Half a Person in Three Places Is Not a Saving

The objection that saved the project

Halfway into a labor and lines review at a Midwest protein processor, the continuous improvement leader stopped the recommendation deck. "If we save half a person in three different areas across three different shifts, I don't know that I could actually tactically execute that." Then he sharpened it. "I can get all of this production done in six hours instead of eight by adding this automation. I'm not going to send my people home after six hours, and I can't add more capacity because I'm at a cook limit."

He was not resisting the work. He was refusing to book a number he could not collect. The deck in front of him was full of real, defensible hour reductions. Almost none of them named a position, a shift, or a date, and the cook step downstream capped anything the upstream lines might free up. What he asked for was the bread crumb trail from a theoretical two hours of daily productivity to money in the bank.

That objection was the most useful thing that happened in the engagement.

Why the fractions never add up

Labor is bought in whole people, on whole shifts, in specific areas. An improvement that frees four tenths of an operator in trim on first shift, three tenths in pack off on second, and half a person in sanitation on third has not freed a person. It has freed three fractions sitting in three different supervisors' crews, and no supervisor can act on a fraction. Each one is individually correct and collectively uncollectable.

The measurement system hides this. The improvement list is denominated in hours. The plant is denominated in conversion cost per pound. Nothing reconciles the two, so a project can report a strong hour reduction while the cost per pound the controller sees moves by zero. At the plant above, that gap has a number on it: conversion cost runs eight cents a pound heavier than at the sister plant, on a site that carries about one fifth of production pounds and roughly 47 percent of contribution.

Aggregate measurement makes it worse. On the same engagement the team stopped measuring pounds per labor hour in aggregate and moved to a segmented basis, steak days measured separately from barbacoa days. A mixed schedule averages away exactly the variation you staff against. The average day does not exist on the floor, and no crew is ever built for it.

There is a second reason fractions evaporate: the mix under them keeps moving. At that same plant, a brisket program starting the first week of October changes 100 percent of the trim labor and the trim process, an entire product family moves to the sister site, and the crew that ran it becomes a flex team that runs roll stock at the back of the building, packs off after high pressure processing, then comes back and racks. A saving pinned to a specific station in a specific product flow has a shelf life measured in weeks.

Two ways to convert hours into cash

There are only two honest conversion routes, and both of them end in a headcount document.

The first is level loading. Finishing two hours early is not a saving when nobody goes home and the downstream cook step will not accept more volume. The same work done by fewer people across the full eight hours is a saving. That was the reframe in the room: the question is not whether the line can be faster, it is whether the line is overstaffed for the eight hours the volume actually needs. Stated that way, the recommendation becomes "remove two positions, keep the eight hour shift, capture roughly $100,000," which a plant manager can execute and a controller can verify.

The second is a flex crew. Fractions become real when one pool absorbs them across lines instead of stranding them inside single areas. The plant above is already doing this informally: five or six people a day come off the trim line onto other lines, backfills are being held rather than filled, and an entire product crew is about to be repurposed as a flex team. Informal works until the person holding it in their head takes a week off. Doing it deliberately means the pool has a size, an owner, a written rule for who moves when, and a cross training matrix deep enough that two absences on the same day do not break it.

The rule worth adopting: do not accept a labor saving that cannot name the position it removes. Anything else is a hypothesis with a dollar sign attached, and hypotheses do not clear the P&L.

What a well run labor model looks like

Every approved improvement carries an area, a shift, a position and an effective date, and it appears on the staffing plan before it appears in a savings report. Conversion cost per pound is reported by day type, so a steak day and a barbacoa day are never averaged into a number that describes neither. The flex pool is a headcount with an owner, not a habit that lives in one supervisor's memory. Cross training depth is tracked per position, and the plant knows how many simultaneous absences it can absorb before throughput drops. When the product mix changes, the labor model is reissued in the same week, not the same quarter.

The unit you book in

Hours are the unit the analysis produces. Positions are the unit the plant pays. A project that never converts between them will generate a long, accurate, unbankable list, and it will do it while everyone in the room agrees the work is good.

The plant carrying eight cents a pound more conversion cost than its sister site did not get there by missing improvements. It got there by booking them in the wrong unit.