It was a Tuesday.
I remember it as a Tuesday because the customer's production line stopped on a Monday — and by Tuesday morning, I was already on a call that nobody in the room wanted to be on.
The part was a precision bracket. Nothing spectacular. Forty-seven grams of stamped and machined steel. Its job was to hold a sensor in exactly the right position inside a dashboard assembly. Without it, the dashboard couldn't be installed. Without the dashboard, the vehicle couldn't be completed. Without completed vehicles, the customer's plant — 1,400 workers, two shifts — had nothing to build.
The bracket cost less than a dollar to make. The production stoppage cost the customer approximately $180,000 a day.
We had missed the delivery by nine days.
The bracket cost less than a dollar to make. The production stoppage cost the customer approximately $180,000 a day. And the signal was readable three weeks before the line stopped.
In the review that followed — the kind of review that feels less like a meeting and more like a sentencing — the programme manager tried to explain what had happened. He was good at his job. Genuinely. He had managed forty-plus programmes in his career, delivered most of them on time, and knew the supply chain for this part better than anyone in the room.
But he had missed the signal.
The Signal Was Always There
The sub-supplier — a small press shop in a satellite town three hours away — had changed their shift schedule six weeks earlier. The change reduced their effective output by about thirty percent. Nobody had told us. The sub-supplier hadn't thought to mention it because they believed they could catch up. They couldn't.
And by the time the delay appeared in our project tracking system, the window to act had already closed.
The signal had been there. The data existed. The press shop's delivery performance over the previous four weeks had shown a clear pattern of slippage — each time by a few days, each time with a confident revised date that was never met. If anyone had been watching the right number, they would have seen it three weeks before the line stopped.
Nobody was watching the right number.
In over 200 manufacturing programmes investigated, every single failure had a visible signal at least three weeks before it became irreversible. The signal was not hidden. It was simply not being read.
Why Manufacturing Projects Really Fail
We blame the wrong things. We blame the supplier who was late. We blame the PM who didn't escalate. We blame the system that wasn't updated. All of these are symptoms.
The real reason manufacturing projects fail is this: the system was never designed to surface the signal before it became a crisis.
Most project tracking tools record what happened. They do not interpret what is happening. They are calendars with memory — useful for reporting to the past, useless for seeing into the near future.
When a PM updates a task date in a spreadsheet, the spreadsheet stores the new date. It does not notice this is the fourth time that date has moved. It does not calculate that the cumulative drift is now nine days. It does not weight the danger by the fact that the task is sub-supplier controlled. It does not generate a risk number. It does not surface this task to the top of anyone's attention.
It stores a date.
The Three Components of a Real Risk Signal
Every meaningful risk signal in a manufacturing programme has three components — and it is the combination of all three that makes it readable.
1. Frequency — how many times has this task slipped?
A task that has never slipped and slips once is different from a task that has slipped four times. The fourth slip is not just another event. It is the fourth data point in a pattern that tells you the original plan was never achievable.
2. Magnitude — how much has the date moved, cumulatively?
Individual slips of two days look manageable. Nine cumulative days of drift tells you that something structural is wrong — not a weather delay or a one-time event, but a fundamental gap between what was planned and what is actually possible.
3. Control type — who owns this task?
If it is internal, you have options. If it is a supplier, you have pressure but limited control. If it is a sub-supplier, you have almost no direct lever. The same slippage pattern means completely different things depending on who is responsible.
It is the combination of frequency, magnitude, and control type — read together — that transforms individual data points into a genuine risk signal.
The Three-Week Window
In almost every manufacturing programme failure I have investigated, there was a point — approximately three weeks before the delivery failed — at which the situation was still recoverable.
Three weeks is the recovery window. Resources could have been redirected. Alternative suppliers could have been activated. Customer expectations could have been managed proactively. The cascade could have been stopped.
After that window closed, options narrowed rapidly.
The cruelest thing about the three-week window is that it closes without announcement. There is no notification that says: you now have three weeks to act. The window simply closes — and by the time you realise it has closed, you are already in the crisis.
This is why the signal matters so much. The signal, read at the right time, is the notification that the window is open.
A signal read three weeks early costs you a few hours of investigation and one difficult conversation with a supplier. A signal read too late costs you the delivery, the customer's trust, emergency airfreight, and — in the worst cases — the customer entirely.
What Changes When the System Is Designed to Warn You
When the signal finds you — consistently, reliably, based on data rather than opinion — something changes in the culture of the programme.
The PM stops managing perceptions. Because the system shows the truth regardless of what the PM reports, there is no benefit to softening the picture. The data is the data. The only question is what to do about it.
The review meeting stops being a performance. Instead of hearing that everything is fine, the meeting surfaces what is actually at risk — early enough to fix it. The discomfort is productive. It produces action. It changes outcomes.
That is what a purpose-built manufacturing project execution system does. Not just track tasks. Surface the signal. Before the window closes. Before the customer calls.
Your Projects Don't Have to Fail
This article is based on Book 1 of The Execution Series — five books on manufacturing programme management written from 25 years of experience. All five books are free. Request your copy below.
Request the Free Series →See the signal surface automatically.
Project Perfect is the execution system that reads the pattern, calculates the risk, and brings the dangerous tasks to the top — before the three-week window closes. Built exclusively for manufacturing.