Keep a supplier's specification limits separate from test results

A limit is a requirement, not a result

Keep them separate so the register does not mistake a requirement for evidence that it was met.

Our public material-register example shows a synthetic report beside downloadable output. Use it to inspect how values and source references travel together. The self-service ParseMill trial is for invoice PDFs; supplier test reports and engineering review need a separate fit discussion.

Even when a number appears within a stated range, a conformance conclusion may depend on method, units, uncertainty and an applicable decision rule. This extraction worksheet records evidence; it does not make a material approval decision.

Use a limits-versus-results schema

The values are synthetic. S2 retains a qualified result rather than an exact value at ten. S3 does not acquire the minimum or maximum as a substitute measurement. S1’s Pass is a source-reported statement and should not be presented as an independent approval by the register author.

ILAC: Guidance on decision rules and conformity identifies decision rules as part of issuing statements of conformity. Preserve the distinction between recording a result and applying such a rule.

Review the field roles

1. Read the complete heading hierarchy, including merged headings above minimum, result and maximum subcolumns. Record the role of each field.
2. Capture raw values, qualifiers and units separately for observations and limits. Verify they belong to the same sample and test context.
3. Store source-reported conformity statements in their own field with a source reference. Keep your organization’s approval status in a separate reviewed workflow.
4. Check cases with blank results, equal boundary values and qualified results. Confirm that no formula silently substitutes limits or generates an unsupported pass/fail label.

NIST: SI notation checklist emphasizes clear value-unit relationships. Compare units explicitly before any technical interpretation, while retaining the original labels and source text.

Catch plausible column shifts

If minimum, result and maximum share similar magnitudes, a misplaced value can look reasonable. Inspect row and column associations rather than relying only on whether the result falls between two numbers. Include a fixture where the result is blank to expose accidental limit copying.

Do not infer a missing decision rule from a familiar industry practice. Record the reported rule if available and ask the responsible technical reviewer for the applicable interpretation when a decision is required. Keep that decision linked to its governing requirements and reviewed evidence.

Check the evidence handoff

For the fixture, the register keeps three distinct numerical roles and one separate source statement. S2’s qualifier and S3’s missing result remain visible. A reviewer can then assess the evidence under the appropriate procedure.

Use adjacent specification and result columns as acceptance criteria for a separately scoped supplier-report pilot. Ask for an output whose labels and associations you can verify, then keep material acceptance with the responsible reviewer. The self-service invoice offer does not perform engineering approval.