A 98% target sounds customer-focused until it is applied to every SKU, in every warehouse, regardless of demand behavior or margin. The result is usually predictable: too much capital tied up in slow-moving inventory, while genuinely critical items still experience stockouts. To configure service level targets effectively, set the required availability at the item-location level and connect it to the actual cost of being out of stock. A service level is not simply a promise to hold more inventory. It is a replenishment policy that determines how much uncertainty the business is willing to cover with safety stock. Higher targets generally increase product availability, but they also raise inventory investment. The right target balances customer expectations, operational risk, supplier lead times, and the financial value of the item. Start With the Service Measure You Intend to Improve Teams often use the term service level to describe several different outcomes. Before setting targets, define the metric your planning process will use. For most inventory planning decisions, the relevant measure is the probability that demand can be fulfilled during replenishment lead time without a stockout. This is the service level used to calculate safety stock and reorder points. A 95% service level means the planning model aims to have sufficient stock for demand during the lead-time period in 95 out of 100 replenishment cycles. That is different from order fill rate, which measures the proportion of units or order lines fulfilled immediately. It is also different from on-time delivery, a broader measure that can be affected by warehouse execution, transportation, and credit holds. All three matter, but they should not be treated as interchangeable parameters in an ERP replenishment setup. When targets are unclear, planners compensate by increasing minimum stock levels. That may reduce visible shortages for a time, but it obscures the real trade-off. A defined service-level policy makes the inventory investment visible and measurable. Configure Service Level Targets by Item Value and Risk A single company-wide target is easy to administer but rarely efficient. A broad SKU portfolio has different demand profiles, customer consequences, margins, and replenishment constraints. Segmenting items gives planners a practical way to apply different targets without managing thousands of exceptions manually. ABC classification is a strong starting point. A items typically account for a high share of sales value, margin contribution, or operational importance. These products often justify higher service levels because a shortage can affect a major customer, production schedule, or revenue stream. B items require a balanced policy. C items usually need lower targets unless they are strategically critical spare parts or required components. A workable starting policy might set A items at 97% to 99%, B items at 93% to 96%, and C items at 85% to 92%. Those numbers are not universal rules. A low-volume replacement part for installed equipment may need a higher target than a fast-moving but easily substituted retail product. The classification should guide the decision, not replace it. Add operational factors where they materially change the cost of a shortage. Items that stop a production line, support contractual availability commitments, have long or volatile supplier lead times, or cannot be substituted may warrant a higher target. Conversely, seasonal items near end of life, low-margin products, and items with readily available alternatives may justify a lower target. The key is to create a limited set of target bands, then assign exceptions deliberately. If every planner can set a different percentage for every item, the policy becomes difficult to audit and impossible to improve. Set targets at the item-location level An item can require different service levels in different stocking locations. A central distribution center serving several branches may need a higher target because its shortage affects downstream availability. A local branch with frequent transfers from the central warehouse may operate effectively with less safety stock. The same logic applies to e-commerce fulfillment centers, production stores, and regional warehouses. Demand frequency, lead time, transfer options, and customer promise all vary by location. A single item-level setting cannot capture those differences. Use Real Demand Behavior, Not a Static Safety-Stock Formula Service-level targets only work when the demand model behind them reflects how customers actually order. Average monthly demand alone is not enough. Two products can each sell 100 units per month while requiring very different inventory protection: one may sell five units every weekday, while the other sells 100 units in one unpredictable order. Forecast error, order frequency, order quantities, and the distribution of sales orders all affect the probability of a stockout. Intermittent demand is especially easy to mismanage with traditional ERP settings. A static safety-stock number may look reasonable on a report but fail when a customer places a larger-than-average order during supplier lead time. A planning system should forecast demand regularly, calculate variability from current history, and translate the chosen service target into safety stock and reorder-point settings. This allows the target to remain stable as a commercial policy while inventory parameters adjust when demand changes. ABCstock applies this approach by using actual order frequency, order quantities, and sales-order distributions in its service-level simulations. Rather than asking planners to maintain fixed minimums, it recalculates item-location parameters from demand and lead-time behavior, then returns approved settings to the operational system of record. This matters because the relationship between service level and inventory is not linear. Moving from 90% to 95% may require a manageable increase in safety stock. Moving from 97% to 99% can require a much larger inventory increase, particularly for volatile or infrequently ordered items. The final percentage points of availability are often the most expensive. Test the Inventory and Purchasing Impact Before Publishing Changes Do not configure targets in isolation. Simulate the outcome before writing new reorder points back to the ERP. A useful simulation shows how a target change affects safety stock, expected availability, purchase quantities, order frequency, and inventory value. For example, raising an A-item target from 95% to 98% may make sense if the item is essential to a major account and has stable supplier performance. But if the supplier has a 16-week lead time and demand is highly variable, the additional inventory may be substantial. The better decision could be to improve supplier lead time, agree on a stocked program, or position stock at a central location instead. Lowering a target should receive the same scrutiny. Reducing a C-item target from 95% to 88% can release cash without meaningful customer impact when the item is rarely requested and easily substituted. It can be a poor decision when the item is a low-volume but mandatory production component. Review the proposed changes in monetary terms as well as units. Finance leaders need to see the working-capital effect. Procurement teams need to understand whether the policy creates more frequent orders or helps consolidate purchases by supplier. Operations needs confidence that the plan protects the items that matter most. Keep Service Targets Governed, Not Frozen Service-level targets are business rules, not permanent master data. Review them when customer commitments, product life cycles, supplier conditions, or warehouse networks change. A quarterly review is often sufficient for policy bands, while demand forecasts and inventory calculations should update much more frequently. Use exceptions sparingly and document why they exist. Common valid reasons include contractual service commitments, strategic customers, critical spare-parts obligations, production dependencies, and temporary launch or phaseout conditions. When the reason disappears, remove the exception. Track the outcomes by item class and location: stockout frequency, achieved availability, safety-stock value, excess inventory, and purchase-order workload. If service targets are consistently missed despite high calculated inventory, investigate forecast bias, supplier lead-time accuracy, transaction discipline, and allocation rules. The target may not be the problem. A disciplined service-level policy gives planners a better question than “How much stock should we hold?” Ask instead: “What availability do we need for this item, at this location, and what inventory investment does that decision require?” That shift makes inventory planning more transparent, more defensible, and far more useful to the business.