The service level represents the desired probability of not getting a stock-out. The more product you stock, the lower the chances of running out of stock, but also the higher the inventory cost. Deciding on the right service level for a certain product is essentially balancing inventory costs with the cost of a stock out.

Whilst inventory costs can often be determined rather easily, the cost of stock-outs are much more complicated to express in dollars. A customer that does not find the product in store might either choose an alternative that is in store, postpone the purchase to a later date, or buy at the competition. What is the real cost to the business to lose a customer?

It is also important to understand the relationship between service level and safety stock. A service level approaching 100% gets very expensive very fast, and a service level of 100% is the mathematical equivalent to infinite safety stock.

Nonetheless, Lokad strongly recommends to avoid using service levels anymore. Indeed, service levels are hard to optimize, and usually poorly reflect actual business priorities. Instead, probabilistic forecasts allow for a far more thorough examination of the expected demand for each product in the next lead time window, and have shown to deliver far greater results than the classical service level perspective.

Learn more in the entry Service level of the Lokad knowledgebase.