A first article report is the closest thing our industry has to a receipt. Like a receipt, it is only useful if you read the line items — and it only tells you about the transaction it records, not about the next one.
What the report contains
A first article inspection report covers the first part produced from a new tool, a new program, or after a meaningful change. Each dimensioned characteristic on the drawing gets a balloon number, and each balloon gets a line.
| Column | Example | Why it matters |
|---|---|---|
| Balloon | #27 | Links the line to one callout on the drawing, so nothing is silently skipped |
| Characteristic | Bore diameter | Names the feature in words, not only by number |
| Nominal | ∅12.000 mm | The design value |
| Tolerance | ±0.010 mm | The band the value must land inside |
| Measured | ∅12.004 mm | What the instrument actually read |
| Deviation | +0.004 mm | Distance from nominal, and how much of the band is consumed |
| Result | Pass | Judgement against the band, ideally computed not typed |
Two details separate a report you can rely on from one you cannot: the deviation column rather than only the pass/fail, and the instrument and setup noted per characteristic. A part that passes at +0.009 mm of a ±0.010 mm band has almost no margin left and will drift out on the next run.
What it proves
Stated precisely, a first article report proves that the process produced one part conforming to the drawing as measured by a documented method on a stated machine. That is worth having. It is the evidence that the tool, the program and the setup can make a good part at all.
- Every dimensioned characteristic was measured, not a convenient subset.
- The values fall inside the drawing's tolerance bands, with the deviations shown.
- The measurement method and instrument are identified per feature.
- The part is traceable to a date, a machine and a lot.
- Any out-of-tolerance item is listed with a disposition, not quietly omitted.
What it does not prove is repeatability. One part has no distribution: it is a single point, and a single point cannot tell you whether the next hundred will land in the same place.
What it leaves out
This is the part of the conversation that saves the most trouble later. Four gaps matter.
- Capability over a run. A first article shows one part inside tolerance. It does not show the spread across a batch. For that you need a small sample measured at intervals, and a capability figure on the critical dimensions only.
- Measurement soundness. The report is only as good as its setup. Probe calibration, part alignment and fixturing are rarely printed. Ask whether the report includes the alignment datum and the probe tip used, because a relocated datum can turn a pass into a fail without the part changing.
- Undimensioned features. Cosmetic defects, flash, sink marks, internal stress and assembly feel are not ballooned, so they are not in the report. They are covered by a separate visual standard or a fit check, and both need to be agreed in writing.
- Fit in an assembly. Individual parts can each pass and still fail together, because assembly is a tolerance stack. If the part goes into a stack, the stack needs its own analysis.
A useful habit. Read the deviation column and rank it by how much of the band is consumed. The two or three characteristics sitting above 70 % of their tolerance are the ones that will cause your next reject, and they are the ones worth naming in a per-lot inspection agreement.
What to ask for
Most buyers either accept no data or demand everything on every lot. Both are expensive. The middle path is a short, named list.
| Your purpose | Ask for | Frequency |
|---|---|---|
| Approve a new tool or program | Full first article report with deviations | Once per tool, plus after changes |
| Monitor production stability | Named critical characteristics with measured values | Per lot or per shipment |
| Satisfy an audit or a customer | Material certificates plus the first article report | Once, retained for the agreed period |
| Investigate a complaint | Raw CMM output for the affected lot, plus the setup notes | On request |
| Compare two suppliers | First article report plus the deviation spread on the same critical features | At qualification |
Ask also that the same characteristics be measured the same way each time. A report that changes method between lots cannot be compared to itself, and comparison is the whole point of keeping records.
On records and retention. We keep inspection records for three years and will state in writing which reports travel with a shipment. The specific set for your parts is agreed with the quotation, because a full report on every lot is a real cost and not every project needs one.
What is a first article inspection report?
A first article inspection report, or FAI, records the measured value of every dimensioned characteristic on a drawing for the first part produced in a new process or after a change. Each drawing characteristic is ballooned with a number, and the report lists the nominal, the tolerance, the measured value and the deviation for each balloon. It is evidence that the process produced a conforming part once, not proof that it will keep doing so.
What does a CMM report not tell you?
Four things. It does not tell you the capability of the process over a run, because one part has no distribution. It does not tell you whether the measurement itself was sound, unless the setup, probe and alignment are documented. It does not cover features the drawing does not dimension, such as cosmetic defects or internal stress. And it does not tell you how the part behaves after assembly, because fit is a stack, not a single value.
Should I ask for CMM data on every lot?
Only for the characteristics that actually matter. A full CMM report per lot is slow and expensive and usually not what protects you: a short list of critical dimensions measured per lot, backed by a first article report and a documented process, gives better protection for less money. Ask for the critical few by name, and ask that the same method is used each time so the numbers are comparable.
How long should inspection records be kept?
Three years is a workable default and matches what we retain. Automotive and medical programmes often require longer, and some require records to follow the part for its service life. State the requirement with the order so retention is set up at the start rather than reconstructed later.
Name the record set before the first lot
Send the drawing and tell us which characteristics protect you. We will put the measurement agreement in the quotation rather than in a later email.
Request a quote