Compare policies
Compare the available safety-stock policies and choose the one with the lowest expected total cost for each SKU.
AInventory evaluates five replenishment policies (Fixed Coverage, Analytic Newsvendor, Proportional Variance, Historical Standard Error, Empirical Bootstrap) over a four-level statistical cascade, choosing the one with the lowest expected total cost. The 2026 User Manual shows a simplified two-policy view (Dynamic/RMSE); we document the five-policy model here. See ESTRUCTURA.md §5.1.
Before you start
- The SKU must have holding and shortage costs configured (see Configure per-SKU parameters).
- Sufficient demand history is required (at least one full cycle) for the statistical policies to be valid.
Steps
1. Open the SKU simulator
From Inventory → SKU List, click the SKU you want to analyze.
The SKU detail panel opens, including the Policy simulator section.
2. Review the total cost for each policy
The simulator displays, for each policy, the expected total cost calculated as:
Total cost = Holding cost × average inventory + Shortage cost × expected stockout units
Look at the Total cost column in the comparison table. The policy with the lowest value is the most suitable for that SKU under current conditions.
The policy with the lowest total cost is highlighted automatically. No manual calculation needed — the system already marks the recommended one.
3. Choose the policy
- To accept the system recommendation, click Apply recommended policy.
- To choose a different policy (for commercial or operational reasons), click that policy's name and then Apply selected policy.
The change is recorded and the projected fill rate is recalculated with the new policy.
The five policies
| Policy | Brief description |
|---|---|
| Fixed Coverage | SS calculated as fixed coverage days × average daily demand. Simple; works well when lead time is very stable. |
| Analytic Newsvendor | Optimizes service level using the critical ratio (shortage cost / holding + shortage cost). Requires a known demand distribution. |
| Proportional Variance | SS proportional to the variance of demand over the lead time. Calibrates well when variability is high. |
| Historical Standard Error | SS based on the forecast RMSE over the lead time period. Directly tied to forecast accuracy. |
| Empirical Bootstrap | SS derived from the empirical demand distribution via resampling. More robust when demand is not normally distributed. |
For detailed theory on each policy, see Replenishment policies.
Expected result
After applying a policy, the panel displays the projected fill rate with the new SS level and the estimated cost change relative to the previous policy.