The machining inspection report: four people, four readings of the same document

---

What we're actually talking about

A word on vocabulary, because it shifts from shop to shop and makes conversations harder than they need to be. Machining inspection report, dimensional inspection report, first article inspection report, quality package: depending on who you're talking to, these sometimes mean the same thing and sometimes don't.

What they all share fits in one sentence. You machine a part from a dimensioned drawing. That drawing carries dozens of dimensions, each with its tolerance. The customer isn't only paying for the part: they're paying for the proof that every dimension meets the drawing. The report is that proof, put into a usable form.

The standard gives it a precise role. Release of products (§8.6) forbids shipping until the planned verifications are complete and documented, with a record of who authorized the release. No report, no release. The document doesn't sit beside the part — it's part of it.

I spent fourteen years in quality before I started building software, and I have never seen a shop where that definition was actually agreed upon. Not because people don't know it, but because each person looks at it from where they stand — and where they stand proves them right.

The machinist: "my part is good, the paperwork is wasted time"

Start with the person who makes the part. For him, the report always shows up at the wrong moment: at the end, once the part is done and there are twelve more to run.

What you should hear in "it's just paperwork" isn't laziness. It's this: he knows the part is good. He watched it come off, he measured it, he adjusted his offset along the way. The part is good — and now he's asked to spend twenty minutes writing that down. From where he stands, he isn't being asked to work. He's being asked to justify himself.

And most of the time, the form proves him right. He retypes values the machine already knows. He copies the work order number that's printed on the sheet right next to him. He fills in boxes nobody ever explained to him. When the only feedback you ever get on a document is a reminder that you forgot to fill it in, you end up filling it in the way you close a door: fast, without looking.

A report filled in from memory at the end of a shift isn't proof. It's a recollection, neatly formatted.

The real cost isn't in those twenty minutes. It's in the gap between the moment of measurement and the moment of writing. In between sits memory — and memory rounds. It rounds toward the expected value, never at random. That's an ordinary human mechanism, and it's exactly what a system of proof is supposed to prevent.

The inspector: the trust trap

Now the person who signs. His position is more uncomfortable than it looks, and it's the one least discussed on the floor.

In many small shops, part of the inspection is self-inspection: the machinist measures his own part. The inspector countersigns. On paper there are two levels of verification. In practice there's often only one, because the second signature comes with no second measurement.

I say that without blame, because it isn't an individual failing: it's what the workload produces mechanically. The inspector knows his people. He knows this one measures carefully and that one moves too fast. He trusts them — and that trust is usually well placed. The problem isn't that it's misplaced. The problem is that it's invisible: nothing in the finished report distinguishes a value that was re-measured from one that was carried over as-is.

Yet he is the one signing the release. The day a part comes back, nobody will ask the machinist why he wrote that value. They'll ask the inspector why he approved it. His signature covers measurements he didn't all take, and he knows it.

Self-inspection isn't the problem — it's a healthy practice, and often the only workable one in a thirty-person shop. The problem is practising it without saying so. A shop that owns its self-inspection defines what gets re-measured, in which cases, and how often; the countersignature then means something specific. A shop that doesn't own it leaves everyone guessing what a signature is worth — and the honest answer, at that point, is: it depends who was in a hurry that day.

The owner: another cost

Third reading, the one from the corner office. It's simple and it's accounting: every hour spent on a report is an hour that machines nothing.

That reading is correct. It's only incomplete, and I won't repeat it here because I've devoted a whole article to it: the hidden cost of machinist time puts numbers on what transcription ties up, and why that bill never shows up on any accounting line.

What I want to add here is what the accounting view misses. The owner sees the report as a burden because he's looking at it from the spending side. But this document is also the only thing that separates his shop from a cheaper one. A customer choosing between two suppliers at the same price doesn't decide on part quality — they haven't seen the parts yet. They decide on the package.

Put differently: the report isn't what costs, it's what sells. Seeing it as overhead is accurate for the current month and wrong for the next contract.

The customer: a go/no-go without ever seeing the part

The fourth reading is the least understood from inside the shop — because it happens somewhere else, and nobody on the floor has ever watched it happen.

At the customer's end, someone opens your package. That person is almost never the one who will use the part. Often they'll never see it. They have your report on screen, a completeness requirement, and a queue of other packages behind yours.

Their job isn't to evaluate your work. Their job is to decide whether the package passes. Are all the dimensions there? Do the values fall within tolerance? Are the signatures and dates present? Are the instruments identified? In aerospace it's tighter still: the first article form has a mandated layout, and one empty field is enough to have the whole thing rejected.

It's worth measuring what that implies, because it runs against the instinct you have on the floor. At that desk, a report that is clean and wrong passes. A report that is right and messy comes back. That isn't bad faith: it's the only thing that desk can do. You can't verify a measurement on paper. You can only verify that what's written holds together.

The customer isn't judging your parts. They're judging the only thing they have in front of them: your document.

That's the perverse effect nobody names. Since form is the only thing verifiable at a distance, form becomes the criterion. A shop that has understood this can produce immaculate packages on approximate measurements — and it holds up for a long time. Until the part that fails at the customer's, when the package comes back out, and every signature suddenly carries its full weight again.

