Throughput

A Plant Leader's Playbook for Managing Throughput

Throughput is the number the P&L actually feels. Here is how to model it from cycle time, uptime, and yield, where the units leak out, and the hourly cadence that gets them back.

Throughput is the only production number the P&L feels directly. Fixed costs are spread across every good unit out the door, so a line absorbing 40,000 dollars a day of overhead at 500 units carries 80 dollars a unit; lift output to 550 and the burden drops to 73 dollars, a 9 percent cost reduction with zero spending. When throughput is your constraint, every additional good unit sells at full margin, so a line making 30 dollars contribution per unit that recovers 50 units a day recovers 1,500 dollars daily, about 375,000 dollars a year. That is why throughput gets managed hourly, not reviewed monthly.

Model throughput as a chain of three factors: theoretical rate, uptime, and yield. A 45 second cycle gives 3,600 divided by 45, or 80 units per hour theoretical. At 88 percent uptime that becomes 70.4 units per hour, and at 97 percent first-pass yield, 68.3 good units per hour. Across 7.5 net run hours, that is 512 good units a shift, not the 600 the nameplate implies. The Throughput calculator stacks these same losses into hourly, shift, and daily projections, and the gap between 600 and 512 is your improvement agenda written as a number.

Benchmark the loss stack, not just the total. Well-run discrete lines hold uptime at 90 to 95 percent, first-pass yield above 98 percent, and realize 80 to 90 percent of theoretical rate overall. Typical lines realize 60 to 75 percent. The diagnostic question is which factor sits furthest from benchmark: uptime at 78 percent points to maintenance and changeover work, yield at 93 points to process control, and a realized rate 15 percent under demonstrated cycle points to minor stops and speed losses that never make the downtime log.

Prioritize levers by the math of multiplication. Because the factors multiply, gains compound: taking uptime from 88 to 92 percent and yield from 97 to 98.5 lifts the example line from 68.3 to 72.5 good units per hour, a 6.2 percent gain from two modest moves. Work the constraint station only; a 10 percent speedup at a non-constraint adds zero units to the system. Typical high-payback moves: SMED to cut changeovers 30 to 50 percent, sensor-triggered minor stop counting on the top three fault locations, and startup parameter recipes that cut first-hour scrap roughly in half.

Watch the standard failure modes. Counting total units instead of good units, which lets a yield problem masquerade as healthy throughput. Chasing rate on non-constraint equipment, which only builds WIP; queues ahead of the bottleneck grow at exactly the rate difference, so a 10 unit per hour mismatch is 80 units of new WIP per shift. Averaging throughput across the week, which hides the fact that Monday runs 15 percent below Wednesday because startup discipline is weak. And treating theoretical rate as the plan, then explaining a permanent 20 percent miss every single day.

Run throughput on an hourly cadence. Post plan versus actual by hour at the constraint; any hour missing plan by 5 percent or more gets a reason code before the next hour closes. Daily, the production meeting reviews the last 24 hours of codes in 10 minutes and assigns one recovery action. Weekly, reconcile the loss stack: how much output was lost to downtime, speed, and scrap, in units and dollars. Monthly, re-validate the theoretical rate and yield assumptions in the plan, because a 43 second demonstrated cycle planned as 45 hides 4.4 percent of capacity.

World-class throughput management shows up as boring consistency: hour-to-hour output varying less than 5 percent, realized rate above 85 percent of theoretical, first-pass yield above 99 percent on mature products, and a loss ledger in dollars that plant leadership reads weekly. Plants at that level typically find 10 to 20 percent more capacity inside existing assets within 18 months, which defers capital and shortens lead times at the same time. The habit that gets them there is small: never let an hour close without knowing why it missed.

Published 2026-07-02.