ABC-XYZ Analysis

Combines ABC Classification (value-based Pareto analysis) with XYZ Classification (demand pattern stability) to create nine distinct segments.

• AX/BX: High-value items with stable demand—ideal for lean, just-in-time strategies.
• AY/BY: High-value items with trending or seasonal demand—require careful forecasting.
• AZ/BZ: High-value items with erratic demand—candidates for make-to-order or high safety stock.
• CX/CY: Low-value items with stable or trending demand—automate replenishment with simple models.
• CZ: Low-value, unpredictable items—review for rationalization or vendor-managed inventory.

Applies the Pareto principle (80/20 rule) to inventory valuation, typically based on annual consumption value.

• A Items: ~10-20% of SKUs, ~70-80% of total inventory value. Require tight control, accurate forecasting, and frequent review.
• B Items: ~30% of SKUs, ~15-20% of value. Moderate control with automated monitoring.
• C Items: ~50% of SKUs, ~5-10% of value. Minimal oversight; focus on administrative efficiency.

The classification is calculated as: Annual Usage × Unit Cost = Annual Consumption Value.

Segments items by the predictability of their consumption patterns using the coefficient of variation (CV).

• X Items: Stable demand with low fluctuation (CV < 0.5). Forecast accuracy is high; suitable for zero-stock or minimal buffer strategies.
• Y Items: Trending or seasonal demand with moderate fluctuation (0.5 ≤ CV < 1.0). Requires time-series forecasting models that account for trend and seasonality.
• Z Items: Highly erratic or lumpy demand (CV ≥ 1.0). Forecasts are unreliable; safety stock must be sized using probabilistic methods or items should be made-to-order.

Coefficient of Variation = Standard Deviation of Demand / Mean Demand.

Each of the nine matrix intersections demands a distinct inventory policy to balance cost and service level.

• AX & BX: Use DDMRP or continuous replenishment with low buffers. High forecast confidence allows lean operations.
• AY & BY: Apply dynamic safety stock that adjusts to seasonal peaks and troughs.
• AZ & BZ: Deploy probabilistic buffers calculated via Monte Carlo simulation or quantile forecasting to cover extreme demand spikes.
• CZ: Consider make-to-order, vendor-managed inventory, or simply holding excess stock given the low carrying cost.

The XYZ dimension directly dictates which forecasting algorithm is most appropriate, preventing model mismatch.

• X Items: Simple models like exponential smoothing or moving averages suffice. Focus on execution, not complex math.
• Y Items: Holt-Winters (triple exponential smoothing) or ARIMA models capture trend and seasonality.
• Z Items: Traditional time-series models fail. Use Croston's method for intermittent demand, demand sensing for short-term signals, or causal models incorporating external drivers.

This structured assignment prevents the common error of applying a single forecast method to all SKUs.

The ABC dimension determines the frequency of planning review and the level of management attention required.

• A Items: Weekly or even daily review. Continuous monitoring of forecast error and supplier performance. Senior planner oversight.
• B Items: Monthly review with exception-based alerts. Automated replenishment with periodic parameter updates.
• C Items: Quarterly review. Use min-max or two-bin systems to minimize administrative overhead.

This tiered governance ensures scarce planning resources are allocated where they create the most value.

The practice of assigning different service level targets to inventory items based on their ABC-XYZ classification rather than applying a uniform policy.

• High-value, stable items (AX): Target 99.5% service with minimal buffers.
• Low-value, volatile items (CZ): Target 90% service to avoid over-investment.
• Strategic items (AY/BY): May warrant higher targets despite lower value due to stockout cost implications.

This ensures capital is allocated where it generates the highest return on inventory investment.

A replenishment trigger level that continuously adjusts based on real-time demand signals and the item's demand volatility clustering behavior.

Unlike static reorder points, this method:

• Raises the trigger for Z-class items during turbulent periods.
• Lowers the trigger for stable X-class items when demand dampens.
• Integrates demand sensing data to react to short-term shifts.

This prevents both premature replenishment of stable items and late reactions for volatile ones.

A demand pattern characterized by frequent zero-demand periods interspersed with sporadic positive demand spikes, typical of many Z-class items.

Standard safety stock formulas fail here because:

• Demand is not normally distributed.
• Average demand is a misleading metric.
• Lead time demand variability is extreme.

Specialized methods like Croston's method and its variants are required to accurately forecast and buffer these items without massive over-stocking.

A safety stock level calculated by balancing the marginal holding cost of additional inventory against the expected stockout cost for each ABC-XYZ segment.

• AX items: High stockout cost, low holding cost relative to value → justify higher buffers.
• CZ items: Low stockout cost, high holding cost relative to value → minimal buffers.
• AY items: Complex trade-off requiring Monte Carlo Buffer Simulation to find the profit-maximizing point.

This moves beyond arbitrary service levels to a financially grounded inventory strategy.

The statistical characterization of historical prediction deviations used to calibrate safety stock for each ABC-XYZ segment.

• X-class items: Tight, symmetric error distributions → standard z-score multipliers work well.
• Z-class items: Wide, skewed, or fat-tailed distributions → require quantile forecasting for accurate buffer sizing.
• A-class items: Even small errors are costly → justify sophisticated Bayesian Safety Stock models that update with every observation.

Ignoring the shape of the error distribution leads to chronic under- or over-buffering.

A Demand Driven Material Requirements Planning inventory buffer composed of green, yellow, and red zones that dynamically resize based on actual demand and lead time factors.

The ABC-XYZ classification directly informs buffer sizing:

• High variability (Z): Expands the yellow and red zones.
• High value (A): Compresses the green zone to minimize working capital.
• Long lead times: Expand all zones proportionally.

The Net Flow Equation (on-hand + on-order - qualified demand) determines current buffer status and replenishment priority.