From inventory target to executable policy: four maturity levels¶

A capital target defines how much the company intends to invest in inventory. An operating policy defines how each item-location will be replenished and protected. The two decisions must connect, but they are not the same decision.

The most common mistake is turning an aggregate target - for example, reducing days of inventory - into proportional cuts for every item and location. The rule is easy to communicate, but it ignores demand, variability, lead time, lots, shelf life, margin, and the consequence of shortage.

The opposite direction also fails: parameters calculated in isolation may protect service without respecting capital availability or network strategy. An executable policy emerges where the two lenses meet.

Two lenses, one decision¶

The financial target should work as a constraint and comparison criterion. The operating policy must explain where inventory will be held, which risk it protects, and how the replenishment flow will execute the recommendation.

A four-level maturity path¶

The levels are not a universal ranking. An organization may operate at level three for regular products while still needing to standardize replenishment for a new line. Diagnose the process and segment, not only the company.

Level 1: standardize before optimizing¶

When allocation depends on spreadsheets, messages, and daily commercial prioritization, the first opportunity is not a sophisticated model. It is a shared, auditable policy.

DRP or Deployment translates demand, balances, receipts, lead time, lots, frequency, and priorities into planned requirements. The planner manages exceptions instead of rebuilding every purchase or transfer manually.

The signal of progress is simple: two people using the same data and parameters should reach the same replenishment recommendation.

Level 2: leave incremental adjustments behind¶

In reactive processes, a shortage period raises the buffer and an excess cuts it. The parameter preserves the memory of incidents but stops representing current risk.

A zero-based review rebuilds policy from:

• expected demand and forecast error;
• average lead time and variability;
• lot, multiple, and replenishment frequency;
• service metric and target;
• shelf life, obsolescence, and shortage consequence.

Segmentation is essential. Items with different velocity, criticality, and substitutability should not receive the same model or target.

Level 3: decide where to protect¶

Single-stage models size each item-location as if its protection were independent. In multi-echelon networks, this can duplicate buffers or protect the wrong point.

Multi-echelon optimization considers dependencies between suppliers, plants, distribution centers, and markets. The question expands from “how much inventory?” to “at which echelon does this inventory best reduce network risk?”.

More mathematical sophistication cannot compensate for weak data or a process that cannot execute the parameters. Level three depends on the discipline built at the previous levels.

Level 4: change the system itself¶

Once parameters and effects are visible, the largest opportunity may not be buffer calibration. It may be an operating-model change:

• reduce lots or increase frequency;
• shorten or stabilize lead time;
• reposition the push-pull boundary;
• adopt postponement or make-to-order for suitable segments;
• revisit channels, locations, suppliers, and capacity;
• share demand, sell-out, and inventory with partners.

At this level, inventory is an outcome of network design, not only a variable to minimize.

How to recognize the next step¶

Data that supports the evolution¶

A gate before moving up¶

Before adding complexity, confirm that:

1. operating parameters are explicit and have an owner;
2. demand, inventory, and receipts share a calendar and unit;
3. the service metric represents the promise made to the market;
4. capital, storage, write-off, and lost margin are kept separate;
5. DRP, purchasing, production, and distribution can execute the recommendation;
6. policy changes are simulated, approved, and versioned.

A mature organization is not the one using the most complex model. It is the one that can explain the decision, execute it, and measure whether the outcome confirmed the hypothesis.

Continue with the economics in How much service is another day of coverage worth?, the product workflow in Inventory policies and optimization, and execution in DRP and Deployment.