Understanding Audit Techniques

Ask a supervisor if allergen cleaning is done properly and you will almost always hear yes. Watch a changeover, read the swab results, check the cleaning record against the production schedule, and interview the operator who does the work, and you may reach a different conclusion. Audit techniques exist to close the distance between what people say and what evidence shows. ISO 19011 calls this the evidence-based approach: conclusions rest on verifiable audit evidence, not on confidence, seniority, or first impressions.

Four techniques do most of the work in an ISO 22000 internal audit: interviewing, observation, document and record review, and tracing. Skilled auditors combine them on the same control, because each technique catches what the others miss. A record can be complete and false. An interview can be honest and mistaken. Triangulation, testing one claim through several sources, is what turns information into evidence.

Interviewing that produces evidence

Interviews explain how work is understood by the people who do it. Use open questions that invite description: walk me through what happens when a delivery fails its temperature check. Follow with focused questions on limits, actions, and records. Avoid leading questions that supply the answer, and avoid turning the interview into a test of memory under pressure. People who fear blame give safe answers. People who feel the audit is fair describe real practice, including the workaround everyone uses and nobody has written down.

Interview at more than one level. A manager describes the intended system. A supervisor describes how it is organised. An operator describes how it runs at 2 a.m. when the line is behind. Differences between those accounts are findings in the making. Note who said what, and verify important claims against records or observation before treating them as evidence. An interview alone rarely proves conformity or nonconformity; it directs you to proof.

Observation of work as it happens

Observation shows practice directly: handwashing on entry, segregation of raw and ready-to-eat product, thermometer use, label checks at changeover, how a deviation is handled in the moment. Observe normal work without directing it. If you must ask someone to demonstrate a task that is not occurring, record that it was a demonstration, because demonstrated behaviour and routine behaviour are different evidence.

Watch for the gap between procedure and practice, and be precise about it. Writing that hygiene was poor proves nothing. Writing that at a stated time, in a stated area, an operative handled raw product then ready-to-eat product without handwashing or glove change, against a named procedure step, is evidence a site can act on. Time, place, and reference turn observation into a finding.

Reviewing documents and records

Documents state intent; records claim history. Review documents for currency, approval, and consistency with the hazard control plan. Review records for completeness, plausibility, and pattern: entries in identical handwriting across shifts, monitoring values that never vary, corrections overwritten without explanation, or checks recorded when the line was not running all undermine confidence far beyond one form.

Sample deliberately. Combine random sampling, so any record might be chosen, with targeted sampling aimed at risk: dates around a deviation, a complaint, a new starter, or a maintenance event. State sample sizes in your notes. A conclusion drawn from three records should not be written as if it covered three thousand.

Tracing: following one thread end to end

Tracing is the strongest technique for system audits. Forward tracing follows a raw material lot from intake through production to finished product and distribution. Backward tracing starts from a finished product or a complaint and works back to ingredients, controls, and records. Vertical tracing stays with one product or process and tests every system element attached to it: specification, hazard assessment, control categorisation, monitoring, verification, training, calibration, and traceability.

Tracing exposes interface failures that department-by-department auditing misses. A specification updated by technical staff but not by purchasing, a flow diagram missing the step where a hazard enters, a calibration gap on the one thermometer used for a critical limit: each appears when you follow a single thread and pull. Choose tracing threads from risk: high-risk ingredients, allergen changeovers, products involved in complaints, and lots produced during deviations reward the effort most.

Testing, reperformance, and professional scepticism

Where you are competent and authorised, independent checks strengthen evidence: taking a temperature with a calibrated instrument, watching a metal detector test with the correct test pieces, or asking for a traceability exercise timed against the site’s own target. Reperformance, repeating a check the site performs, tests both the method and the result. Keep proportion: internal auditors verify the system, they do not take over the site’s monitoring duties.

Professional scepticism runs through all techniques. Accept nothing important on a single source, stay alert to records that look too perfect, and follow inconsistencies calmly until they resolve or become findings. Equally, record good evidence when you find it. An audit that only hunts faults teaches auditees to hide them. Balanced evidence, gathered by sound technique, is what makes conclusions defensible at the closing meeting and useful afterwards.

Key points

• Conclusions must rest on verifiable evidence, gathered by combining interviews, observation, record review, and tracing.
• Open interview questions, asked at several levels, reveal how work is really understood and done.
• Observations become evidence when recorded with time, place, and the requirement they relate to.
• Record review should mix random and risk-targeted sampling and look for patterns, not just completeness.
• Tracing one lot or process end to end exposes interface failures other techniques miss.
• Triangulate important claims through more than one source before treating them as findings.