The Free Half of the Labor Case

The reflex that buries free money

In a working session on a labor-optimization program for a prepared-foods plant, the person carrying the pitch said the thing every operator says. The manufacturing audience is going to read this as a CapEx move, so we need to show a seven-year ROI. That is a sound instinct when you are buying equipment. It is the wrong frame for most of what a labor assessment finds, and the reflex quietly buries money the plant could take this quarter.

Here is the split that gets lost. A labor assessment on a 10-person line, each person around $50,000, does not return one number. It returns two kinds of savings. Some of it is a handful of people who simply should not be in those spots: the right number of people in the wrong configuration, a staffing and flow problem you fix by moving work, not by buying anything. The rest is capital-driven: change a station, add a piece of equipment, and then the role goes away, but only after you spend. Same program, two very different decisions.

Two layers, one hurdle rate

The failure mode is treating both layers as a single investment. Once the whole plan wears a CapEx label, the free layer inherits the capital layer's hurdle rate and its approval queue. A reduction that needs no money, no lead time, and no risk gets parked behind the same seven-year model as the machine purchase. The plant keeps paying for three roles it could have shed on operational evidence alone, because the paperwork treats them as an investment they never were.

This is what capital confidence actually means on a floor. It is not the courage to sign the capex. It is knowing, before you build the case, which dollars require capital and which do not. When the two are fused, you lose twice. The non-capital savings slow to the speed of a capital committee, and the capital savings get muddied by free wins that make the ROI look better than the equipment alone will deliver, which is exactly the number a gain-share or a finance review will later dispute.

The tell is in how the buyer reacts. Frame a mixed plan as one CapEx decision and a manufacturing leader does the rational thing: applies the CapEx test to all of it. Now a payroll cut you could have booked in a month is competing for capital against a new line, and losing, because it was never supposed to be in that queue.

Split it at the assessment

The fix is upstream, in how the assessment is packaged, not in how the case is pitched. Label every identified reduction at the point you find it: non-capital or capital. Non-capital reductions get decided on operational evidence and move now. Capital reductions get their own case, their own ROI, and, before any spend, a bounded validation trial.

That trial step is the discipline that separates confidence from hope. On one plant program, the team laid out two labor paths and, rather than committing to the capital one on a model, drafted a validation charter: a defined trial to prove the change holds on the actual line before the spend is authorized. That is the move. You do not defend a capital labor cut with a spreadsheet; you defend it with a scoped trial that either confirms the number or kills it cheaply. The non-capital cut, meanwhile, was already in motion, because nothing about it needed the trial.

Run the diagnostic on your own floor this week. Pull the last labor or throughput improvement someone pitched you. Draw a line down the middle: dollars that require a purchase order, dollars that do not. If you cannot sort them without reopening the whole plan, the plan fused two decisions that should have stayed apart, and you are almost certainly sitting on free savings that got frozen by association.

What the well-run version reads

Non-capital headcount reductions clear a decision inside 30 days on staffing and flow evidence, with no capex line attached. Capital-driven reductions carry a separate ROI and run a scoped validation trial with a defined window and a pass-fail number before any purchase order is cut. The two savings streams are tracked on separate lines so neither inflates the other, and the person who owns the labor program can tell you, for any given dollar of projected savings, whether it needs capital or not without opening a model.

The seven-year ROI was not wrong. It was answering a question only half the plan was asking. The other half was free, and free money does not need a model. It needs a decision.