FAI Reports: What a First Article Inspection Should Actually Contain

Article

October 26, 2026

“The sample looked fine” and “we have an FAI” are different statements, and the gap between them is where most quality arguments start. A sample is a physical object that somebody looked at. A first article inspection is a data set: this drawing revision, these features, this is what each one measured, this is what it was measured with, and this is who signed it.

The difference shows up at the second delivery, not the first. A sample can only be compared by eye. A data set can be compared against the next data set — which is the whole point, because the question that costs money is never was the first piece good. It is is piece four thousand the same as the first piece.

This article sets out what a report has to contain, how reports get inflated without anyone technically lying, and what to check in the ten minutes after one lands in your inbox.

FAI Is a Data Set, Not a Sample

The distinction is worth stating plainly because it decides what you ask for. If you ask a supplier for “a sample”, you will get a part and a photograph. If you ask for “an FAI to full dimensional, at drawing revision C, with the instruments listed”, you will get something you can hold a later batch against.

The second request is also the one that reveals what kind of supplier you are dealing with. A factory that has been doing this for years has a template, a routine, and a folder of past reports it can show you — redacted — before you place an order. A factory that has not will treat the request as an unusual favour, and the report, when it finally arrives, will show it.

Seven Things the Report Must Show

Seven items separate a document that is evidence from a document that is decoration. None of them is exotic; all of them are routinely missing.

Of the seven, the drawing revision is skipped most often and matters most. If the report cites revision C and you approved revision D, everything in it is a careful measurement of a part you did not order — and it will pass every visual check anyone applies to it.

Five Ways a Report Gets Inflated Without Anyone Lying

Outright fabrication is rare. What is common is a report that is technically true and evidentially empty. Five patterns cover most of it.

“OK” instead of numbers. A cell that says OK has recorded an opinion, not a measurement. It cannot be checked, compared, or used in an argument, which is precisely why it appears.

A sample report presented as a production FAI. The sample was made before the process was set — often on a different machine, by a different person, with more time. It tells you what the process can do on its best day, not what the production run will deliver.

An outdated revision. Covered above, and the single most expensive item on this list, because it invalidates the entire document while leaving it looking complete.

Only the dimensions that were easy to measure. A report with forty of the sixty-two dimensions on the drawing, where the twenty-two missing ones are the tight ones, is not an oversight. It is a selection.

An instrument that cannot resolve the tolerance. A 150 mm vernier with 0.02 mm resolution cannot meaningfully verify a ±0.01 mm tolerance, however careful the operator. The usual rule of thumb is that the gauge should resolve to about one tenth of the tolerance. A report that lists a caliper against a ±0.01 mm feature has given you a number, not a fact.

Three Numbers to Check When It Arrives

None of these requires metrology training. They require ten minutes and the drawing you sent.

The tightest tolerance on the drawing. Find it, then find that dimension in the report. Is the reported actual inside the band, and is the instrument listed capable of the check? If the tightest callout is missing from the report, stop there — the rest of the document has told you what it is worth.

The drawing revision. Compare the revision cited in the report against the one you approved in writing. This takes thirty seconds and catches the most expensive error on the list.

The material lot number. Then compare it against the mill certificate. If the report has no lot number at all, the material claim is a word rather than a record — and a certificate that cannot be tied to a lot does not tell you what is in the bar you are holding.

Full FAI or Sampling: When Each Is Right

A full FAI measures every characteristic on the drawing. A reduced one measures the critical characteristics and relies on process control for the rest. The choice is not a matter of budget preference; it follows from what has changed.

Full FAI when any of these is true: a new part; a new supplier; new or modified tooling; a drawing revision that touches dimensions; a process change, including a change of machine, fixture or finishing subcontractor; or production resumed after a long gap.

Reduced inspection when nothing has changed: repeat order, same drawing, same process, same tooling. In that case a first-piece check on the critical characteristics plus in-process inspection is the proportionate answer, and asking for a full FAI every time trains everyone to treat it as paperwork.

The aerospace industry codified exactly this distinction in AS9102, which separates a full FAI, a partial FAI and a delta FAI. The standard itself applies to aerospace work and not to general machining; the logic transfers regardless, and a supplier who recognises the terms is a supplier who has done this before.

What the Report Cannot Tell You — and What to Put in the PO

An FAI is one part, measured once. It cannot tell you whether the process is stable, whether parts two through four thousand will match, whether the assembly fits, or whether the material is what the certificate says. Those are four separate questions with four separate answers — process capability, in-process inspection, functional testing, and material verification.

What the report can do is fix a reference point. That only works if the purchase order says so, so five lines are worth adding to it:

1. “FAI to full dimensional, on the first piece of the production run, at drawing revision ___.”
2. “Report to list the instrument used for each dimension and the material heat or lot number.”
3. “No production shipment before written FAI approval.”
4. “Any deviation to be raised in writing before shipment, with a proposed disposition.”
5. “A new FAI is required after any drawing revision or process change.”

One honest note about us. Our product page states that first articles are typically ready in one to two weeks with inspection reports, and that PPAP or your own template can be followed. The sentence on a page is not the test. The test is a redacted report from a past job, and it is worth asking for one before you place an order — from us, and from anyone else shortlisted. A supplier who cannot produce one has not been producing them.