to the same underlying shift: the operational value of AI in inventory planning comes from doing familiar analysis more often and at finer granularity, not from replacing the method itself.

The same logic applies to forecasting accuracy more broadly. For more on how forecast error compounds across a catalog this size, see the guide on the cost of inaccurate demand forecasts.

Applying the matrix in a real operation

Running an ABC XYZ analysis by channel or vertical surfaces patterns a company-wide view misses, as two examples below show. Elias manages supply planning for a food and beverage franchise network running the FaaS model with an external forecasting partner rather than an in-house data team. Perishability made the XYZ layer critical for his catalog. Items with a CV above 1.2, mostly limited-time promotional items and weather-sensitive beverages, got shifted to a shorter review cycle and a conservative safety stock rule, while stable staples moved to fully automated reordering. Within one full quarterly cycle, the franchise network’s planning team spent noticeably less time manually adjusting orders for the staple items and could concentrate that attention on the erratic tier where it actually mattered.

Priya oversees inventory for a multi-brand beauty retailer managing DTC and wholesale channels with different demand rhythms for the same products. An ABC XYZ pass run separately by channel, rather than blended together, revealed that several products classified as AY overall were AX in wholesale and AZ in DTC, where promotional spikes drove far more variability. Splitting the classification by channel let her team apply tighter safety stock in wholesale and a wider buffer in DTC for the same products, instead of averaging the two into a policy that underserved both. This kind of split only works if the underlying demand signal is accurate to begin with. See how AI demand planning handles channel-level variance.

FAQ

What is the difference between ABC analysis and XYZ analysis?

ABC analysis ranks products by financial value, usually annual consumption value. XYZ analysis ranks the same products by how predictable their demand is, using the coefficient of variation. Used together, they show both how much a product matters and how hard it is to plan for.

How do you calculate XYZ analysis?

Calculate the coefficient of variation for each product: standard deviation of demand divided by mean demand over a defined period, typically 12 months of monthly or weekly sales. A lower CV means more stable demand; a higher CV means more erratic demand.

What are the 9 categories in ABC XYZ analysis?

Combining three ABC tiers (A, B, C) with three XYZ tiers (X, Y, Z) produces nine segments: AX, AY, AZ, BX, BY, BZ, CX, CY, CZ. AX is high value and stable demand; CZ is low value and erratic demand.

How often should you redo an ABC XYZ analysis?

Most retailers review quarterly at minimum, and monthly for fast-moving categories like electronics or promotional-heavy beauty and F&B. Products with a coefficient of variation near a threshold boundary are the ones most likely to have shifted since the last review.

Can ABC XYZ analysis be used for new products with no sales history?

Not directly, since the coefficient of variation needs historical demand data. Most teams assign new products a placeholder tier (commonly moderate risk) based on comparable products, then reclassify once enough sales history accumulates, typically after 8 to 12 weeks.

What is an example of an AX product versus a CZ product?

An AX product might be a retailer’s best-selling staple item, high revenue and steady week-to-week demand, well suited to automated reordering with minimal safety stock. A CZ product might be a low-priced, rarely-purchased accessory with sporadic demand spikes, where the cost of holding safety stock outweighs the revenue it protects.

Conclusion

ABC XYZ analysis remains a dependable way to decide where planning attention belongs: value tells you what matters, variability tells you how hard it is to get right. The method breaks down less because the formula is wrong and more because most teams run it once and let it go stale while products quietly drift between segments. Recalculating it regularly, and at the product x location level rather than company-wide, is what keeps the classification matching reality. See how Metreecs applies this continuously to your own catalog. Book a demo.

Sources

• McKinsey & Company, "Succeeding in the AI supply-chain revolution" - early AI-enabled adopters improved logistics costs by 15%, inventory levels by 35%, and service levels by 65% versus slower-moving competitors. https://www.mckinsey.com/industries/metals-and-mining/our-insights/succeeding-in-the-ai-supply-chain-revolution
• Gartner, "Gartner Says Top Supply Chain Organizations are Using AI to Optimize Processes at More Than Twice the Rate of Low Performing Peers" (February 2024 press release). https://www.gartner.com/en/newsroom/press-releases/2024-02-20-gartner-says-top-supply-chain-organizations-are-using-ai-to-optimize-processes-at-more-than-twice-the-rate-of-low-performing-peers