The Inspection Pack Is a Claim, Not a Certificate: Reading Dimensional Evidence Before Sign-Off

A precision part arrives with an inspection pack. The cover sheet is neat, the columns are populated, and every characteristic appears to sit inside its tolerance band. For many buyers, that is the end of the technical conversation. The purchase order is closed, the goods are booked in, and the report is filed. The difficulty is that an inspection report is not a certificate of conformity in any absolute sense. It is a claim about a measurement process, made at a particular time, on a particular sample, using a particular method. Read without that context, a page of green ticks can be more comforting than informative.

A report records a method, not a population

The first question to ask of any dimensional report is what population it describes. A batch of 500 machined housings may be accompanied by a report showing five parts measured in full. That is not necessarily wrong. Sampling is a legitimate part of industrial quality control, and 100% inspection is not always technically or commercially sensible. But the buyer needs to know whether the five parts were selected to represent a stable process, or selected because they were the first five to hand.

A useful inspection pack should state the sampling plan, the batch size, the number of parts examined and the acceptance rule applied. If the report is silent on those points, the ticks describe the sample, not the batch. That distinction matters most where a characteristic is safety-related, where the process is new, or where the supplier has limited history on the part. In those cases, a small sample can be a reasonable screening tool, but it is not evidence of process control across the whole delivery.

Measurement uncertainty sits behind every number

A CMM report often presents dimensions to two or three decimal places, which creates an impression of certainty. In reality, every measurement carries uncertainty. The machine may be calibrated and the laboratory may hold UKAS accreditation, but the uncertainty budget for a particular feature depends on the probe, the styli, the temperature, the part geometry and the measurement strategy. A reported value of 24.991 mm against a limit of 24.990 mm is not simply a pass. It is a pass whose margin may be smaller than the measurement uncertainty itself.

Buyers do not need to reproduce the uncertainty budget. They do need to know whether one exists and whether the supplier has applied a decision rule. Under ISO 14253-1, the treatment of measurement uncertainty near specification limits should be agreed, not assumed. A report that marks a feature as conforming without stating the uncertainty or the decision rule is offering a conclusion without its basis. That is not a reason to reject every part, but it is a reason to ask how the supplier decides when a result is close to the limit.

Datum callouts are part of the specification

Datum structures and GD&T callouts are often treated as drawing decoration by commercial teams, yet they determine how the part is set up and how the CMM program is aligned. A positional tolerance without the correct datum sequence is not the same requirement, even if the reported number looks acceptable. The inspection report should show which datum features were used, how the part was restrained, and whether the measurement software applied the drawing's datum precedence.

This is particularly important where a drawing uses composite position tolerances, datum shift or maximum material modifiers. A report may show a single position value and a green tick, but the underlying evaluation can be sensitive to the alignment method. If the supplier cannot provide the CMM program revision, the alignment report or a clear statement of the datum simulators used, the buyer has little basis for judging whether the reported result matches the drawing requirement. The tick may be real, but its meaning is not yet established.

Questions that separate evidence from closure

The practical response is not to demand a metrology lecture with every delivery. It is to ask a short set of questions proportionate to the risk. For a routine part with a stable history, a light touch is appropriate. For a new supplier, a first article, or a safety-related feature, the following questions are worth putting in writing.

• What population does this report represent, and how were the measured parts selected?
• What is the measurement uncertainty for the reported characteristics, and what decision rule was applied near the limits?
• Which datum features were used, and how was the part restrained during measurement?
• Which revision of the drawing and the CMM program does the report relate to?
• Were any characteristics outside tolerance, and if so, how were they dispositioned?
• Is there evidence of process control, such as trend data or SPC, or only a pass/fail snapshot?
• Can the report be traced to the specific batch, material certification and heat or surface treatment records?

None of these questions is hostile. Each one treats the inspection pack as a technical document rather than an administrative formality. That is the correct posture for a buyer of precision parts. The supplier who can answer them quickly is usually the supplier who understands the difference between measuring a part and controlling a process. The supplier who cannot may still be capable, but the buyer is entitled to know where the evidence stops.

A green tick is a useful signal. It is not a substitute for understanding the method behind it. In precision supply, the inspection report should open a conversation about capability, not close one about paperwork.