Lead Time Variability, Stockouts, and Why Forecast Accuracy Is Not the Whole Story

Forecast accuracy gets enormous attention in supply chain planning. It is the metric that planning teams are most commonly measured on, the improvement that most planning technology investments promise to deliver, and the explanation most often given for stockouts and service failures.

The attention is deserved - better forecasts do improve supply chain outcomes. But the focus on forecast accuracy as the primary lever for service level improvement can obscure the other factors that drive stockouts, sometimes more powerfully. Lead time variability, inventory positioning, allocation decisions, and inventory aging risk each play significant roles that are frequently under-measured and under-managed.

Understanding the full picture of what causes stockouts is not an academic exercise. It is the difference between making the right investment to improve service and making the wrong one.

The Real Causes of Stockouts: A More Complete Picture

Stockouts rarely have a single cause. When a customer cannot receive the product they ordered, the immediate cause might be an empty bin - but the chain of events that led to that empty bin almost always involves multiple contributing factors.

The most common contributing factors, in approximately decreasing order of how often they appear in stockout root cause analyses, are:

Demand forecast error: Actual demand exceeded the forecast, so safety stock was insufficient to cover the gap. Supply disruption: A supplier delivered late, delivered short, or did not deliver at all - and the safety stock was depleted before the supply arrived. Inventory positioning error: Stock existed in the network but was in the wrong location - excess at one depot, zero at the depot where demand occurred. Allocation decision failure: When supply was constrained, it was directed to the wrong demand - a lower-priority customer received stock that a higher-priority customer needed. Lead time variability: The supplier's actual lead time exceeded the planning assumption, depleting safety stock before replenishment arrived.

Notice that only the first factor - demand forecast error - is directly addressed by improving forecast accuracy. The other four require different interventions: supplier performance management, network-level inventory balancing, allocation decision frameworks, and safety stock calibration for actual lead time variability. Looking at these together reveals bigger opportunities than focusing on forecasting alone.

Lead Time Variability: The Overlooked Driver

Of all the factors that drive safety stock requirements, lead time variability is the one most consistently under-measured and under-managed. This is partly because it is a supplier performance problem rather than an internal planning problem - and the tendency is to treat it as background noise rather than a variable that can be actively managed.

The mathematics of safety stock calculation makes the impact of lead time variability clear. The standard safety stock formula includes both demand variability (the standard deviation of demand during the lead time) and lead time variability (the variability of the lead time itself multiplied by average demand). When lead times vary significantly, this second term can dominate.

Consider a supplier whose stated lead time is 14 days. If that lead time is consistent - always arriving on day 14 - the planning system can set safety stock based purely on demand variability over 14 days. If the same supplier's actual lead time ranges from 10 to 21 days, the planning system must hold enough stock to cover 21 days of demand in the worst case, even though average demand over 14 days is the central scenario.

Reducing that supplier's lead time variability - not necessarily reducing the average lead time, just reducing the variance - directly reduces the safety stock required to maintain the same service level. This can deliver larger inventory reductions than equivalent improvements in forecast accuracy, particularly for categories where demand is relatively stable but supply is not.

The implication is that supplier lead time performance data should be tracked, analysed, and incorporated into safety stock calculations - not as a background assumption but as an active variable. Suppliers with high lead time variability should carry a safety stock premium; improvements in their reliability should flow through directly to inventory reduction.

What Forecast Accuracy Can and Cannot Do

Better forecast accuracy genuinely improves supply chain performance. When the demand signal is more accurate, less safety stock is needed to cover the gap between forecast and reality, replenishment decisions are better calibrated, and production and procurement plans are more aligned with actual demand.

But forecast accuracy has diminishing returns. The relationship between accuracy improvement and inventory reduction is not linear - reducing forecast error from 30% to 20% delivers more benefit than reducing it from 10% to 5%. Beyond a certain point, further accuracy improvement is expensive (requiring more sophisticated models, more data, and more computational investment) and delivers limited incremental benefit.

