Service Level vs. Fill Rate: Which One Your Retailers Measure

What is the difference between service level and fill rate?

The two terms are used interchangeably in planning meetings, and they measure different things. A target of "95% service" means one thing to the planner sizing safety stock and another to the retail buyer reading a scorecard.

Cycle service level is a yes-or-no test applied to each replenishment cycle. Fill rate is volumetric: it gives partial credit for a partial shipment. What is fill rate covers the unit, line and order versions of the fill-rate formula in detail.

How do the two metrics diverge on the same inventory?

For example, take one SKU over 52 weekly cycles with average demand of 100 units a week and a stockout in 4 of those weeks.

For example, both plans report the same 92.3% service level. Plan A barely disappointed anyone. Plan B missed four full weeks of orders, and a retail buyer's in-full score would show it.

Is fill rate always higher?

Usually, but not by rule. A cycle that stocks out still ships most of its units, so fill rate tends to sit above cycle service level. The gap narrows or reverses when:

• Cycles are short. Frequent small replenishments mean more cycles exposed to a stockout, and each miss is a larger share of that cycle's demand.
• Shortages are deep. If a stockout tends to wipe out most of a cycle's demand, the two numbers converge, as in Plan B.
• Demand is lumpy. A few large wholesale or retail orders per cycle make each shortfall large in units.

There is no general conversion between the two. Calculate both from your own order history.

Which metric does a retail scorecard actually charge you on?

Retail compliance programs measure delivery against the purchase order, and the in-full part of that measure is fill rate. Walmart's on-time, in-full (OTIF) program is the best-known example. Its published goals are 90% on-time for prepaid suppliers, 98% for collect, and a 95% in-full goal measured on cases, with a fine of 3% cost of goods on cases that miss the goal.

Two consequences follow for planning:

• A high cycle service level does not protect the in-full score. The scorecard counts cases short, so one deep stockout on a large purchase order can cost more than several shallow ones.
• Allocation matters as much as stock. When a SKU is short across several open purchase orders, how you split the available units decides which accounts see the miss.

How to avoid retail chargebacks covers the compliance side, including the delivery windows that make up the on-time half.

How does safety stock map to each metric?

Safety stock for a cycle service level target

The standard formula is built on cycle service level:

Safety stock = Z × σ_LT

σ_LT is the standard deviation of demand over the lead time, and Z is the standard normal value for the target. Z is 1.28 for a 90% cycle service level, 1.65 for 95%, 2.05 for 98% and 2.33 for 99%. Z sets the probability that lead-time demand stays below the reorder point, which is the definition of cycle service level. What is safety stock walks through the inputs, and safety stock formulas compared covers the variants that add lead-time variability.

Safety stock for a fill-rate target

A fill-rate target needs one more input: the order quantity. The calculation runs in three steps.

1. Allowed units short per cycle = (1 − target fill rate) × order quantity.
2. Find the safety factor k at which the expected units short per cycle, σ_LT × G(k), equals that allowance. G(k) is the standard normal loss function, read from a table or calculated in a spreadsheet.
3. Safety stock = k × σ_LT.

For example, with an order quantity of 1,000 units, a σ_LT of 200 units and a 99% fill-rate target, the allowed shortfall is 10 units per cycle, so G(k) must be 0.05. That gives k of about 1.25 and safety stock of about 250 units. A 99% cycle service level on the same SKU would need Z of 2.33 and 466 units.

The order quantity matters because a larger order means fewer cycles, so fewer exposures to a stockout. A SKU ordered once a quarter reaches a given fill rate with less safety stock than the same SKU ordered weekly.

How should targets differ by SKU?

Not every SKU needs the same target. Set them by what a miss costs:

• Retail-scorecard SKUs: set the target on fill rate, at or above what the retailer's in-full measure requires, and check the safety stock that implies.
• High-volume DTC and Amazon SKUs: a cycle service level target works, because each customer order is small and a stockout is close to all-or-nothing for that period.
• Long-tail SKUs: lower targets. The last few points of service level cost the most inventory, because the Z value rises faster the closer the target gets to never running out.

State which metric a target refers to whenever it is written down. "95%" means nothing until someone says 95% of what.

Common mistakes

• Reporting one metric as the other. A 95% fill rate says little about how often you ran out.
• Sizing safety stock with Z for a fill-rate commitment. The result is usually more stock than the fill-rate target needs on large orders, and less than it needs on small, frequent ones.
• Ignoring backorders in the fill-rate calculation. Decide whether a backordered unit that ships later counts as filled, and apply the rule consistently. What is a backorder covers the accounting.
• Fill rate higher than demand. Units shipped cannot exceed units ordered. A result that does means shipments are being matched to the wrong orders.

How does Planster handle safety stock?

In Planster you set safety stock in days of cover for each SKU at each warehouse, and Planster converts those days into units from the forecast for the weeks ahead. The forecast comes from a statistical model fitted to that SKU's own history, so the buffer rises into a seasonal peak and falls after it. Planster has no service-level setting. To turn a cycle service level or fill-rate target into days of cover, ask the agent: it can help you calculate the number with the methods above, and you set it. Reorder points and order quantities are recalculated as the forecast and lead times change, and purchase orders are drafted for your team to approve. See reorder recommendations for how the calculation runs.

Planster has no prebuilt fill-rate or OTIF report; reporting is custom. Measure those from your shipment and retailer scorecard data, and use the results to decide whether a SKU needs more or fewer days of cover.