Begin with the consequences of a missed defect
Not every characteristic deserves the same inspection effort. A cosmetic mark on a hidden surface differs from a dimension that prevents assembly or a defect with a safety consequence. Identify the important failure modes and the evidence needed to detect them.
Check the basic identity first: part number, revision, quantity, material documentation where required, and shipment condition. A perfectly measured obsolete revision is still the wrong shipment. Preserve packaging and lot labels until the parts are entered into your records.
NIST describes acceptance sampling as a method for deciding a lot's disposition, not a method for precisely estimating its quality. That distinction matters: a sample-based accept decision does not mean every unit was verified or that the sample's defect percentage is an accurate population estimate.
Define the lot and the selection method
A lot should have a meaningful identity related to how the parts were produced and delivered. If a shipment contains multiple material lots or process setups, determine whether they should be treated separately. Mixing them can hide a localized problem.
Select samples using the agreed method rather than taking the easiest parts from the top of a box. Convenience selection can miss damage or variation elsewhere. Record which units were inspected and any relevant location or lot information.
A sampling plan includes more than sample size. It specifies acceptance and rejection rules and is chosen with consideration of risks. NIST's sampling-plan material describes those tradeoffs. Do not copy an AQL label or a sample count without understanding the associated decision and the applicable plan.
A hypothetical shipment of small brackets
An inventor receives one hundred brackets in four labeled bags. The critical questions are correct revision, mounting fit, and surface damage that could interfere with assembly. The inventor initially plans to inspect five pieces from the first bag because they are easy to reach.
A better plan first confirms whether the bags represent one lot or different setups. The selection and acceptance approach is agreed before inspection. Identity and shipping damage checks may be broad, while dimensional checks follow the justified plan appropriate to the application. This example does not prescribe a universal sample size.
Suppose one inspected bracket fails the mounting requirement. The inventor follows the predefined disposition rather than replacing it with another sample until the count looks acceptable. The affected lot is kept separate while the supplier and inventor investigate the cause and agree on screening, rework, replacement, or another appropriate action.
If the defect is found only in one identified setup, traceability can narrow the investigation. Without the bag and lot records, all one hundred parts may remain equally suspect, increasing delay and cost.
Make the measurement method part of inspection
Inspectors need a consistent definition of the feature and a suitable method. A go/no-go fixture may be useful for a functional interface if its design and maintenance are appropriate. It should not be assumed to establish every dimensional requirement merely because the part fits.
Record actual values when they are useful for diagnosis, not only pass marks. A trend toward a limit can matter even when all sampled parts pass. Small samples do not automatically support process-capability claims, but they can identify a question worth discussing with the supplier.
For appearance, define viewing conditions and examples where practical. Without a shared reference, minor surface judgments can become inconsistent. Separate functional defects from cosmetic preferences while preserving the criteria agreed in the order.
Keep disposition and supplier feedback connected
Use a receiving record with lot identity, quantity, inspection method, sample selection, results, and disposition. Mark accepted, held, and rejected material so it cannot be mixed accidentally. Record any approved rework and its verification.
Send factual evidence to the supplier when a problem appears: part revision, lot, requirement, actual result, method, and photographs. Ask for containment and investigation appropriate to the issue. A replacement shipment may solve immediate supply but does not necessarily prevent recurrence.
The deliverable is a repeatable receiving process with a known decision rule. It should detect important problems, preserve evidence, and support a fair supplier conversation. It cannot turn an arbitrary small sample into certainty about every part in the box.
Plan for destructive checks
If an inspection consumes or damages a sample, identify that cost and the sample's final disposition before receiving the batch. Do not return a tested-to-damage unit to accepted stock. The need for destructive testing is one reason a sampling approach may be appropriate, but it does not remove the need to understand the plan's risks.