Technical reference
and sampling test single controls. An audit tests the whole system: whether what's written is right, what's done matches it, and what's recorded is true. It works only if the auditor is independent and every finding is followed through to a fix.
Audits suffer from a sampling problem. An auditor sees only a small part of what happens, often on a day people expect them. That's why evidence that can't be staged carries more weight: a record traced to a batch, a probe reading, a shift that nobody expected to be visited.
Scores can mislead. An average hides a single serious finding, and a checklist reward answers rather than judgement. Many programmes therefore cap the grade when a serious finding exists, whatever the percentage. A rising score with the same repeat finding is a warning sign, not a success.
Parameters
| ISO 22000 | Planned intervals, objective auditors |
|---|---|
| BRCGS Food Safety Issue 9 | At least 4 audit dates a year; every activity at least annually |
| SQF Code | At least annually; independent where practical |
| Open food areas | : hygiene and fabrication inspections at least monthly |
| Independence | Not your own work |
| Closure | on site, not accepted on paper |
Records that prove it
- Audit programme, with changes and reasons
- Each audit: scope, auditor, date, checklist, evidence
- Findings: requirement, evidence, rating, owner, date
- Root cause, action and who verified it
- Missed, expired or cancelled audits, with the reason
- Summary for management review
- Checklist version, with who changed what and when
CCP or prerequisite?
That depends on the for your process. The number, if it has one, belongs to your plan.