Procurement Lead Time Forecast Error

A buyer expects a supplier to deliver within twelve days, but recent orders take longer, and procurement lead time forecast error compares the lead time predicted when an order was placed with the actual elapsed time under a consistent start and endpoint. It tests forecast quality, not only supplier punctuality.

Define the forecast timestamp by saving the lead time that planners genuinely used before the outcome was known, since a later overwritten forecast is not an honest baseline. Define lead time carefully, as APQC describes supplier material lead time from purchase order transmission to the buyer's receipt, while a business may also track approval-to-receipt or order-to-usable-stock, which are different intervals.

Specify the unit, since calendar days, working days and hours can differ across weekends and regions and one convention should be kept per series, and match the actual endpoint so that a forecast of physical arrival is compared with arrival, not quality release. Include buyer delays only when intended, because a purchase order held internally for approval is not supplier lead time under a transmission-to-receipt rule.

Separate forecast types, since a supplier's quoted time, a planner's statistical estimate and a contracted target are different, and track item and supplier because lead times vary by product, origin, shipping mode and season and a pooled average can hide chronic errors for a critical component. Measure signed error as actual minus forecast, which is positive when delivery takes longer than expected under this convention, and absolute error, because positive and negative misses can cancel in a mean signed error while absolute error captures typical magnitude.

Use percentage error cautiously, since it becomes unstable when forecast lead time is near zero and days of error may be easier to understand. Keep open orders visible, as a report of completed receipts alone ignores current orders already later than their predicted arrival, and handle partial delivery by stating the completion rule, since a first pallet arriving on time is not the full order if the forecast was for complete receipt.

Account for changes, because a buyer who changes quantity or specification may reset the planning estimate and the original and revised forecasts should be kept separately, and record extraordinary disruption, as a port closure can create a genuine miss and the forecast process may need a scenario adjustment without erasing the old result. Do not blame the supplier from this metric alone, since customs, buyer approvals or chosen carrier can affect actual elapsed time depending on contract terms.

Assess bias over time, because a recurring positive signed error suggests planners are systematically expecting deliveries too soon under the defined cohort, and check samples by showing median and tail alongside the mean. Compare horizons, since a same-day estimate made just before arrival should be more accurate than an estimate made at order release and the two should not be mixed, and test forecast revisions by saving each revision timestamp and the new expected arrival so managers can see whether changes improved planning ahead of time.

APM guidance on bias warns that optimistic project estimates can understate time, and procurement forecasts also deserve a check for systematic optimism, without assuming the same cause. Assess operational effect, because the same two-day error may be harmless for buffer stock and severe for a production bottleneck, so pair the metric with stockout or disruption measures.

Review prediction inputs, updating them without using hindsight data in old forecasts, and use results to recalibrate by changing the forecast model or supplier planning assumptions and comparing later orders on the same basis.