A sampling plan needs a rationale, not a number

A sampling plan usually shows up in a quality file as a table: lot size, sample size, acceptance number, sometimes an AQL borrowed from ANSI/ASQ Z1.4 or a comparable standard. What the table doesn't show, and what a second question from an auditor tends to expose, is why that particular plan was chosen for that particular characteristic. ISO 13485:2016 Clause 8.1, the general clause that opens the standard's measurement-and-improvement section, requires an organization to determine appropriate methods — including statistical techniques — and the extent of their use, and to document that determination. A sampling table with no traceable reasoning behind it has produced a number, not the clause's actual deliverable.

Clause 8.1 sits above the clauses it enables

Clause 8.1 opens Section 8 of ISO 13485:2016 before the section's more familiar sub-clauses take over — feedback and complaint handling, regulatory reporting, internal audit, and monitoring of processes and product under 8.2, nonconforming product under 8.3, data analysis under 8.4, corrective and preventive action under 8.5. What 8.1 requires is the planning layer underneath all of them: the organization has to plan and implement the monitoring, measurement, analysis, and improvement processes needed to demonstrate product conformity, ensure QMS conformity, and maintain the QMS's effectiveness, and that planning has to include determining appropriate methods — statistical techniques among them — and how far each one gets used. Every sampling plan a manufacturer runs, from incoming inspection to final release to the sample size behind an internal audit, is one instance of a method Clause 8.1 requires be chosen on purpose, not carried forward from whatever the last project used.

A cited standard is a method, not a justification

Naming a sampling standard and applying it correctly are not the same act as satisfying Clause 8.1. A record that cites “ANSI/ASQ Z1.4, General Inspection Level II, AQL 1.0” has named a method; it hasn't shown the extent-of-use determination the clause actually asks for unless it also connects that specific inspection level and AQL to the severity of failure for the characteristic being sampled. A critical dimension tied to a patient-contact function and a cosmetic surface attribute don't belong under the same AQL by default, and a plan that applies one Z1.4 table across an entire inspection scheme without differentiating between them has skipped the judgment call the clause exists to force, not just the paperwork around it.

The same clause governs full inspection and none at all

“Extent of their use” cuts in both directions from a partial sample. Choosing to run 100% inspection on a characteristic, and skipping the sampling risk that comes with anything less, is a documented decision Clause 8.1 covers just as much as choosing a sample size — and so is choosing to run no incoming inspection at all on a lot from a supplier qualified and monitored under Clause 7.4. A quality system that treats partial sampling as the default and full or no inspection as the unusual exception has the clause's own posture backwards: every point on that spectrum needs the same kind of traceable reasoning, not just the middle of it.

A sampling plan's rationale has a shelf life

The risk profile a sampling plan was built against doesn't stay fixed. A new supplier, a tooling change, or a shift in the defect rate a nonconforming-product record starts to show can move a characteristic's actual risk well past what an old AQL was chosen to catch, and the plan doesn't reset itself just because the paperwork behind it hasn't changed. A CAPA tracker is usually where that drift first becomes visible in practice, but Clause 8.1's own determination is what should have been reviewed on a cadence tight enough to catch it earlier, rather than waiting for a trend line to force the question.

Where this meets the rest of the file

An inspection-and-sampling rationale worksheet that ties each cited standard and AQL back to a documented risk basis, and treats full inspection and no inspection as decisions owed the same reasoning as any partial sample, is previewed in the launch catalog. If your program justifies its sampling plans differently, the shelf takes that correction directly.