Not long ago I sat in a meeting about a meeting about an escape. Nine people, ninety minutes, one projector, and not one of us within sight of the line shipping suspect parts. We produced an action plan, a RACI chart and a follow-up slot. The line changed shifts twice. I have been in quality long enough to know which of those artefacts mattered, and it was not the calendar invitation.

Three headlines, one disease

This week SME launched its Manufacturing Leadership Academy. In the same seven days, Wisconsin committed $6.6M to training thousands of workers in AI and advanced manufacturing, and a columnist at The Manufacturer argued that the best manufacturing leaders do not spend their lives in meetings. Read together, the three stories describe one condition. Academies, grants and columns treat a symptom – a capability gap you can close with curricula. The disease is that leadership has left the floor. Somewhere in the last decade the meeting room became the default workplace of the manufacturing leader, and no framework installed from a slide deck reverses that. Frameworks do not stand anywhere. People do.

What an academy can install – and what it cannot

I built a QA/QC department for a 900-plus-employee greenfield plant from zero: empty offices, no procedures, no habits, a training matrix I wrote myself, role by role. That experience taught me exactly where a course stops working. A curriculum installs codified knowledge, and it installs it fast. Six months in, every inspector could read a control plan, run an MSA, structure an 8D and recite the IATF 16949 clauses governing their station. That half of capability is purchasable, and an academy – or a state grant – buys it well.

What no course installed was the reflex to stop a line. The instinct to walk to the part before opening the laptop. The willingness to tell a customer we shipped it, here is the containment, here is the clock without checking with legal first. Those behaviours arrived one way: new engineers standing beside a leader who modelled them while a real defect sat on the table. That department went on to cut defect costs by 70% and hold customer satisfaction at 98%, and the ratio behind those numbers tilted hard toward floor hours over classroom hours. If your training budget buys only curricula, you are buying half the capability.

The floor is the only classroom with consequences

Two decades in automotive and aerospace quality taught me this about escalations: the critical ones are won at 06:30 in the containment room, not in the boardroom. QRQC works because the part is in the room, the operator is in the room, the data is raw and the clock is visible. An hour later there is a containment decision, a root-cause direction and an owner. I have sat through eight-hour alignment meetings that produced less. The difference is consequences. The floor charges tuition – in scrap, in €, in the customer's next audit. The meeting room charges nothing, which is exactly why it is so well attended.

A standard is not taught. It is witnessed.

Leading quality initiatives across a 2,000-plus workforce on multiple sites, I could predict a site's performance from one habit: whether the plant manager walked the line at shift start or read about it afterwards. Same standards, same tools, same training matrix. Different feet.

The AI wave sharpens the point rather than softening it. Siemens and NVIDIA are pushing AI-driven PCB inspection from the digital twin onto the factory floor, P&G is scaling AI-based inspection across its operations, and Wisconsin's money leans the same way. I design and deploy AI systems myself. They are superb at flagging an anomaly at 02:00. They cannot walk to the machine, weigh a borderline part and decide. Faster detection raises the price of absent leadership: when the system flags an escape in seconds, we did not know dies as a defence, and all that remains is who showed up.

How to hire for scar tissue instead of polish

If the floor is the classroom, hire people who have already paid tuition. Interview polish is a learned skill; scar tissue is not. The guidance I give recruiters and quality directors:

  • Ask for the worst escape they owned, in the first person. You want a cost figure, a containment decision and a structural change. "We" throughout, with no price attached, is a slide – not a failure.
  • Ask when they last stopped a line and what it cost. Good leaders know both numbers without notes.
  • Run the walk test. Put a defective part on the table and watch whether the first questions concern the part or the paperwork.
  • Read the CV for launches, greenfield builds, containment rooms and de-escalations. Certifications are the baseline, not the signal – I hold a shelf of them myself, and I would trade none of the floor hours that came after.

A candidate with fifteen years in the trade who has never had a customer's quality engineer camp in their plant has been standing far from the failure. That tells you where they will stand for you.

Key takeaways

  • Buy floor hours, not just curricula. Pair every classroom day with a structured assignment on a live defect, and budget backfill so learning does not cannibalise the line.
  • Track time-to-floor after an escape the way you track scrap. Presence is a metric, not a mood.
  • Let grants buy what they buy well – standards, tools, APQP and PFMEA mechanics – and leave judgement formation to the floor, under a leader who models it.
  • Hire for named failures with price tags. A candidate who cannot cost their worst escape has not had one.

Spend the grant. Run the academy. Certify the matrix – the codified half of this craft is real and worth every euro. Then put your hours where your escapes are. The plant learns the standard from where you stand, not from what you signed off, and no curriculum yet written can stand at a line at 06:00 in your place.