I can almost hear the chopper starting. But what if Butch had taken a wrong turn and pulled into our factory?
Our capacity slides said stainless had 48% free capacity. Anybody reading that number in an office would have scheduled more work into it.
The floor was not idle in the afternoons. Customer demand ran about 120 pieces a day. We were making 180 to 200. Roughly 80 extra pieces went into stock for monthly consignment replenishment at two customer locations, 500 to 1,000 pieces at a time.
The capacity number was correct within the way we had calculated it. It described a plant that did not exist.
A system could have displayed it faster. First, we needed to ask why production and customer demand were so far apart.
02 / The production record
The answer was in the paperwork
When I needed to know what customers actually bought, I pulled 16 random days of shipping paperwork spread over six months and counted the pieces by hand.
The count showed 45 to 60 stainless pieces a day and about 75 galvanized. Roughly 120 pieces of daily demand, against about $600,000 a month in sales. I did not see a seasonal pattern in those sampled days.
That hand count became the basis for our takt calculation, the headcount at each station, and the distances in the 9,000 square foot layout.
The data was in the building the whole time. Nobody had decided which question the record was supposed to answer.
03 / What we ask of MES
What exactly died?
In a recent LinkedIn post, Stephen Tagg writes, “MES is dead.” He means the reputation left by long, costly implementations and systems that are difficult to change. He explicitly says the technology itself remains useful.
A factory still needs to know what was released, what was made, what was held, and what shipped. ISA 95 gives those operations and their connection to business systems a common framework.
Our problem was more basic. “Capacity” told us what the equipment could produce under our assumptions. “Demand” told us what customers bought. We were using the first number to make decisions that required the second.
A different name for MES would not have caught that. A better screen might have helped, if both numbers were on it and someone owned the decision between them.
04 / Owning the software
Buy it, build it, or both?
Kunal Wagh asks buyers to look past the demo at exceptions, older equipment and integration with existing systems. In a second post, he asks who will test, maintain and support the software over the years.
Those questions belong together.
I would buy a system for work it already does well, provided it works during a real shift and our team can support it. I would build a focused application where our process needs something different and we are prepared to maintain it. I might do both.
AI makes that conversation more interesting. A team can explore an idea by building a first version. But the first version still needs clear definitions. What counts as demand? Does a shipment replenish consignment stock or fulfil a customer purchase? Who corrects the record when those events are confused?
Those answers determine whether the application helps us release the right work.
05 / Take it to the floor
One test for Monday
Before committing to a system, I would run this test:
- Choose one product family. Put customer demand, production and consignment withdrawals in one view.
- Agree what each number means. Name the person who corrects the record when it is wrong.
- Use that view at the next release decision. Did it change what we chose to make?
If it did, we have learned what the software needs to preserve. If it did not, adding more screens will not make the decision clearer.
Whether MES survives as a name is a vendor’s problem. Mine is whether the record tells me enough to change a release decision on Tuesday morning.
Which status in your system do you already know not to trust, and what do you check instead?
I can picture Butch parked beside the loading door, looking at the MES screen. Fabienne gets off the chopper and looks at the stacks of sheet metal.
I imagine she would have a question.
“If they only need 120, who are the other 80 for?”
I would want an answer before we started the next batch.
The floor tells the truth.
MarekContinue the conversation
Which status do you check on the floor?
Tell me what the system says, what you check instead, and which decision depends on it.
inFind me on LinkedIn