Cost Estimation

What OEE Losses Cost Per Unit and How to Quote Around Them

A money-first view of factory performance: how downtime, scrap, and speed loss inflate cost per unit, how to price against effective capacity, and where estimates quietly break.

Cost per unit is not driven by your nameplate rate; it is driven by your effective rate after losses. A cell that costs 480 dollars an hour to run, all in, and produces 600 good units an hour looks like 0.80 dollars per unit. Drop effective output to 450 good units an hour through downtime and scrap and the same 480 dollar hour now spreads across fewer units, pushing cost to 1.07 dollars, a 33 percent increase with no change in wages or material. The mistake that wrecks quotes is dividing fixed hourly cost by capable output instead of realistic output, so start every estimate from a demonstrated OEE, not a hopeful one.

Break the hourly burden into its parts so you can see which loss hits which bucket. On a typical machined part, material might run 2.10 dollars, direct labor 0.65 dollars at a loaded 42 dollar per hour operator across the cycle, machine time 0.55 dollars from a depreciation plus energy plus maintenance rate near 33 dollars an hour, and overhead applied at 0.90 dollars. That is 4.20 dollars of true cost. Only material scales cleanly with volume; labor and machine time are time based, so any minute lost to a stop is paid whether or not a unit comes out, which is exactly why availability loss is expensive.

Downtime cost is the clearest money lever and the easiest to underprice. Take a line whose fully loaded run cost, labor plus machine plus allocated overhead, is 600 dollars an hour and whose lost margin per idle hour reflects units not sold. The Downtime Cost Calculator separates the direct burn, roughly 10 dollars a minute here, from the contribution margin forgone. If each unit carries 3.50 dollars of margin and the line makes 600 an hour, an hour down forfeits 2,100 dollars of margin on top of 600 dollars of burned fixed cost, so a single hour is closer to 2,700 dollars than the 600 many estimators book.

Scrap is more expensive than its yield number suggests because you pay for a bad unit twice. At 3 percent scrap on a 4.20 dollar part, the naive hit is about 0.13 dollars per good unit. But a scrapped unit consumed the full material and machine time, and it displaced a good unit you could have sold, so the loaded cost of scrap includes forgone margin plus any rework labor. Fold scrap into cost per good unit, not cost per unit produced: divide total cost by good count. That single denominator change moves a 4.20 dollar part to about 4.33 dollars and keeps you from quoting a price that never covers real yield.

Speed loss is the quietest cost because nothing looks broken. If the machine can hold 1.0 second per unit but chronically runs 1.15 seconds, you have lost 13 percent of capacity while paying the same operator and the same machine rate. On a 42 dollar loaded labor hour plus 33 dollar machine hour, that 13 percent is roughly 9.75 dollars an hour of value you cannot recover, and across 4,000 hours a year that is 39,000 dollars per cell. Quote from the sustained real cycle time, and treat any planned improvement as future capacity, never as a discount you give away today.

Overhead allocation is where defensible quotes fall apart. Spreading a fixed overhead pool across an inflated production forecast makes every unit look cheap until volume misses and the pool has to be recovered over fewer units. If your annual overhead is 1.2 million dollars and you allocate against 1.5 million units but only make 1.2 million, your real overhead per unit jumps from 0.80 dollars to 1.00 dollars, a 25 percent miss baked into every quote. Allocate against a conservative demonstrated volume tied to your actual OEE, and revisit it quarterly as effective capacity changes.

Improvement projects need the same money discipline as the quote. The IoT ROI Calculator and Automation Payback Calculator both hinge on converting recovered availability or reduced scrap into dollars at your real margin. A sensor package that recovers 5 points of availability on a 600 dollar an hour line running 4,000 hours yields roughly 120,000 dollars of regained capacity value a year; against a 60,000 dollar install that is a six month payback. The error is crediting the project with revenue you have no demand to sell; only count recovered capacity you can actually fill with orders.

A defensible quote reads back cleanly to the buyer. State the effective rate and the OEE it assumes, list material, labor, machine time, scrap, and overhead as separate lines, and show cost per good unit rather than cost per unit produced. Add a contingency tied to the gap between your demonstrated OEE and the quoted assumption, commonly 3 to 8 percent, so a bad week does not erase margin. Quotes built this way survive a purchasing audit because every number traces to a rate, a time, or a yield the customer can check against your own floor data.

Published 2026-07-01.