Take the Free Labor Savings First

Two buckets, one bad habit

Walking a sous-vide protein facility with an equipment integrator, the scoping conversation kept returning to one split. The line and labor review had found real savings, but they divided cleanly into two buckets that behave nothing alike. One bucket needs equipment: tasks that only come out of the crew if a machine takes them over, with quotes, integration, and lead times attached. The other bucket needs nothing. Stations were simply carrying more people than the measured work content of the line, plainly enough that the reviewer's honest reaction was that it was shocking.

The habit that costs plants money is bundling those two buckets into a single automation project. It feels natural, it is all "labor savings," and the combined number makes the capital case look stronger. But the moment the free bucket is stapled to the machine bucket, it inherits the machine's timeline: justification cycles, integrator selection, quotes, approval, and often a year or more of equipment lead time. Payroll that could have stopped this quarter keeps running to support a purchase decision it does not depend on.

Why the buckets get merged

The mechanism behind the bundling is organizational, not analytical. Crew sizes are inherited, last year's crewing plus adjustments, and nobody re-derives them from work content because that conversation is uncomfortable without an external trigger. An automation project is a socially acceptable trigger: nobody's judgment is questioned if a robot changes the math. So the no-capital reductions hide inside the capital case, where they double as ROI padding.

There is a second-order cost to that padding. When the free savings are counted in the machine's payback, the machine looks better than it is. Strip out the labor that would come out anyway and some automation cases stop clearing the hurdle rate at all. Which means the merged bucket does not just delay the free half; it can buy the wrong machine with it. The honest sequence is the reverse: take the no-capital savings first, let the line settle, then price the equipment against the labor that truly remains.

The measurement layer matters here too. On this operation, the performance numbers coming off the lines had been reading impossibly high before the sensors were reconfigured, the kind of miscalibration that makes crews look busier than they are. You cannot split the buckets on numbers like that. A week of getting the instrumentation honest preceded any staffing conclusion, because a staffing change justified by a broken number gets reversed the first time someone audits it.

Running the split

The application is a two-column exercise on the existing labor study. Column one: reductions achievable with current equipment, each with the operational change that enables it, a rebalanced station, a combined role across two lines, a schedule change that eliminates a partial crew. Column two: reductions that require the machine, priced into the capital case and only those. Column one gets a 90-day plan owned by operations. Column two goes to the integrator.

The plant in question had no preferred integrator and was open to recommendations, which is common and is one more reason not to chain the free bucket to the machine: integrator selection alone can take a quarter. The stage gate for the capital decision was directional numbers; the stage gate for the no-capital bucket was simply a decision to act on measured work content.

What a well-run split looks like

Crew size per line per shift traces to a measured work-content number that was validated after instrumentation was fixed, not before. The no-capital reductions are implemented or scheduled inside 90 days of the study. The automation justification contains only machine-dependent labor, and it still clears the hurdle, or it dies honestly. And a year from now, headcount per thousand units shows two distinct steps in the trend line: one when the crews were rightsized, one when the equipment landed.

The savings nobody has to buy are the ones most often left waiting. The machine was never the obstacle. The split was.