Operations Mistakes

Costly Mistakes in Lean Scheduling and Operations, and How to Catch Them

A troubleshooting guide to the recurring errors that break lean scheduling and operations numbers, from mixed time units to phantom capacity, each with a concrete symptom and fix.

The most frequent error is mixing time units inside one calculation. Symptom: a Capacity Planning result that is 60 times too high or too low, or a Lead Time answer in the wrong order of magnitude. Root cause is feeding minutes into a field that expects hours, or entering 480 available minutes as 480 hours per shift. The fix is a units audit before you trust any output: confirm every input carries the label shown beside the field. One plant caught a batch quote that was off by 60x because cycle time was logged at 45 seconds but entered as 45 minutes, inflating the run from 11 hours to 33 shifts.

The second mistake is treating gross capacity as usable capacity. Symptom: schedules that look feasible on paper but slip 15 to 25 percent every week. Root cause is loading a work center at 100 percent of calendar hours and ignoring availability and yield. If a cell has 16 hours per day at 88 percent availability and 92 percent yield, usable output is 16 times 0.88 times 0.92, or about 12.9 hours, not 16. Loading to 16 guarantees overruns. Run the numbers through Capacity Planning with real availability and yield pulled from the last 90 days, not nameplate specs.

A third error is using an average cycle time when the distribution is skewed. Symptom: Lead Time estimates that are right on quiet weeks and badly wrong during peaks. Root cause is that a mean hides the tail; if 80 percent of jobs run 4 hours but 20 percent run 14 hours, the mean of 6 hours understates the queue that forms behind the long jobs. The fix is to plan protected buffers off the 85th percentile, not the average. A promise date built on the mean will miss roughly one order in five when mix shifts toward the long tail.

Ignoring setup and changeover time inside Batch Size decisions is a classic. Symptom: tiny lot sizes that look lean but crush available capacity with setups. Root cause is optimizing for inventory without pricing the changeover. If a changeover costs 45 minutes and you cut batch size from 500 to 100, you add four extra setups per 500 pieces, or 180 extra minutes, roughly 15 percent of an 8-hour shift gone. The fix is to balance carrying cost against setup cost explicitly in Batch Size, and to reserve small lots for parts with genuinely fast changeovers under 10 minutes.

Stale or averaged Machine Hour Rate data quietly poisons every downstream quote. Symptom: jobs that win easily but lose money, or a Unit Cost that never matches the actual close. Root cause is a machine rate built on last year's utilization and depreciation, often assuming 2,000 productive hours when the asset really ran 1,300. That understates the hourly rate by more than 50 percent. Rebuild the rate at least twice a year from actual run hours, current energy cost, and current labor, then feed the corrected figure into Machine Hour Rate and re-check any quote issued in the interim.

Confusing utilization with productivity leads teams to chase the wrong lever. Symptom: a work center reported at 95 percent utilization that still misses due dates. Root cause is counting all busy time, including rework and slow running, as productive. A machine can be 95 percent utilized and only 70 percent effective if it spends time making scrap. The fix is to separate the three OEE components and confirm which one is actually short. Pushing utilization from 95 to 98 percent buys almost nothing if yield is the constraint; that 25 point yield gap is where the real hours hide.

Double-counting or omitting overhead in Inventory Turns and ROI comparisons is common. Symptom: an Inventory Turns figure that looks strong at 8 turns while cash is still tight, or a Manufacturing ROI that swings wildly between analysts. Root cause is inconsistent numerator and denominator, using cost of goods sold in one calculation and sales value in another, or valuing inventory at standard cost in one place and average cost elsewhere. The fix is to lock one valuation basis. Turns computed on COGS over average inventory at the same cost basis is the only version that compares cleanly across periods.

Finally, planners trust a schedule built on infinite capacity assumptions. Symptom: a beautiful Gantt chart where every operation starts on time and nothing ever finishes. Root cause is that MRP and simple schedulers assume every resource is available whenever needed, which no bottleneck ever is. If one furnace is loaded to 130 percent while the plan shows green, the whole downstream sequence is fiction. The fix is a finite check: run the constraint work center through Capacity Planning, confirm load stays under about 85 percent to leave room for variation, and resequence before release rather than after the miss.

Published 2026-07-01.