Troubleshooting

Common MRP and Production Scheduling Mistakes and How to Catch Them

A troubleshooting guide to the MRP, reorder point, and scheduling errors that quietly wreck plans, each with a symptom, a root cause, and a fix you can put a number on.

The most expensive MRP mistake is a wrong or stale lead time. Symptom: parts land 2 to 3 weeks after the planned need date even though MRP said they were on time. Root cause is a static lead time field, often the day the part was set up, while the real supplier average has drifted from 21 to 38 days. MRP offsets orders by that field, so every planned release is late by the drift. Fix: recompute lead time from the last 12 actual receipts using the Manufacturing Lead Time calculator, load the P50 plus one standard deviation, and reschedule. A 17 day error on a 200 line BOM can push a whole build out a month.

Reorder points set on average demand alone cause chronic stockouts. Symptom: a part hits zero once a quarter despite a reorder point that looks reasonable. Root cause is a missing safety stock term, so the ROP equals demand times lead time with no variability buffer. If daily usage averages 40 but spikes to 90, and lead time is 12 days, a demand-only ROP of 480 covers nothing on a bad week. Fix: rebuild it in the Manufacturing Reorder Point calculator with safety stock at Z times sigma times the square root of lead time. At 95 percent service, Z is 1.65, which typically adds 300 to 600 units and kills most stockouts.

Loading a schedule to 100 percent of capacity guarantees you miss dates. Symptom: the plan is theoretically full yet actual output runs 15 to 20 percent under plan every week. Root cause is planning at nameplate hours while ignoring the 12 to 18 percent lost to changeovers, breaks, and minor stops. A cell rated at 120 hours a week rarely delivers more than 100 usable hours. Fix: run the Production Schedule Capacity and Rough Cut Capacity Planning calculators at a demonstrated capacity factor of 0.82 to 0.88, not 1.0. Planning to 105 usable hours instead of 120 removes the phantom 15 hours that never existed.

Netting errors from bad on-hand and allocation data corrupt every downstream order. Symptom: MRP tells you to buy a part you already have three bins of, or skips one you are out of. Root cause is inventory accuracy below 95 percent plus open allocations that were never cleared, so net requirement math starts from a false on-hand. If the record says 500 but the shelf holds 340, MRP under-orders by 160 on every run. Fix: cycle count A items weekly to hold record accuracy above 97 percent before trusting the MRP Material Requirements calculator. Reconcile allocations older than 30 days first.

Ignoring forecast bias silently inflates or starves the whole plan. Symptom: you are consistently long on some SKUs and short on others, every single month. Root cause is a forecast measured only by MAPE, which hides direction, while the real problem is bias, a persistent over or under call. A forecast that is 8 percent high every month builds a standing pile of excess that MRP keeps ordering to. Fix: track signed bias and tracking signal, not just error magnitude, in the Forecast Accuracy calculator. Keep the tracking signal between minus 4 and plus 4; anything outside means reset the base before it feeds MRP.

Promising delivery dates off gross inventory instead of available balance breaks customer trust. Symptom: sales confirms a ship date, then the order cannot ship because the stock was already committed elsewhere. Root cause is quoting from on-hand rather than available to promise, which nets out existing commitments and scheduled receipts by time bucket. If on-hand is 800 but 650 is already allocated across three orders, only 150 is truly promisable this week. Fix: quote from the Available to Promise calculator, bucketed by week, so a 150 unit inquiry gets a real date instead of an optimistic one that slips.

Treating the master schedule as a wish list rather than a load check overloads specific work centers. Symptom: the plant looks balanced in aggregate but two machines run at 130 percent while others sit at 60 percent. Root cause is releasing the master schedule without validating it against resource load, so total hours balance while individual centers do not. Fix: run every proposed MPS through the Master Production Schedule Load calculator and flag any resource above 95 percent before release. Rebalancing 30 hours off an overloaded center onto an idle one recovers on-time delivery without buying capacity.

Ordering in the wrong lot size quietly bleeds cash at both extremes. Symptom: either frequent tiny purchase orders with high freight and setup, or huge buys that sit as dead stock. Root cause is ordering to convenience or supplier minimums instead of the economic tradeoff between ordering cost and carrying cost. If setup is 250 dollars and annual holding is 4 dollars a unit on 12,000 units of demand, the economic quantity near 1,225 beats both a 200 unit reactive order and a 5,000 unit bulk buy. Fix: recompute with the Economic Order Quantity calculator whenever demand shifts more than 20 percent or setup cost changes.

Watching the wrong backlog signal hides a plan going off the rails. Symptom: backlog dollars look flat and stable, so nobody reacts, yet promised dates keep slipping. Root cause is tracking backlog level instead of burn rate, so a stalled queue that clears exactly as fast as it fills reads as healthy while aging orders pile up behind it. Fix: track burn-down slope and days-of-backlog in the Backlog Burn Down calculator. If the queue holds 900 hours and you clear 150 a week, that is a 6 week horizon; when clearance drops to 100 a week, the horizon jumps to 9 weeks even with flat dollars, which is your early warning.

Published 2026-07-01.