Why all four are right at once

Put the four readings side by side and none of them is unreasonable. The machinist is right: he's made to retype what's already known. The inspector is right: he's asked to guarantee what he couldn't always verify. The owner is right: those hours produce nothing. The reviewer is right: they can only judge what's in front of them.

So the conflict isn't between people. It's structural. A single document is being asked to be a defence for one, a commitment for another, an expense for the third and proof for the fourth. No form pulls that off, however willing the people filling it in.

And there's a consequence we'd rather not admit. Because nobody truly reads this document while things are going well, it faithfully records how each person works around it. The fields left blank every time, the values that are a little too round, the signatures applied in a batch on Friday, the instruments never identified: all of it accumulates while no one is looking.

Then comes the day someone does look — a customer return, a dispute, an audit, a part that fails. That day the report stops being a formality and goes back to what it always was: an exact portrait of the shop that produced it. Nobody wrote it for that purpose. It does it anyway.

And the standards have made it harder

Something has to be added here, because the situation I've described has grown worse in recent years rather than better.

A shop working for aerospace, defence or automotive no longer deals with ISO 9001 alone. It deals with AS9100, with IATF 16949, and above all with first article inspection requirements formalized to an extreme: mandated forms, required fields, every drawing dimension ballooned and reported, every characteristic accounted for. These requirements aren't absurd — they come from real accidents and hard-won lessons nobody wants to repeat.

But their combined effect on a thirty-person shop deserves to be named: we've built an administrative monster that protects the paperwork better than it protects the part. By demanding ever more formal proof, we've made producing that proof so heavy that it gets done under exactly the conditions that undermine it — fast, at the end, from memory, by someone with twelve other things to do.

The harder the demand for proof gets, the worse the conditions under which it's produced. That's a paradox, and it's ours.

This isn't an argument against standards. It's an argument for no longer treating compliance as clerical work bolted onto the trade.

Produce the proof once

The way out isn't "be more rigorous". That doesn't work, and it would insult people already doing their best in the time they're given. The way out is structural, and it fits in one line: produce the proof once, at the moment of measurement, then let each role read what concerns them.

In practice, that means three things.

Capture while measuring, not at the end of the shift. It's the single change that removes the gap between the act and the writing — and therefore memory, and therefore rounding. This isn't a matter of discipline. It's a matter of where the form is.

Never make anyone retype what's already known. The work order number, the drawing dimensions, the tolerances, the min and max: all of it already exists somewhere. That's exactly the work that automated drawing reading, framed by purpose-built software, absorbs properly — provided it flags what it isn't sure about instead of quietly sliding it into the table. What stays human is judgment: choosing the instrument, settling an ambiguous dimension, ruling on a borderline case.

Make verification visible. This is the point that resolves the inspector's discomfort. If the system distinguishes a value measured by the machinist from one re-measured in inspection, the countersignature regains a precise meaning. Self-inspection becomes a stated policy rather than a tacit arrangement — and in front of an auditor, that difference is considerable.

None of this is spectacular. Not one of these three points demos well in a meeting room. Together, they let a single record serve all four readings without asking anyone to do the work twice.

None of it is new either, and that should be said plainly: it's what good metrology tools and manufacturing execution systems — MES — already do when they're genuinely implemented. The instrument, the measuring arm or the digital gauge feeds its value straight in, with no re-keying. Drawing reading is automated, with uncertain areas flagged rather than hidden. Self-inspection and independent inspection are two distinct steps, with visible status. And every measurement carries its instrument, its valid calibration, its author and its timestamp. The technology exists. It isn't the hard part.

The real obstacle isn't technical

Because the hard part is elsewhere, and I'd rather say so than let anyone believe software settles this on its own.

The obstacle is cultural and organizational. It means getting people to accept that filling in the report is no longer administrative drudgery that follows the work, but the natural record of the act of measuring. That isn't a turn of phrase: it's a complete reversal of the document's status, and therefore of who owns it.

That reversal unsettles everyone at first. The machinist, because he's asked to record as he measures rather than afterwards. The inspector, because his work becomes visible: what he actually re-measured stops being a grey area. The owner, because those minutes have to be accepted as part of the work rather than overhead. I've seen technically flawless rollouts fail on this point alone, and more modest tools succeed because the shop understood what was really being asked of it.

---

In closing: who, in your shop, actually reads this document?

I'll leave you with a few questions that answer themselves from your own files, not from my examples.

Across your last ten reports, how many values were written at the moment of measurement, and how many reconstructed afterwards? When your inspector countersigns, what is genuinely re-measured — and is that written down anywhere, or does it depend on the day? If a customer asked you tomorrow for the calibration certificate behind a dimension measured three months ago, how long would it take you? And the question that contains the other three: when was the last time someone in your shop read an inspection report for any reason other than checking that it was filled in?

If that last answer is "at the last dispute", you don't have a rigour problem in your shop. You have a document with four readers, none of whom was ever equipped to find what matters to them in it.

And if your reports still live in spreadsheets or in PDFs filled in by hand at the end of the shift, you already know you're the case described here. The day a serious customer or an auditor digs in, the document will talk. It always does.

The principles described in this article are the ones that guided the development of Asterion Solutions, a suite of trade-specific tools built for small manufacturers who want to structure their quality without multiplying paperwork.