Audit Execution: Evidence Gathering, Interviewing, and Sampling
This domain tests how you collect evidence that will survive challenge: verifiable, relevant, representative, and recorded so another person could re-check it. Expect questions on question technique, traceability exercises, and sampling risk.
What counts as evidence
Records and documents, observation of activity and conditions, interviews, and physical evidence (product, equipment, environmental results). Interviews alone are the weakest form; triangulate - a claim that "sanitizer is always checked" should be confirmed by observation and by the log. Evidence must identify what, where, when, and which record (document number, lot, date), not a vague impression.
Interviewing
Ask open questions first ("walk me through what you do if the metal detector alarms"), then closed questions to confirm. Avoid leading questions ("you do calibrate daily, don't you?"). Interview at the workstation, not the conference room, and talk to the person doing the task, not only the supervisor. Check understanding: ask them to demonstrate.
Sampling
Audit sampling is judgemental: choose by risk (new lots, changeovers, night shift, prior failures), not only the neat folders. State the sample in the report ("5 of 40 lots"). Product sampling for testing may follow statistical plans such as ISO 2859-1, but that is separate from audit sampling of records. A sample that misses the failure is sampling risk; a sample selected by the auditee is not a sample.
Worked example
Traceability: lot 240501 produced 12,000 units. Tracing forward finds 11,400 shipped, 300 in the warehouse, 100 retained - 200 units (1.7%) unaccounted for. Reasoning: the mass-balance gap exceeds the site's stated 0.5% allowance, so this is a nonconformity in traceability, not a rounding issue. Trace backward to raw receipts to see whether the gap originates in production yield, rework, or destruction records before writing the finding.
Traps that cost marks
- Accepting "we always do it" without observation or record.
- Letting the auditee choose the sample.
- Sampling only conforming records; excluding rework, deviations, and complaints.
- Hearsay recorded as objective evidence.
- Using statistical-sampling language for what was a judgemental sample.