Companies need to know how well they are fulfilling customers' needs. Commonly used metrics in practice are fill rate and service levels. In traditional inventory management, fill rate and service levels are used to determine safety stock requirements in order to achieve the desired fulfilling percentage. Although often used interchangeably, these two metrics are measuring different things and show significant differences in safety stock values. The goal of this article is to show how demand uncertainty interacts with fill rate, service level and safety stock.
Conceptually, the difference between service level and fill rate lands in the perspective in which they look at fulfillment. Service level is calculating the fraction of replenishment cycle that ends with all customer demand met, while fill rate is calculating the fraction of demand that is satisfied from available inventory. In other words, service level is measure fulfillment from a customer angle while fill rate is measuring it from a product angle.
Let's pretend that a lockdown hits because of a pandemic. People decide to go to the local store and buy paint to fix their house while being quarantined. On average, the store sells 1 bucket of paint everyday. Assuming that the store keep 50 units of inventory in stock and the lead time to replenish is 1 day. Now, 50 customers walk in and each of them buys 1 bucket of paint, each of them can walk away satisfied because they got what they were looking for (units were available). But, one additional customer walks in the store, he wants to purchase 50 buckets of paint (he has a very big house). The store would not be able to satisfy this customer because it already sold the 50 units on hand. In terms of service level, the store is doing well because 50 out of 51 customers' orders were fulfilled, meaning that service level is slightly over 98%. However, in terms of fill rate, the store was only able to fulfill 50 units out 100 units meaning that the fill rate is 50%.
From this example, we see how fill rate and service level behave differently when demand is erratic (steady demand of 1 unit and a sudden 50 units order).
This erraticity of the demand is representative of real world supply chains. This is why service levels and fill rates are determined alongside with safety stocks which is a fraction of the stock on hand serving the purpose of buffering against uncertainty during the exposure period (lead time and review time).
As we may guess by now, a safety stock calculated from the service level will look different from a safety stock calculated with the fill rate.
Showing the steps on optimally determining service level and fill rate and calculating safety stocks is beyond the scope of this article but formulas are accessible and can be found online. What is of interest for us is the economic and inventory quantities differences in using fill rate versus service level to determine safety stock while facing uncertainty.
Let's take a look at this table:
We are here assuming that demand averages 100 units for one product. Safety is calculated for service levels and fill rate values ranging from 80% to 99.99%. We also injected some variability in the average demand to show how safety stock varies from fill rate and service level in regards to variability.
We can see how choosing one method of establishing safety stock over the other has tremendous consequences on the amount of safety stock kept in hand. Establishing safety stock with service level is resulting in keeping a much higher safety stock for this example. Now let's pretend that the cost per unit of safety stock is 2$, the table below shows the savings realized from using fill rate to determine safety stock over service level.
The savings realized just in unit costs (holding cost is omitted), is pretty significant between safety stock calculated with service level and fill rate. We can also notice that savings from service level to fill rate increases as uncertainty increases too. In 99% of real world supply chain cases, uncertainty is always a factor to account for in forecasting and replenishment decisions. This is why service levels and fill rate are useful because they account for the trade off between cost of stock (having inventory units on hand) and cost of stockout (not being able to service units or customers). In other words, we want to buy and hold enough stock to satisfy the most customers's demand but holding too much stock could result in losses. The two graphs above illustrate this trade off pretty well, as we can see that the amount of units to keep in safety tends to grow exponentially as the service level or fill rate grow toward 100%.
To conclude this brief article, it is very important to be aware of the difference between fill rate and service level. Claiming that business is striving because service levels are over 95% could be miss-guided and may not be reflective of the reality. In addition, being aware of what algorithm your ERP is using to determine safety stock is also of primary importance as we saw that significant savings could be achieve using the proper method. Doing this simple analysis can give a better idea of what method to choose. As uncertainty will always be there in supply chain, and most likely continue to increase as more options become available to customers, being critical about the tools we are using, knowing them and understanding where results and values are coming from is the very first step toward a more optimized supply chain.