A food safety internal audit exists to answer one question with evidence: is the hazard plan we wrote the plan we are actually running. Everything else follows from that. The plan names hazards, the controls that manage them, the limits those controls must hold, who monitors them, how often, and what happens when a limit is missed. Each of those is checkable on the floor and in the records, and the gap between the written plan and the running line is where every meaningful finding lives. Prerequisite programmes, supplier approval and traceability then wrap around it, and each has its own short verification.
Verify the plan against the line, in that direction
Read the plan, then go and watch the step. Is the control point where the plan says it is, is the limit the one on the paperwork, is the person monitoring it the person named, and are they doing it at the stated frequency rather than at the end of the shift from memory. Auditing in the other direction, by watching the line and then finding the paperwork that matches, misses the controls that were dropped entirely, because nothing on the floor advertises the absence of a check nobody does.
Monitoring records and the corrective actions behind exceedances
Sample monitoring records across the period, not the last week, and look specifically for the entries that are out of limit. A period with no exceedances at all is worth a second look, because processes vary and perfectly clean records more often mean unrecorded readings than perfect control. For each exceedance, the chain is the record, the immediate action on the product, the correction of the process, the decision on affected product, and the sign-off. A missing product decision is the finding that matters, because it is the one with product still in the market attached to it.
Prerequisites, suppliers and traceability
Prerequisite programmes are audited by sampling: cleaning schedules with completed records and verification, pest control reports with the actions closed, glass and plastic registers reconciled, allergen segregation observed rather than described. Supplier approval is audited by picking a delivered material and asking for the approval file behind it. Traceability is audited by a mock recall run against a real batch code, timed, and the test is not whether the paperwork exists but whether the quantities reconcile at each step.
Verification is not the same as monitoring, and auditors check you know
Monitoring is the routine check that a control is holding. Verification is the separate activity that confirms the monitoring itself is working: reviewing records, calibrating the thermometer, testing the finished product, and reassessing the plan when the process changes. Sites that fold verification into monitoring end up with a system that checks itself with the same instrument that might be wrong. An internal audit is itself one of the verification activities, so its records are part of the evidence a certification body or a customer will sample.
Questions people ask about food safety internal audit
How often should a food safety internal audit be run?
At a frequency your programme justifies by risk, with the whole scope covered across a cycle, plus a reassessment triggered by change: new product, new equipment, new supplier, new line layout, or an incident.
Can the person who wrote the plan audit it?
Not independently. They can support the audit as a technical resource, but the finding has to be raised by someone who did not design what is being tested, or the audit cannot detect a design error.
What is the most common food safety internal audit finding?
Records complete but verification missing, and corrective actions that address the batch without addressing the cause. Both look tidy on file and both leave the same hazard live.
Does a mock recall have to be a full exercise?
It has to be realistic enough to produce a reconciliation and a time. A paper exercise that assumes the records are correct tests nothing, because the failure mode being tested is precisely that they are not.