Here is an uncomfortable observation from two decades of plants: the sentence that does the most work in any consolidation press release is never the one about people or products. PMGC Holdings announced this week it has combined two aerospace manufacturing companies into a single California location, "enhancing manufacturing and operational efficiency and reducing costs." Fine words. What the sentence omits is what will define the next eighteen months: two AS9100 quality systems, two part-numbering schemes, two libraries of customer approvals, two sets of special-process accreditations — now sharing one roof and, eventually, one certificate. The org chart merges in a day. The quality systems collide for a year and a half, and the collision is where the money actually goes.
I have seen this from both directions. Leading quality across a multi-site, 2,000-plus workforce under AS9100 and IATF 16949 taught me what it costs to keep baselines aligned when they were never designed to be identical. Building a greenfield QA/QC department for 900+ people at SNOP from an empty office taught me the other lesson: standing a system up from nothing is the honest version of what a merger pretends to be — deciding, part number by part number, which of two living systems survives. At least in a greenfield nobody argues the old way still works. In a consolidation, everyone does.
Two certificates, one baseline
An AS9100 certificate has a scope, and scopes do not merge by press release. When two registered sites become one, the certification body has to be notified, the scope redefined, a transfer or re-registration audit scheduled. Until that audit closes, you are in a gap. Both legacy quality systems must stay fully alive, both internal audit programmes must keep running, both document-control loops must stay closed — while the organisation around them is already being told to "standardise".
Document control is where it gets ugly. Two revision conventions. Two change-management workflows. Two ways of numbering a traveller, a router, a nonconformance report. I have walked into integrations — FOREAST work with clients like ArcelorMittal included — where the merged site ran parallel document masters for eight months because nobody had budgeted the weeks of reconciliation. Nobody budgets it because it doesn't appear on a machine price list. It appears as an auditor finding twelve months later.
A merger combines square footage in a quarter and quality baselines in a couple of years — and only one of those appears in the press release.
The knowledge leaves before the machines do
Consolidation redundancies are never distributed evenly. They land hardest on quality engineering, inspection and planning — the functions leadership perceives as overhead. The result is predictable: the two people who knew why a key characteristic tolerance was tightened in 2019, which supplier lots historically drifted, and how the penetrant inspection line behaves in August humidity are gone before the first machine is disconnected.
Tribal knowledge does not survive an org-chart line item. Key characteristics, special-process subtleties, customer-specific quirks buried in old 8D records — this is the invisible inventory of a quality system, and it has no stock location. At SNOP I documented as I went because there was no predecessor to lose. In a merger you have two predecessors, both assumed to be captured in the QMS. They never fully are. Run structured knowledge-extraction sessions — interviews, shadowing, review of closed NCRs — before the redundancy notices, not after.
Moved is not qualified
Here is the part finance never sees coming: relocating a machine resets its capability. A five-axis mill that was holding Cpk 1.8 in Building A is, after rigging, transport and re-levelling in California, an unknown process until proven otherwise. That means first article inspections — real ones, full AS9102, not a rubber stamp — on production parts approved years ago under the old process signature. Customers who approved parts from Site A do not automatically approve them from Site B with the same machines. Re-approval packages, PPAP-equivalent evidence in aerospace terms, go out the door one by one.
Special processes compound this. NDT, heat treat, chemical processing, welding — accreditations such as Nadcap are tied to processes and locations, and a moved process is a changed process. The calibration programme merges too, and the two legacy programmes will disagree about intervals, methods and masters. The merged site needs one calibration backbone, reconciled before the first external auditor asks to trace a gauge. Then there is the current wave of computer-vision inspection tools — Tristar AI just raised $12.4 million to automate exactly this space — and the temptation to bolt new technology onto an unmerged baseline. Don't. Validate the baseline first; automate second. An unqualified process inspected by AI is still an unqualified process, just detected faster.
Key takeaways
- Map both QMSs — document control, part numbering, approvals, calibration — before move-in day, not during the first joint audit.
- Extract tribal knowledge on key characteristics and special processes before redundancy notices go out; it has no second copy.
- Budget FAI, customer re-approval and requalification as capex, not overhead — that is what they are: the price of the moved capability.
- Keep both legacy systems fully audit-alive until the merged scope certificate lands; the registration gap is a risk window, not a holiday.
The press release claims the savings on day one. The savings become real the day the merged site passes its first external audit as one entity and ships its first escape-free quarter afterwards. Everything between those two dates is the quality function quietly earning back a promise somebody else made. Having stood up a system from zero and integrated systems that already existed, I can tell you which is harder: the merger, every time. A greenfield has no ghost arguing that the old revision was fine. Consolidation done well is eighteen months of disciplined, unglamorous reconciliation. Done badly, it is two failures sharing one address, at the cost of a certificate and a customer base that took decades to earn.