Should the safety stocks built in Supply Chain inventory buffers be based on demand variability?

Inventory buffers in Demand Driven Supply Chains are set at ‘Maximum demand during RLT, adjusted for supply reliability’. When we define maximum demand, it has a built-in safety stock, over and above the normal demand. Common practice is to base the safety stock on variability of demand.

However, companies keeping safety stocks based on demand variability end up keeping relatively higher safety stocks. That’s because demand variability is not entirely unpredictable. There are several elements of variability which can be predicted to a large extent. For example, weekly cyclicity in case of retailers, month-end skews in FMCG, impact of promotions and festivals.

If we use these underlying factors as demand drivers, it is quite possible to predict a substantial part of the demand variability and build it in the demand signal. Safety stock should then be only to the extent of demand prediction errors, which are much smaller than the overall demand variability. As a result, we get better product availability with lower inventory.

We must change our thinking from demand variability to demand predictability!