Labor Cost

Managing Labor Cost per Unit on the Shop Floor

Labor cost per unit decides whether a quote wins and whether a line earns its keep. Here is how to measure it honestly and manage it daily.

Labor typically runs 20 to 35 percent of cost of goods sold in discrete manufacturing, and it is the number a customer's buyer will attack first. A quote built on $1.60 per unit of labor that actually runs $2.10 gives away 31 percent of your labor line on every shipment. On a 500,000 unit annual contract that is a $250,000 hole nobody sees until year-end. The plants that win reprice work every quarter because they know their labor cost per unit to the penny, by product, by line, by shift.

The math is simple and most plants still get it wrong. Labor cost per unit equals crew size times loaded wage, divided by good units per hour. Four operators at a $32 loaded rate cost $128 per hour. At a 45 second cycle the line makes 80 units per hour, so labor is $1.60 per unit. Let output slip to 60 units per hour with the same crew and you are at $2.13, a 33 percent increase with zero change in payroll. The Labor Cost calculator runs this in seconds; the discipline is feeding it honest numbers.

Loaded wage is where the first lie usually lives. Take base pay and multiply by 1.25 to 1.45 to cover payroll taxes, benefits, insurance, and paid time off. A $22 base operator costs $28 to $32 loaded. Then decide what counts as crew: if a material handler feeds three lines, charge each line a third of that person. Plants that count only direct heads on the line understate labor cost per unit by 15 to 25 percent, and that error flows straight into bad quotes.

Benchmark against your own contribution structure, not industry folklore. In high-volume assembly, labor under 10 percent of selling price is strong; 10 to 18 percent is normal; above 20 percent means automation or rebalancing should be on the table. Track the ratio of actual to standard labor per unit as well. World-class plants run 95 to 105 percent of standard consistently. If you are running 120 percent of standard, the standard is fiction or the line is broken, and either one costs money.

The levers, in order of typical payback: line balance, crewing, cycle time, and cross-training. A balance study that moves 8 seconds of work off the bottleneck station lifts an 80 unit per hour line to 95, cutting labor per unit from $1.60 to $1.35 for free. Flexing crew to demand matters just as much: running 4 operators on a day that needs 60 units per hour costs $2.13 per unit when 3 operators could hit it at $1.60. Overtime at 1.5 times pay adds 50 percent to the marginal unit; use it for real demand, never to cover absenteeism.

Watch the classic failure modes. First, averaging labor across a mixed-model line: if product A takes 30 seconds and product B takes 90, a blended $1.80 average means you are overpricing A and losing money on every B. Second, ignoring the denominator: labor per unit must divide by good units, so a 5 percent scrap rate silently raises labor cost 5.3 percent. Third, treating indirect labor as free. Supervision, quality, and material handling often add 30 to 50 percent on top of direct labor, and quotes that skip it bleed.

Run it as a cadence. Daily: post crew size, output, and labor per unit on the line board and have the supervisor explain any day more than 10 percent over standard at the tier meeting. Weekly: review labor variance by product with the planner and rebalance crews for next week's mix. Monthly: reload actual cycle times, wage rates, and yields into your standards and recheck every active quote against them. Standards older than 12 months are almost always 10 to 20 percent stale.

World-class looks like this: labor per unit known by product and posted within 24 hours, actuals within 5 percent of standard, crews flexed to demand daily, and every quote rebuilt from current loaded rates. Plants that get there typically pull 10 to 15 percent out of labor cost per unit in the first year without cutting a single head, purely from balance, crewing discipline, and honest denominators. The ones that do not are the ones surprised at year-end.

Published 2026-07-02.