Field Notes: Engineering QA/QC and the Audit Trail
A senior engineer once told me the most dangerous phrase on any project is we checked that. Not because the check did not happen — but because we checked that is a memory, and a memory cannot be handed to an auditor, a regulator, or the engineer who inherits the project two years from now.
Here is the thing about engineering QA/QC that nobody says out loud: an undocumented quality check and no quality check at all are, at the moment it matters most, the same thing. If you cannot produce the record, you cannot prove the quality — and on a capital project, proof is the whole point.
The check that cannot be produced never happened
Field note from more projects than I can count: the work was reviewed, the calculation was verified, the weld was inspected — and none of it was written down in a way anyone could retrieve later. Then a failure, an insurance claim, or a routine audit asks for the evidence, and the team is left reconstructing from memory and email fragments. The quality was probably real. The record was not. And in that gap live delays, liability, and a great deal of expensive doubt.
What a documented QA/QC process actually looks like
It is not more bureaucracy. It is the same checks you already do, captured as you do them:
Defined checkpoints. Everyone knows which reviews and inspections are required at which stage. It is not left to individual diligence and good intentions.
A named reviewer. Each check has an owner and a date, so we checked that becomes she verified it on the fourteenth.
The evidence attached. The marked-up drawing, the test result, the inspection photo — stored against the item itself, not buried in someone's inbox.
A closed loop. Every issue found is tracked through to resolution, so a caught problem cannot quietly reopen or be forgotten.
The audit trail is the deliverable
Reframe the whole thing and it gets easier. On a capital project, the audit trail is not overhead sitting on top of the quality work — it is the quality work's only durable form. The bridge, the building, the system all get handed over. So does the proof that they were built right. Ten years from now, when someone asks whether a critical connection was ever verified, the honest answer is whatever the record says. And if there is no record, the honest answer is: we do not know.
You can test this tomorrow morning without a single new process. Pick one completed quality check from last week and try to produce its evidence — the who, the when, the what — in under five minutes. If you can, your QA/QC is real and provable. If you cannot, the quality might well be there, but you cannot prove it — and on the day it actually matters, that is the only kind of quality that counts.
Quality you cannot retrieve is quality you cannot defend, and on a capital project the two are the same thing. More field notes on engineering and QA/QC land every week on the XNM blog.
