Aviation Audit Records: How to Structure the Report Pack and Findings

Here is a short audit record. Everything in it is fictional.

AUD-EXAMPLE-2026-014 · ExampleMRO · Line station EXA · 12 March 2026

Scope. Tool control at line station EXA: loan tool issue and return, calibration status of torque-measuring tools, quarantine of unserviceable tools. Out of scope: stores receiving and base maintenance tooling.

Criteria. ExampleMRO procedure QP-EXAMPLE-07 "Tool control and calibration", revision 3.

Auditee roles. Shift supervisor, stores clerk. Station engineer not available (leave); follow-up raised.

Interview notes. Shift supervisor stated overdue tools are "usually pulled by stores at the start of the month." Stores clerk stated the calibration register is reviewed "when someone has time"; showed the register, last review entry dated 20 January.

Observation. A torque wrench past its calibration due date remained available for issue and is listed on three work orders after that date.

Draft finding — for auditor review, not issued. Requirement: QP-EXAMPLE-07 rev. 3 requires tools past their calibration due date to be withdrawn and quarantined. Evidence: S2, S3. Gap: TW-EXAMPLE-07 was not withdrawn after its due date and was issued on three work orders. Classification: for the auditor under ExampleMRO's finding procedure. Immediate question for the organisation: what assessment is needed for the tasks on the three work orders?

Notice what the draft finding does not say. It does not say "tool control is ineffective" or that the three aircraft were released with incorrectly torqued fasteners. It says the tool was issued. Whether that affects the tasks is a question for the organisation to assess, and the record asks it.

That is the whole craft of an audit record: every sentence can be traced to something sampled, and nothing is made stronger than the evidence. The rest of this article takes the pack apart field by field.

1. Identification

Reference, organisation, location, dates, auditor, auditee roles and the audit programme item it belongs to. It sounds administrative, but months later the first question from a reviewer is often "which audit was this, and was it the planned one?" A record that cannot be placed in the programme cannot show coverage. How that programme is built, with a sample 12-month plan, is in the audit programme guide.

2. Scope statement

Weak: "Maintenance audit."

Strong: the scope in the example above, including what was out of scope.

A scope that says only "maintenance audit" does not show whether the auditor checked line maintenance, stores, certifying staff authorisations, defect control or the interface with planning. Stating what was excluded matters as much. If stores receiving was out of scope, nobody should later read the record as evidence that stores receiving was fine.

3. Criteria

Weak: "EASA requirements and company procedures."

Strong: the named procedure and its revision, or the requirement reference, as used on the day.

Criteria are what the evidence is measured against. If the reference is not yet confirmed, the record says so: requirement not yet confirmed. It does not guess one. A finding against the wrong revision of a procedure is a finding the auditee can reject.

4. Sampling table

Weak: "Records checked, some issues found."

Strong: a row per sample, showing the record, the population, the sample size, how items were selected and the result.

This is the field that most often decides whether a finding survives. "Two of twenty sign-out entries lacked a return signature" supports a very different conclusion from "tool control failed." If the auditor sampled two work orders, the record should not imply the whole system failed. Recording the selection method, whether every third entry or auditor-selected, also shows the sample was not chosen to prove a point.

Give each sample a short reference (S1, S2…). Findings can then cite evidence by reference instead of retelling it.

5. Interview notes by role

Record the role, date, what the person stated, what records they showed and any follow-up. Use roles rather than names where the organisation's practice allows.

The discipline is keeping four things apart. From a planning audit:

• "Planner stated the weekly check is normally done on Friday" is an interview note.
• "Three sampled weekly planning checks for May did not contain sign-off" is objective evidence.
• "The planning control may not be consistently completed" is interpretation.
• "Raise a finding" is a decision for the auditor and reviewer.

When those categories are mixed, the report becomes hard to defend. When they are separate, the reviewer can see exactly what each conclusion rests on. Objective evidence vs opinion in aviation findings covers the method in more depth.

6. Observations

An observation states what was seen, tied to sample references, without the finding's conclusion. Not every observation becomes a finding. Some are recorded as concerns or opportunities to improve, and many organisations also record positive observations. The S4 row above, a quarantine log with both entries tagged and signed, is one worth writing down: it shows the control working where it was used.

7. Findings

A finding needs five parts:

• requirement: the criterion that applies;
• evidence: what was sampled or observed, by reference;
• gap: how the evidence differs from the requirement;
• classification: decided by the authorised person, not pre-filled;
• status: draft, reviewed, approved or issued, according to the company process.

Weak: "Tool control poor."

Strong: identify the tooling procedure, state which tool records or shadow-board checks were sampled, describe the missing calibration label or sign-out entry, and leave classification to the auditor or quality manager.

"Tool control poor" is not a finding. It is an opinion without an address. The auditee cannot answer it, and the root cause analysis will be as vague as the finding. Also keep corrective action out of the finding: "Stores should review the register weekly" is a solution, and it belongs to the auditee's response, not the auditor's statement. Can AI draft aviation audit findings safely? sets out the guardrails for the drafting step.

8. Open questions and follow-ups

Anything the auditor could not close on the day: the station engineer not interviewed, a record requested but not produced, a reference to confirm. An open question is honest. A gap papered over with assumed evidence is not.

9. Response and CAP evidence expectations

State what the auditee must return and by when: containment, root cause, corrective action, owner, due date, and the evidence that will show completion. Setting the closure evidence expectation at issue saves an argument at closure. Better corrective action plans covers the review side.

Labels for evidence that is not ready

A draft audit record is safer when its weak points are labelled, not smoothed over. A reusable set:

• requirement not yet confirmed
• sample size not stated
• evidence unclear
• interview note only
• classification requires human decision
• draft for audit review

These labels are not bureaucracy. They are how an uncertain note avoids becoming a confident but unsupported finding. A reviewer who sees interview note only against a draft finding knows exactly what to ask for before it is issued.

The wording must also stay traceable to the notes. If a draft finding says two authorisation records were expired, the record references must be available to the auditor in the controlled evidence trail. If it cites a procedure, the document and section need confirmation.

Where AI fits, and where it stops

Most of this structure is preparation, which is where AI is useful. From rough notes, it can build the sampling table, sort interview statements from record evidence, apply the labels above, and check each draft finding for a requirement, evidence references and a gap without a hidden corrective action. It can ask for the missing scope exclusion or the procedure revision instead of inventing one. AI for aviation audits covers the wider audit workflow, before and during fieldwork.

It stops at judgement. AI does not choose the sample, weigh the evidence, classify the finding or issue the report. The auditor reviews every line, and the approved output is entered, attached or retained through the organisation's controlled process.

In Avioverse, Metis assists with preparation, while people operate the Audits workflow. Each audit run has Plan & team, Checklist, Report and Findings. The plan records objective, scope, criteria and sampling, and findings can be raised during fieldwork. Team audits record allocated roles, an accountable finding handler and an independent reviewer. One internal audit followed from checklist to closed finding is in Aviation Audit Software: From Checklist to Closed Finding.

Avioverse Audits preserves issued reports and their history. Reopening a report preserves the previous issue, and communicated findings retain their original statement. Accepting corrective actions does not close a finding; an authorised reviewer closes it explicitly, with a recorded risk decision. For information on our controls and support for organisational assessment, see AI assurance.

A record built this way does not remove accountability. It lets the accountable auditor see the work clearly before signing, issuing or entering anything into the controlled system.