I have watched more than one team set a “98% service level,” feel good about it, and then field angry calls about backorders. The reason is almost always the same: they set a Type 1 service level and assumed it promised a Type 2 outcome. Those are different numbers. Conflating them is one of the most common and most expensive mistakes in inventory planning.
Two different questions
- Cycle service level (Type 1) answers: what is the probability that I do not stock out at all during a replenishment cycle? It is about the chance of any shortage, regardless of size.
- Fill rate (Type 2) answers: what fraction of demand do I satisfy directly from stock? It is about the volume of unmet demand, not whether a shortage happened.
A “98%” attached to the first means “I expect to avoid a stockout in 98% of cycles.” A “98%” attached to the second means “I expect to ship 98% of demanded units on time.” Those promise very different things to a customer, and they require different amounts of stock.
Why the gap matters
A 98% cycle service level can coexist with a disappointing fill rate when the stockouts that do happen are large. If you stock out badly in the 2% of cycles where it goes wrong, you can miss a lot of units while your Type 1 number still reads 98%. Fill rate captures that; cycle service level hides it.
The math, briefly
Safety stock under a cycle-service-level target uses the familiar form:
safety_stock = z * sigma_LTwhere z is the standard-normal quantile for your target probability and sigma_LT is the
standard deviation of demand over lead time. That z targets Type 1.
Fill rate is driven instead by the expected shortage per cycle (ESC), the average number of units short when a shortage occurs:
fill_rate = 1 - ESC / Qwhere Q is the order quantity. ESC depends on the demand distribution and the safety stock, via
the standard loss function. The practical consequence: to raise fill rate, a larger order quantity
Q actually helps (it spreads the same expected shortage over more units), which is not obvious if
you only think in Type 1 terms.
What to do about it
- State which one you mean. Write “Type 2 fill rate 98%” or “Type 1 cycle service level 98%” in the policy. The ambiguity is the bug.
- Set the target customers feel. Customers experience fill rate, not cycle probability. For most operations, fill rate is the honest service promise.
- Size safety stock to that target. If you promise a fill rate, compute safety stock from the
loss function and ESC, not from a
zyou picked for a cycle-service number.
Service level and fill rate both belong in your inventory KPIs, but they are not interchangeable. The next time someone says “we run at 98%,” the useful question is: 98% of what?
Implementing this at your scale?
The walkthrough above comes from production work. AvanSaber’s inventory practice has implemented variations of this pattern across multiple customer engagements.
If you are building this and want expert review of your design, or would rather have the team that built this build yours, book a discovery conversation or describe your situation at [email protected].