The reorder point formula has an implicit assumption built into it that most planners know is wrong but treat as acceptable: that supplier lead times are fixed. The formula calculates when to reorder based on a fixed lead time multiplied by demand rate, plus safety stock. If the lead time is 14 days, the formula assumes 14 days. Every time. If the actual lead time varies between 10 days and 22 days, the formula is systematically wrong in a way that produces either stockouts or excess inventory depending on which way the lead time swings.
For suppliers with lead time variability in the 10-15% range, the fixed-lead-time assumption is a reasonable simplification. For suppliers with lead time variability in the 30-50% range - which is not uncommon in industrial and specialty distribution - the assumption actively breaks the reorder point calculation. The fixes are not complicated, but they require measuring what most planning systems do not track: actual lead time history per supplier.
The Mechanics of the Problem
A reorder point set for a 14-day average lead time will produce the expected service level if lead times actually average 14 days. If lead time sometimes runs to 22 days, the inventory that was supposed to last 14 days after the purchase order was placed now needs to last 22 days. If safety stock was calibrated for a 14-day lead time, it is not sufficient for a 22-day lead time. The stockout probability on that extended-lead-time delivery is significantly higher than the target service level implied by the safety stock calculation.
The standard textbook treatment is to incorporate lead time variance into the safety stock calculation: safety stock = Z * sqrt(average lead time * demand variance + average demand squared * lead time variance). This formula accounts for lead time variance as a static parameter, which is an improvement over assuming a fixed lead time. The problem is that lead time variance in most planning systems is set once, during system implementation, from whatever data was available at the time, and never updated.
A supplier that was reliably delivering in 12-14 days three years ago, and now delivers in anywhere from 10 to 26 days due to production scheduling changes, has a dramatically different lead time distribution than when the parameter was set. The formula is still running on 2021 data about a 2026 supply chain relationship.
Measuring Actual Lead Time Distribution
The first step in addressing lead time variability is measuring it. Most ERPs capture both the purchase order date and the receipt date for every inbound shipment. The difference between these two dates is the actual lead time for that delivery. A year of purchase order history for a given supplier produces a lead time distribution that is far more informative than any parameter set during system implementation.
What you want to understand from this distribution: the average, the standard deviation, and the 90th or 95th percentile. The percentile is more useful than the standard deviation for safety stock purposes, because it directly answers the question "what lead time should I plan for to achieve 90% service level?" If the 90th percentile lead time for a supplier is 21 days, then planning for 21 days (not the 14-day average) is what produces 90% service level against lead time variation alone.
This analysis requires pulling a few years of purchase order history and calculating the receipt-minus-order-date for each line. With ERP access, this is a 30-60 minute analysis for any individual supplier. For a distributor with 30 suppliers, doing this once and establishing quarterly update cycles is achievable without dedicated analytics resources.
Probabilistic Lead Time in the Reorder Point Calculation
Once you have the actual lead time distribution, there are two ways to incorporate it into inventory planning. The simpler approach is to replace the fixed lead time parameter with the 85th or 90th percentile lead time (depending on target service level), while keeping the rest of the reorder point calculation unchanged. This is conservative and easy to explain, but it tends to produce higher inventory levels than necessary because it does not account for the fact that demand and lead time are independent variables whose joint variance is smaller than treating the worst case of each separately.
The more accurate approach - which any forecasting system with proper safety stock calculation should support - is to use the lead time distribution parameters (mean and standard deviation of lead time) in the full safety stock formula. This correctly models the combined uncertainty of both demand variability and lead time variability, producing safety stock levels that are appropriately sized for the joint risk rather than oversized from stacking worst cases.
For planners without access to a system that supports the full probabilistic calculation, the simpler 85th-percentile approach is a significant improvement over fixed lead times, even if it is slightly conservative. In practice, the over-stock from conservatism on lead time is usually less costly than the stockout risk from using an optimistic fixed lead time.
Segmenting Suppliers by Lead Time Reliability
Not all suppliers have the same lead time variability. One of the most actionable outputs of a lead time audit is a supplier reliability segmentation: which suppliers are consistently predictable (low coefficient of variation) and which are highly variable (high coefficient of variation).
High-variability suppliers warrant both more safety stock on the items they supply and more frequent monitoring of order status. For critical items supplied by high-variability suppliers, the case for finding an alternate source or holding higher buffer inventory is directly calculable from the lead time data. The carrying cost of the additional safety stock should be compared against the cost of the stockouts that safety stock prevents.
Low-variability suppliers, conversely, may be carrying more safety stock than they need if the original safety stock calculation was set assuming industry-average lead time variance. If a supplier has consistently delivered within a two-day window for three years, the safety stock calculation should reflect that reliability. Over-sizing safety stock on reliable suppliers is a working capital cost that does not purchase any service level improvement.
The Reorder Point as a Dynamic Parameter
The underlying principle is that the reorder point should be a dynamic parameter updated quarterly from actual performance data, not a static parameter set during system implementation and left unchanged until a stockout prompts someone to look at it.
The data to support quarterly recalculation exists in any ERP that tracks purchase order history. The calculation is not complex. The barrier is almost always organizational: establishing the habit of treating reorder points as outputs of a data process rather than configurations set by someone three years ago.
For a mid-market distributor with 50 suppliers, quarterly lead time review on the top 20 suppliers by purchase volume is a half-day process. Over the course of a year, it tends to produce a noticeable improvement in service level on supplier-variability-driven stockouts and a reduction in excess safety stock on reliable suppliers. The math on the ROI is generally favorable. The barrier is the habit, not the work.
