GIFSQ Food Safety Desk · **Published:** September 30, 2026 · **Sources:** Codex CAC/RCP 1-1969, ISO 19011:2018, 21 CFR Part 117

External auditors find what internal audits missed. That gap — between what your system says on paper and what happens on the floor at 2 a.m. — is exactly what an internal audit exists to close. Done well, it’s the cheapest corrective action you’ll ever take. Done as a tick-box exercise, it’s the most expensive surprise you’ll ever get.

Plan the audit before you walk in

Define scope and criteria in writing. Scope: which areas, lines, or shifts. Criteria: the standard you’re auditing against — your own procedures, 21 CFR Part 117, a GFSI scheme, customer requirements. An audit without written criteria is an opinion walk.

Pick auditors who don’t own the area. Independence isn’t bureaucracy; it’s the whole point. A sanitation supervisor auditing sanitation will see what they expect to see. Cross-train auditors from other departments or sites.

Review documents first. Pull the procedures, last audit’s findings, corrective action records, and recent monitoring data before the opening meeting. Half your findings will already be forming — you’re just confirming them on the floor.

Notify the area, but don’t rehearse it. Give supervisors the schedule and scope so people and records are available. Never pre-share the checklist questions; you want normal operations, not a performance.

Conduct it on the floor, not in the office

Open with a short meeting. State the scope, the criteria, how findings will be reported, and when. Ten minutes. Then get out onto the floor.

Follow the product flow. Start at receiving and walk forward to shipping. Flow-following catches handoff failures — the allergen changeover between products, the rework bin nobody documented — that department-by-department auditing misses.

Ask, watch, then verify. Interview operators (“show me how you check the metal detector”), observe the task, then pull the record. When all three agree, the system works. When they don’t, you’ve found a finding — and the record usually tells you which version is true.

Write findings as facts, not adjectives. “The CCP monitoring log for Cooker 2 is missing entries for March 3–5” is a finding. “Monitoring seems inconsistent” is a complaint. Every finding needs: what you saw, where, when, and which requirement it breaks.

Grade severity honestly. Critical: direct food safety risk or regulatory breach — product on hold today. Major: system breakdown that could become critical. Minor: isolated lapse in a working system. Observation: no nonconformity yet, but a trend worth flagging. Inflating minors into majors destroys credibility; softening majors into observations destroys the system.

Close with a meeting the same day. Present findings to area management while memories are fresh. No ambushes in the report later — if it’s a surprise in the report, you failed the closing meeting.

Write the report people actually fix things from

One finding per item. Each carries: the factual description, the requirement it violates (procedure number, regulation clause, standard), severity, and the evidence (record ID, photo, interview). Skip the background essay — auditees fix facts, not narratives.

Set response deadlines by severity: critical findings get immediate containment plus a corrective action plan within days, majors within 30 days, minors within 60–90. Track every one to closure; an open finding from the last audit is itself a finding in this one.

Corrective action: fix the cause, not the symptom

Correction stops the bleeding: hold the product, re-clean the line, retrain the operator. Corrective action removes the cause: why did the CCP log go unfilled for three days — and what changes so it can’t happen again? Most failed corrective actions are just corrections wearing a fancier name.

Demand root cause on every major: 5 Whys, fishbone, or plain structured thinking — the method matters less than refusing to accept “human error” as a cause. “Human error” is where investigation goes to die; the real cause is the system that let the error reach the product.

Verify effectiveness, don’t assume it. Re-audit the area, check the records for the next 30 days, confirm the trend moved. Close the finding only with evidence.

—

The 40-point checklist

Audit program and management (1–5)

1. A documented internal audit schedule exists, covers all food safety areas, and is being followed — not just written.
2. Auditors are trained in audit technique and independent of the areas they audit.
3. Findings from the previous audit are all closed with evidence, not just marked complete.
4. Management review actually discusses audit results, complaints, and corrective actions — minutes prove it.
5. Food safety objectives are set, measured, and reviewed; missed targets trigger action.

