The Ceiling Is Utilization, Not Rate

Four days with a stopwatch, and two numbers that did not match

Our field lead spent four days on the trim line at a Midwest protein processor with a stopwatch and a clipboard. About 100 cuts observed. Average cut around 20 pounds. Average cycle time just under 53 seconds. That works out to roughly 1,300 pounds per hour per trimmer while they are actively trimming.

Then he asked the plant for its own data. Not an estimate, the trim time sheet the plant already keeps, showing when trim starts for a cut, when it ends, and what came off. He filtered it to flat cuts only, assumed 21 people on the line, and asked what the all in rate came to.

377 pounds per hour.

Same line, same people, same week. One number says 1,300 and the other says 377, and both are correct. He re-ran it at 15 pound cuts and at 10 pound cuts to be fair to the mix argument. The gap held every time.

The second number has no owner

Every plant knows its rate. Somebody has timed the task, and that number lives in a standard, a line sheet, or a supervisor's head. Almost no plant knows its utilization, meaning the fraction of the paid hour that the rate is actually being delivered.

That is not an accident. The measurement systems on the floor report presence and quality, not time at task. On another line at that same processor, the roll stock machines were reporting essentially 100 percent quality, which the team's own words called out as making no sense. Somebody swapped that line off the add on loggers and read the PLC directly, and only then did a true OEE appear. The instrumentation had been reporting a number that felt fine and was not real.

The same thing happens with labor, just without the sensor. Staffing gets set from the observation that people are present and busy looking. The trim supervisor at this plant staffs 27 to hold 23 on at all times, and the crew we watched ran closer to 20. Nobody in that chain is lying or lazy. They are managing to the only variable they have been given, which is bodies on the line. Time at task is the variable that actually governs output, and it is nobody's number.

So the gap gets absorbed. When output falls short, you add a person, because a person is the lever you control today. That decision is cheap once and expensive forever, because it moves the labor baseline and the baseline is what next year's budget starts from. Meanwhile the reason the gap exists, in this case people waiting on supply and stepping away to sharpen knives, never gets attacked, because it was never named.

There is a human tell for this. The plant leader who owns that line spent close to a year insisting there was no opportunity, that his operation was as efficient as it could be, and escalated the point above his own head more than once. Two months into the work, with about seven positions already out, the position had shifted to yes, we had planned for that. That reversal is not cynicism. It is what happens when the only number a leader has to defend himself with is headcount. Give him the utilization number and he has something better to defend.

Take one person off and find the break-even

The instinct at 377 against 1,300 is to announce a target and cut to it. Do not. The gap is a measurement, not a headcount plan. Part of that gap is real work that has to happen somewhere.

The sequence that works is smaller and slower and it survives contact with the floor.

Take one person off the line. One. Then define, before the shift starts, what the metrics have to read for the supervisor to say this is fine. Daily production complete. Labor hours not up, because moving a person into overtime elsewhere is not a saving. Work in process still sufficient for the downstream operation, because starving the next step just relocates the problem. When those hold for a run of shifts, take the next person and repeat.

That is how this plant is moving from its own stated baseline of 23 toward 20, one step at a time, with the supervisor holding the pen on the break-even each time.

While you are at it, look at the positions that do not appear in any standard. In the ready to eat room at this same site, a person was stationed just past the X ray machine on one visit, moving bags into bins. On the next visit the position was empty and nothing downstream had missed. The role that covers it is two and a half loaders, two full time and one splitting time between loading and rework, and the rework half is not really half. Get the baseline instead of the opinion. Two people unload a rack in six minutes, one person does it in twelve, and against the actual production schedule that is about six hours of straight loading time. Now the staffing question has an answer instead of an argument.

What a well-run line reads

Two rates are posted for every direct line, the rate at the task and the rate all in, and the ratio between them is on the weekly review with a name next to it. When the ratio moves the wrong way, the first question is what is consuming the hour, not who is missing from the line. Uninstrumented positions do not exist; every station on the floor traces to a standard with a time behind it. Automated data that reports a perfect number is treated as broken until someone proves it from the controller. Staffing changes move one position at a time against a written break-even that the area supervisor signed before the shift, not a target announced in a conference room.

The plant did not have a labor problem. It had a measurement problem that could only express itself as a labor problem, so that is the form it took, for about a year, at full cost.