We Did the Work Right. But With the Wrong Tool on the Wrong Problem

Half of all the Extruded Profiles went directly to the scrap bin

We ran five weatherstrip profiles on an extrusion line. Each one had its own compression force target and its own tolerance window. We had been running them long enough to think we understood the process. But the scrap kept coming. The hit rate stayed poor. We were burning through material and time, and nobody had a clear answer for why.

So the group somehow decided we should build a FMEA to solve our problem.

We got a good cross section of people together. We mapped every failure mode across all five profiles. We scored severity, occurrence and detection. We picked it apart, debated it, and kept going until we felt we had it right.

Three months later we still had a scrap problem.

So we changed approach.

Someone said put the FMEA down and find the worst profile. Run a Pareto, focus the effort, stop the bleeding.

Profile 2 was the biggest offender. We adjusted the die. We worked the cooling. We stayed on it until the numbers seemed to improve.

Profile 2 improved.

Then Profile 4 became the problem.

We fixed Profile 4.

Then Profile 5.

We were working hard and carefully but scrap rate was still unacceptable …a different profile, every month.

The problem was never Profile 2.

It was never any individual profile.

Our process was not capable of holding compression force targets consistently across five different profiles under normal production conditions.

The compound durometer was shifting batch to batch. Melt temperature was inconsistent across changeovers. The process settings were not robust enough when moving from one profile to another. And the compression force measurement process had never been properly validated across five different geometries.

We had a process problem. We were chasing it one profile at a time. Every time we closed the gap on one profile the process found another one to fail on. We were solving the output while the process behind it stayed broken.

So what was the right tool?

It was not PDCA. PDCA works well when you already have a good idea of the root cause and need to test a fix quickly. You plan a change, try it, check the result and act on what you learn. It is a genuinely useful tool. It just was not built for this situation.

Our root cause was genuinely unknown. There were multiple variables moving at the same time — incoming compound consistency, melt temperature stability, changeover robustness, and a measurement process that had never been validated across five different profile geometries. We did not know which of those was driving the variation. We were not even sure it was just one of them.

That called for DMAIC.

You define the problem clearly. You measure what is actually happening across the process. You analyze where the variation is really coming from. You improve the right thing once you know what it is. Then you control it so it stays fixed.

DMAIC is built for exactly this kind of problem. Something complex and chronic with multiple unknown variables and a process that needs proper measurement and analysis before you can confidently improve anything.

We had that tool available yet the group built an FMEA instead, someone with enough authority had mislead the team.

Two mistakes running at the same time.

The first was the wrong tool. A FMEA is built to prevent problems before they happen. It looks forward and asks what could go wrong. We already had a problem running on the floor every shift and we did not know why. DMAIC was the right answer. How could a group of well-trained individuals have missed this?

The second was the wrong problem. When we finally moved fast, we focused on the product. The biggest number on the Pareto selected our worst product. The process just continued to crank out the scrap.

We did good work both times. Wrong tool. Wrong problem. Both times.

The question we should have asked on day one.

Is our process actually capable of holding compression force within tolerance across five different profiles under normal production conditions?

Not which profile is worst. Not which failure mode scores highest on a risk matrix. Just — is the process capable?

For us the answer was no. And until we asked that question nothing else we did was going to matter.

A DMAIC project aimed at that single question would have moved us further in two weeks than three months of FMEA and profile chasing ever did.

This touches every part of the business.

It was not just a quality problem or an engineering problem.

Production was chasing yield numbers the process could not reliably deliver.

Engineering was fixing profiles and the process continued failing in the following months.

R&D was developing compounds the process was not stable enough to take advantage of.

Sales was making commitments the operation genuinely could not keep.

Everyone was working hard. Nobody had stopped to ask whether the process was actually capable of doing what we were asking it to do.

Peter Drucker said it decades ago. Efficiency is doing things right. Effectiveness is doing the right things.

We were efficient. We worked carefully and thoroughly and with good intentions.

But we were not effective. We had the wrong tool pointed at the wrong problem for a long time before we understood what was actually going on.

The problem was never the profile.

It was always the process behind it.

Are you doing things right — or are you doing the right things?