Missed Work Has No Room In The Live Week

The rule a scheduler wrote for herself

I sat in on an interview with the production scheduler at a prepared foods plant last week. She described the change that took her from firefighting to something close to control, and it was not a system, a dashboard, or a new planning module. It was a rule she now applies without asking: whatever was planned last week and did not get made does not go into this week. It goes at least two weeks out.

Her reasoning was plain. If you drop last week's miss into the live week, you have to take something out to make room. That something becomes next week's miss. She used the word domino, and she was right: the first shortfall never gets paid for, it just gets passed along, and by the fourth week nobody can tell whether the plan is late or the plant is slow.

The second rule was the same shape. She will not schedule a job whose ingredients are not in the building. Not ordered, not confirmed, not on a truck. In house. A month earlier the plant was waiting midweek on film for a Wednesday run, and every job sequenced behind it was holding a slot against material that might not arrive.

Both rules do the same thing. They stop the schedule from making promises the week cannot keep.

A week is a commitment, not a container

The reason this works is that a production week is not empty space waiting to be filled. By the time it starts, it has already been spent. Crews are staffed to the run lengths in the plan. Sequence is fixed by the constraints that cannot be negotiated away: allergen and test requirements that stop two items sharing a line, a multi day hold between an upstream step and packing that no amount of urgency compresses, a small number of lines that can actually run the format. The schedule is the ledger of those commitments.

So when a late change arrives, usually from the commercial side, and asks to swap the item running today for a different one, the edit looks free. In the planning system it is two clicks. On the floor it is not an edit, it is a re-pricing of the week.

Here is what actually moves. The crew was built for the original run length, so a shorter run means people were called in against volume that is no longer there and the line finishes early with the shift already paid. The item that was displaced still has to be produced, so it returns to a week that was already full, which means the next miss is now scheduled rather than random. If the swapped item needs a hold step, it cannot simply be pulled forward, because the hold is calendar time, not queue time. And the workaround that gets reached for under pressure, forcing a faster path to compress the hold, consumes capacity on a shared resource that other products depend on.

None of that shows up as a schedule variance. It shows up as overtime, as attainment, as a line that ran inefficiently, and as a commercial team that has learned lead times are negotiable because they have been absorbed before.

Stabilize before you automate

The same plant had just automated part of this. The weekly job list used to be pulled out of the ERP by hand, costing hours of someone's week, and now it comes out automatically. That is real value and it was worth doing. It is also worth being precise about what it bought: a faster path to the same plan. It did not make the plan hold.

This is the decision I want operators to be clear on, because it comes up as a tooling request almost every time. When a schedule is unstable, the instinct is to automate the scheduling. Faster replanning, better visibility, a tool that reshuffles the week when something slips. What that buys is the ability to produce a new plan more quickly, which is only useful if the old plan was failing because it was slow to produce. It almost never is. It is failing because the week is absorbing variance it was never staffed to carry, and a faster replan just lets you absorb it more often.

My rule is simple. Stabilize the week before you automate the plan. Put the two constraints in first, that missed work goes out and unmaterialized jobs do not get scheduled, and hold them for a month. Then automate, because now you are automating a plan that means something.

The stabilizing move is not a system change at all. It is deciding who is allowed to change the live week, and requiring that anyone who does restates the crew plan at the same time. A change that has to name its own labor consequence gets made far less often, and the ones that survive are the ones that were genuinely worth the disruption.

What a stable week looks like

Attainment is measured against the schedule as published before the week began, not against the version edited on Thursday. Misses from the prior week appear in a future week, not in the current one, and you can point to where each one went. Every job on the current week has its materials physically received. In week changes exist, because customers are real, but each has a named approver and an accompanying crew adjustment, and you can count them: a stable week has a handful, not a daily stream. Run lengths in the plan match the crews that were called in for them. And the commercial team quotes the lead time the process actually needs, including hold time, rather than the lead time a customer asked for.

The test

Pull your last four weeks. For each item that did not get made in its planned week, write down which week it was produced in. If most of them moved by one week, you are not recovering, you are carrying the shortfall forward and paying for it in shift labor the whole way. Fix where the miss lands before you buy anything that promises to reschedule faster.