Who Spends the Hour You Just Recovered

A rate gain that nobody had agreed to spend

A capacity project can deliver exactly the rate it promised and still return nothing. Not because the measurement was soft, and not because the constraint was in the wrong place. Because the hours it freed up had no claimant, and hours with no claimant get absorbed by the plant inside a week.

I ran into the cleanest version of this question last summer, working through how to position two different kinds of optimization work. The person I was building it with asked the obvious thing: do you tell a plant it keeps the same crew and gets more throughput, or do you tell it the crew gets smaller? I said the crew, and I said it fast. Then I spent the rest of the conversation working out why I had been so quick about it, because the answer was not that throughput matters less. It was that labor and throughput are not the same kind of result, and only one of them arrives already attached to money.

That is my filter now. When I have to choose which question to open with, I open with the labor question, because its conversion is already owned. And I do not take a capacity project seriously until somebody outside the plant has committed to spending the hours it recovers.

Labor arrives attached to a line item. Throughput does not.

A headcount reduction is countable the month it happens, by somebody who already owns the number. Department payroll exists in the budget, gets read monthly, and has a name against it. Nobody has to decide what to do with the savings, because the savings are the absence of a cost.

A throughput gain is the opposite shape. It produces capacity, and capacity is time at a resource. Time is not a line item. It becomes money only if one of three things happens, and all three are decisions somebody outside the plant has to make.

The commercial side sells the incremental volume, at a margin that still holds at the new mix. Volume currently running at a co-packer or a second site comes back in house, which requires a qualification and a transfer decision. Or the plant releases the resource the hours were sitting in: a position, a shift, paid overtime, a weekend.

Each of those has an owner. In a mid-market manufacturer, none of those owners is the person who approved the throughput project. That is the gap. Operations can deliver the rate, verify it honestly, and still produce nothing, because the conversion is somebody else's decision and nobody asked them in advance.

This is a different failure from putting the constraint in the wrong place. That one I have written about separately: where flows converge, delivered rate is set by the pattern of arrivals into the shared station rather than by any machine's rated speed, which is why you level the arrivals before you price the line. That is a diagnosis problem, and operations can solve it alone. The conversion gap cannot be solved alone. You can get the constraint exactly right, level the arrivals, prove the delivered rate against a real baseline, and still have no buyer for the hours.

When there is no buyer, the hours do not wait in a ledger. They get absorbed, in one of four ways, all of which look like success on a rate report. The line finishes the schedule early and the crew stands. The shift gets quieter instead of shorter. The former constraint idles while the next shared resource starts backing up, which moves the binding constraint without anyone having decided to move it. Or the plant runs the same hours and makes more units, which is inventory, not revenue, until somebody sells it. None of the four changes a cost per unit.

Make the conversion a precondition, not a conclusion

The fix is a sequencing discipline, and it costs nothing except the willingness to ask early.

Write the conversion path into the project when you scope it, before the study, not after the result. One path, named. Not capacity and labor and growth together, which is three paths and therefore none.

Name the owner outside operations who commits to that path, by name, and the month it lands. Commercial for a sales path. Supply chain, or the site that currently runs the volume, for an insourcing path. The plant manager and HR for a resource-release path.

State the single number that will move and its pre-project value, pulled from the system of record rather than estimated. Units shipped on that family. Outside spend on that SKU group. Department payroll. Overtime hours. One number, one source, one date.

Then pressure-test the path on its own terms. If it is sales, get the margin at the new mix, not the current mix, because a capacity gain usually shows up first on the product that was being rationed, and that is rarely the richest one. If it is insourcing, the qualification clock sets the conversion date, not the install date. If it is resource release, say so plainly, because what you have is a labor project with an equipment component, and it should be priced and governed as one.

The diagnostic that catches a missing path is a question asked in month-after form. Not what is the business case, which invites a model. In the month after this lands, what moves, who reports it, and against what number? A project with a real path answers in one sentence. A project without one answers with a range.

There is an honest version of the no-path answer, and it is sometimes the right call. Some capacity projects exist to remove a risk, or to hold option value against a demand case that has not closed yet. That is a legitimate reason to spend money. It is not a savings case, and it must not be carried inside a savings number. A proposal, an approval, an installation, a verified rate, and a converted result are five different states, and only the last one reaches the P&L.

What the well-run version reads

Every open capacity project shows one conversion path, one named owner outside operations, a commit month, and one number with its pre-project value. That number is read in the commit month from the system of record and recorded next to the rate result, so a project can show a verified rate gain and an unconverted result at the same time without either one hiding the other. Capacity recovered and capacity converted are reported as two separate numbers, and the gap between them has an owner. A project whose conversion has not landed within a quarter of the rate gain stays on the list as open.

Put a name next to each one

Take your current capacity project list and, for each line, write the name of the person outside operations who will spend the hours it recovers, and the month they will spend them. The ones you can fill in without calling a meeting are the real projects. For the rest, you now know what the next conversation is, and it is not with the integrator.