Setting the Right Service Level Without Letting Inventory Costs Run Away
Every supply chain manager eventually meets the same tension: hold more inventory and you almost never disappoint a customer, but you tie up cash and warehouse space; hold less and you free up money but risk a stockout at the worst possible moment. After the disruptions of 2020 and early 2021, plenty of organizations swung hard toward holding more of everything. That instinct is understandable, but buying blanket safety stock across the board is an expensive way to feel safe. The better approach is to set service levels deliberately, item by item, and let cost and risk decide where to spend.
Service level, in plain terms, is the probability that you can fill demand from stock on hand during the replenishment lead time. A 95 percent cycle service level means that, on average, you expect to run short in roughly one replenishment cycle out of twenty. Pushing from 95 to 99 percent sounds like a small step, but the safety stock required climbs much faster than the number suggests, because you are buying protection against rarer and rarer demand spikes.
Start by segmenting, not by guessing
The mistake is treating every SKU the same. A short ABC or criticality analysis usually reveals that a small share of items drives most of the revenue, and a different small share would cause real harm if it ran out. Sort your catalogue before you set a single target.
Rank by impact. Classify items by revenue contribution and by the consequence of a stockout. A cheap fastener that halts a production line is more critical than its price implies.
Assign target service levels by tier. Reserve your highest targets for items that are both high-value and high-consequence. Let low-impact, easily substituted items sit at a lower, cheaper target.
Account for lead-time variability. An item with an unpredictable supplier needs more buffer than an item with a steady one, even at the same target. Variability, not just average lead time, drives safety stock.
Do the cost math before you commit
Once you have target service levels, translate each into a safety-stock quantity using your demand and lead-time variability, then price it. Multiply the extra units by your true carrying cost, which is more than warehouse rent: it includes capital, insurance, handling, obsolescence and shrinkage, often 20 to 30 percent of unit value per year. On the other side, estimate the cost of a stockout, including lost margin, expediting fees and the harder-to-quantify damage to a customer relationship.
If raising a target from 95 to 98 percent adds modest stock for an item that customers will abandon you over, spend it.
If the same jump adds large carrying cost for an item customers will happily backorder, hold the line.
Revisit targets when demand patterns shift; a level set during a shortage may be wasteful once supply normalizes.
This is also where pooling helps. Centralizing stock across regions, or sharing buffer between similar items, reduces the total safety stock needed to hit the same service level, because aggregated demand is proportionally less variable than the sum of its parts. Hybrid teams managing inventory remotely should make sure the data behind these decisions is shared and current, not locked in one person's spreadsheet.
Make it a habit, not a one-time fix
Service-level targets are not set-and-forget. Track your achieved fill rate against the target, watch how often you expedite, and review the cost of carrying versus the cost of shorting at least quarterly. The aim is not a perfect record of never running out; that record is a sign you are overspending. The aim is to be short only on the items, and only as often, as the economics justify.
If you want a disciplined, defensible way to set these targets across your catalogue, XNM's procurement, sourcing & contract management can help you build the segmentation and cost models that make the trade-off explicit.