Every moulder I know has had this conversation. A part is failing at launch, everybody is standing around the press, and somebody says, “Didn’t someone flag this?” Where is the mould decision record.
Usually someone did. The harder question is what happened to the warning after it was raised.
I lived this one on a defroster grille program. I have written about it before as an $800,000 lesson in listening. This time I want to look at it from a different angle, because I think listening was only half of the failure. The other half was that the decision to set the warning aside never became something anyone could pick up later and read.
The warning was in the room
The FEA engineer on that program had been doing this work for fifteen years. His analysis flagged high stress concentrations at the mounting points and warpage, and the mould flow simulation showed choked flow and uneven cooling. His recommendations were specific: redesign the mounting interface, add nozzles to balance the fill, add material in certain spots, reconsider the attachment method, and improve the cooling.
The toolmaker quoted the changes at $50,000 and eight weeks. The nozzles alone were $10,000 each. We were already pushing to hit the customer’s launch date, so the team decided there were no funds available and moved forward with the original design.
Then the toolmaker raised concerns about surface finish on the mounting features. Those got discussed too, and talked down. The part would be textured. The customer would accept the raw, untextured sample parts, and once the mould was grained the blemish would be hidden.
Here is the line from my original post that I keep coming back to. Engineers raised concerns, meetings acknowledged them, and then decisions got made as if those concerns had never been voiced.
Every problem we hit at launch had been predicted. Parts cracking under the stress of installation. Parts warping as they cooled. Marks showing through the texture instead of hiding under it. The customer rejected the parts at approval sampling. We ran at 25 percent scrap for three months, tried Velcro at $4 a vehicle, and finally built an $800,000 replacement mould that included every change the FEA engineer had asked for in the first place and then some.
Where a warning goes to die
On a moulding program, information does not sit still. It moves through a chain of handoffs, and every handoff is a place where it can drop.
The studio hands the shape to design. Design hands the part to the toolmaker. The toolmaker hands the mould to the floor. Each group has its own meetings, its own files and its own people. The studio handoff on the grille was restrictive. The message was essentially: meet these surfaces, with a maximum deviation of 1.0 mm, allowed only in non-aesthetic areas and pending approval.
A warning raised in a design review lives in the memory of the people who were in that room. When the drawing goes to the toolmaker, the warning does not go with it. When the mould goes to the floor, nobody at the press knows the cooling was a known compromise. When warpage shows up at first shots, the setup guy starts chasing process settings for a problem that was decided months earlier in an office he never sat in.
That is how you end up spending three months on band-aid fixes. Nobody can see that the root cause is a decision, not a defect.
It is rarely a knowledge problem
I say this a lot, and the grille program is the clearest example I have. We did not lack knowledge. The FEA engineer knew. The mould flow knew. The toolmaker knew. The knowledge was in the building.
What was missing was a cheap, reliable way to carry it forward. Writing a decision down properly takes effort, and on a program running behind, effort is the first thing that gets cut. That is an overhead problem, not a knowledge problem. The fix is not to hire smarter people. It is to make capturing the decision so simple that it gets done even when everyone is busy.
What a decision record looks like
I am not talking about a 40-page report. I am talking about one page or less, filled in at the moment a significant warning gets accepted, deferred or rejected. Something like this:
– The warning, in one sentence, and who raised it.
– What it would cost to act on it now, in dollars and weeks.
– What we decided, and who made the call.
– What we expect to see if the warning turns out to be right.
– Who needs to know downstream: toolmaker, process, setup, quality, customer.
– When we will look at it again.
The fourth line is the one that earns its keep. On the grille program it would have read something like “if FEA is right, expect cracking at the mounting points and warpage at first shots.” Anyone looking at the first parts off the press would have known within a day that we were looking at a design decision, not a processing problem.
The record also changes the decision itself. It is easy to say “there’s no budget” in a meeting. It is harder to sign your name next to “$50,000 and eight weeks” on one side and “expected failure: parts crack at installation” on the other. Writing it down makes the trade visible to everyone, including the person making it.
Making it survive the handoffs
A record only helps if it travels. A few habits make that happen.
Keep the decision records with the job, not in someone’s inbox. When the drawing package goes to the toolmaker, the open decisions go with it. When the mould goes to the floor for trials, the setup lead and the process tech get the short list of known compromises before first shots, not after.
Read the open records at every gate. At the mould design review, at the tool trial and at approval sampling, pull out the open decisions and ask one question: is what we expected to see showing up?
Close them out. When a deferred warning proves harmless, say so and close it. When it proves right, the record becomes a lesson learned for the next program, already written. That is how a checklist grows from real experience instead of from a template.
You can start without software. A shared folder and a one-page form will do the job at first. If you want better tools later, fine, but every working automated system started as a working manual one.
Start with the next decision
You do not need to go back and document every program you have running. Start the next time someone in a design review says “I’m concerned about…” and the room decides to move on anyway. Write it down. Name who made the call. Write what you expect to see if the concern is right. Send it to the people downstream.
If we had done that on the grille program, I cannot say for certain we would have made a different call. But the call would have been made in the open, and we would not have spent three months trying to fix a mould that was doing exactly what the analysis said it would do.
On past programs we tracked all of this in meeting minutes and considered it done. We logged who said what, when and why. It turns out nobody reads meeting minutes in the rush of a busy day, or at the launch of a new program. That is why I now keep decisions, lessons learned and tooling insights in one central database, where they show up when the next program needs them.
I have spent 35 years in moulding and product development and hold a PMP. Most of what I do now is help moulders front-load the design stage and hold on to the knowledge already in their shop, as a project manager when you need one. If you have a program heading into tooling and a few concerns that got talked down along the way, I would be glad to talk it through. A free Diagnostic Map conversation is a good place to start, to find where information is dropping between your handoffs.