Start with the consequences of shortages
For each part, record what its absence would stop and whether a fallback process exists. An inexpensive item can halt an important task. An expensive module need not always be stocked if the supplier provides an acceptable repair route.
Separate consumption from failures
For regularly consumed materials, you can use withdrawal history and delivery lead times. Monthly averages can be misleading for rare failures. Such cases require a decision about criticality and availability, not simply multiplying an average.
Record a reorder point
A working reorder point can be estimated as consumption during the delivery lead time plus an agreed buffer. The HumanoidPoint calculator shows this simple model. It does not predict failures or choose your safety stock level.
Monitor usable stock
Label reserved, damaged and in-transit parts separately, as well as those compatible only with certain revisions. Check storage requirements. Packages on a shelf are not the same as working parts available for installation.