Sanitation (6–11)

1. A master sanitation schedule lists every task — what, how, with what chemical, by whom, how often — and completed tasks are signed off.
2. Cleaning is verified, not assumed: visual checks plus ATP swabs or microbial testing on a defined frequency, with results trended.
3. Chemical concentrations are checked with test kits at defined frequencies; out-of-spec results trigger re-cleaning.
4. Allergen changeovers follow a validated procedure, verified each time with visual inspection plus protein swabs or equivalent.
5. Cleaning chemicals are stored away from food areas, labeled, and used at label dilutions — no unmarked spray bottles.
6. Environmental monitoring (pathogen swabbing zones 1–4) runs on schedule, positives trigger documented corrective action and intensified sampling.

Allergens (12–16)

1. An allergen map identifies where each allergen is used, stored, and where cross-contact could occur.
2. Labels are verified against the actual recipe at every product changeover — wrong-label checks are documented.
3. Allergenic and non-allergenic products are segregated in storage and production; scheduling runs allergen-free first where applicable.
4. Rework containing allergens is identified, labeled, and only used back into products containing the same allergen.
5. Supplier allergen declarations are on file and current; new suppliers are assessed for allergen risk before approval.

Pest control (17–20)

1. A licensed pest management contractor services the site on schedule; licenses and insurance are current.
2. Service reports are reviewed by site management — not just filed — and recommendations are actioned with dates.
3. A current map shows every bait station, trap, and insect light; devices match the map on the floor.
4. Any pest activity noted in reports triggered documented corrective action: findings, proofing repairs, trend review.

People: hygiene and training (21–25)

1. Training records prove every employee completed food safety induction before starting work, plus role-specific training.
2. Refresher training happens on schedule, and its effectiveness is checked — not just attendance sheets.
3. A health screening policy excludes ill workers from food handling; return-to-work criteria are defined and followed.
4. GMP rules (handwashing, protective clothing, jewelry, eating/drinking) are posted — and actually followed on the floor during the audit.
5. Visitors and contractors sign in, receive hygiene instructions, and are supervised in production areas.

Receiving and storage (26–30)

1. An approved supplier list exists; suppliers are approved against defined criteria and re-evaluated on schedule.
2. Incoming goods are inspected on receipt — temperature, condition, labeling, allergen status — and rejections are recorded.
3. Cold storage holds 4°C (40°F) or below, freezers hold −18°C (0°F) or below; temperature logs are complete and reviewed.
4. Stock rotation follows FEFO/FIFO; expired or damaged goods are segregated and clearly marked, never in general stock.
5. Allergens are stored segregated and labeled; chemicals are stored separately from all food and packaging.

Production and process control (31–35)

1. CCP monitoring records are complete, in real time, and signed — no backfilled logs, no gaps.
2. Out-of-limit CCP results triggered the defined corrective action: product hold, assessment, disposition records exist.
3. Thermometers and scales carry current calibration stickers; calibration records reference a traceable standard.
4. Foreign material controls (sieves, magnets, metal detection, x-ray) are checked at defined frequencies with test pieces; failures trigger holds.
5. Hold and release procedures work: nonconforming product is identified, segregated, labeled, and only released by authorized sign-off.

Maintenance (36–37)

1. Preventive maintenance runs on schedule for food-contact equipment; missed tasks are rescheduled, not silently dropped.
2. Lubricants used where incidental food contact is possible are food-grade (H1); maintenance chemicals are controlled like any other chemical.

Traceability and recall (38–40)

1. Lot coding on finished product is legible and links to raw material lots — one step back, one step forward, demonstrable today.
2. A mock recall was completed within the last 12 months, met the site’s time target, and its gaps were actioned.
3. Recall procedures name the team, roles, contact lists, and regulatory notification steps — and the contacts are current.

—

Use this checklist as your starting point, then make it yours: add your scheme’s clauses, your customers’ requirements, your plant’s known weak spots. The best internal audit program is the one that finds your problems before anyone else does.