More fundamentally, forecast accuracy addresses only one of the five factors that cause stockouts. An organisation that achieves excellent forecast accuracy but does not address lead time variability, positioning errors, or allocation failures will still experience service level problems - just with fewer demand-related causes.

The productive framing is to treat forecast accuracy as one of several levers in a service level improvement programme, not as the primary or only lever. When a stockout occurs, the question should be "which factor caused this?" - not "was the forecast wrong?" The answer determines the right intervention.

Inventory Positioning as a Service Lever

Inventory positioning - where in the distribution network stock is held - has a significant and often under-recognised impact on service level. An organisation with adequate total inventory but poor positioning will experience stockouts at specific locations while excess sits unused elsewhere. This pattern is both costly (the excess carries holding cost) and damaging to customer relationships (the stockouts create service failures).

The positioning problem is particularly acute for organisations with high demand uncertainty at the location level. When it is genuinely unclear whether the next unit of demand will come from depot A, depot B, or depot C, holding stock centrally and using rapid replenishment to serve actual demand is typically more efficient than positioning stock peripherally based on a forecast that may be wrong.

This is one of the core arguments for centralised or semi-centralised inventory strategies - not because centralisation is inherently superior, but because it allows risk pooling. Demand variability that would require separate safety stocks at each peripheral location can be partially absorbed by a central buffer, reducing total safety stock while maintaining service levels.

Getting positioning right requires understanding which SKUs benefit from centralisation (high variability, moderate demand) and which require local positioning (time-sensitive, high-volume, predictable demand). This is a classification exercise that most organisations have not completed for their full SKU base, and the inventory benefits of doing it can be substantial.

Inventory Aging: The Other Side of the Service Equation

Most discussion of inventory and service focuses on the shortage side - stockouts and service failures caused by too little inventory in the right place. But the excess side has its own set of risks that deserve equal attention: inventory aging and obsolescence.

Slow-moving inventory creates risk in two forms. The first is direct financial loss through markdowns, write-offs, or disposal costs when stock ages past its useful life or reaches the end of a product lifecycle. The second is opportunity cost - warehouse space, working capital, and management attention consumed by inventory that is not contributing to service or revenue.

Early visibility into inventory aging risk is substantially more valuable than late visibility. An item that is flagged as slow-moving three months before its expiry or end-of-life date can be rebalanced to a higher-demand location, offered through a promotional channel, or included in a discount programme - all at a lower cost than the alternatives available when the problem is discovered with two weeks remaining.

Yet most organisations do not have systematic early-warning processes for inventory aging. The alert - if it comes at all - comes when the item has already aged past the point where the lower-cost interventions are still available. Building proactive aging risk identification into the planning process is a direct investment in margin protection.

How These Factors Interact - and Why That Matters

The five factors that contribute to stockouts do not operate independently. They interact in ways that can amplify each other if not managed together.

A supplier disruption that causes a stockout is more damaging when forecast accuracy is poor, because less safety stock was held to absorb the disruption. A positioning error is more costly when lead time variability is high, because the time required to correct the imbalance through lateral replenishment is longer and less predictable. A poor allocation decision during a constrained period creates a service failure that a better forecast would not have prevented.

Understanding these interactions is what makes supply chain decision intelligence valuable beyond any single analytical module. A platform that can identify the combination of factors contributing to a service risk - and model the impact of interventions that address multiple factors simultaneously - consistently produces better outcomes than point solutions that address demand forecasting, lead time management, or network optimisation in isolation.

Frequently Asked Questions

The Bottom Line

Stockouts are a multi-factor problem that requires a multi-factor response. Organisations that look beyond forecast accuracy - to lead time variability, network positioning, allocation discipline, and aging risk management - consistently find larger and more durable service level improvements than those that treat forecasting as the primary lever.

The question is not "how accurate is our forecast?" It is "which of the five factors is most contributing to our service failures - and what is the fastest path to addressing it?"