Inventory Optimization Playbook

Problem

Most manufacturers apply the same replenishment logic and safety stock rules across their entire catalog, which means high-value stable items are under-served while slow-moving erratic items accumulate into dead stock. The result is capital trapped in the wrong places: excess inventory on C-items that nobody needs urgently, and stockouts on A-items that stop production lines. Companies operating below median inventory turns are typically carrying 25-40% more working capital than necessary, and write-offs on obsolete stock quietly erode margins every quarter.

Step-by-step approach

1. 1

   Run ABC classification on annual cost of goods consumed

   Pull 12 months of consumption data valued at standard cost. Rank every SKU by total annual consumption value and classify: A-items are the top 80% of value (typically 15-20% of SKUs), B-items the next 15%, and C-items the remaining 5%. Do not use revenue — use COGS or standard cost, because that is what is sitting in your warehouse. This tells you where your capital is concentrated and where a stocking error costs real money.

2. 2

   Add XYZ demand variability overlay

   For the same 12-month period, calculate the coefficient of variation (standard deviation divided by mean) of monthly demand for each SKU. Classify as X (CV below 0.5 — stable, predictable), Y (CV 0.5-1.0 — moderate variability), or Z (CV above 1.0 — erratic or lumpy). Combine with ABC to create a 9-cell matrix. Your AX items are your bread and butter — high value, highly predictable. Your CZ items are your dead-stock-in-waiting. Treat them differently.

3. 3

   Set differentiated service level targets by segment

   Assign target service levels based on the matrix: AX and AY items get 97-99% fill rate targets because a stockout stops your line or loses a major customer. BX and BY get 93-95%. CX and CY get 85-90%. AZ items are the tricky ones — high value but erratic — and they need a case-by-case review, not a blanket policy. CZ items should be evaluated for make-to-order or discontinuation. Write these targets down and get sign-off from sales and operations.

4. 4

   Recalculate safety stock using segment-appropriate methods

   For X-items with stable demand, use standard statistical safety stock formulas based on demand variability and lead time variability — the math works well here. For Y-items, add a demand-sensing adjustment and review monthly. For Z-items, do not use statistical safety stock at all — it will either be absurdly high or meaninglessly low. Instead, use min-max with a cap, or switch to make-to-order. Recalculate quarterly as demand patterns shift.

5. 5

   Run a dead stock purge and establish ongoing governance

   Identify every SKU with more than 180 days of supply on hand and no consumption in the last 90 days. Build a disposition plan: return to supplier, markdown, reclassify, or write off. Then set up a monthly excess and obsolete review with finance — if it is not on the calendar, it will not happen. Track inventory turns by ABC segment quarterly and compare against your targets. A-item turns should improve within 60 days of implementing differentiated policies.

What good looks like

Top-quartile operations run differentiated stocking policies across every segment of their inventory — not a single reorder point formula applied to thousands of SKUs. They review ABC-XYZ classifications quarterly, adjust safety stock parameters when demand patterns shift, and hold a monthly excess-and-obsolete review with financial accountability. Their planners spend time on the 20% of SKUs that drive 80% of value, and they automate replenishment on the stable, low-value tail.

Common failure modes

Inventory optimization initiatives fail most often because teams do the segmentation analysis once, set new parameters, and never revisit them — demand patterns change and the classifications go stale within two quarters. The second failure mode is applying statistical safety stock formulas to erratic-demand (Z-class) items, which produces either absurd buffer levels or false confidence in thin coverage. Third, many organizations segment inventory but do not actually differentiate their service level targets, which means the segmentation was an academic exercise with no operational consequence. Finally, skipping the dead stock purge means you are optimizing around a base that includes items nobody will ever use — you cannot turn inventory that is not moving.

This playbook is based on:

• EyeOn — ABC/XYZ Segmentation for Inventory Optimization, Journal of Business Forecasting (2024)
• ISM — Safety Stock Formula and Inventory Policy (2025)
• APICS Magazine — Inventory Stratification Optimizes Results (2018)
• ASCM CPIM Body of Knowledge — Planning and Inventory Management (2024)