Compliance Monitoring and Safety Management: Where the Line Actually Falls

Two professionals in a control room reviewing compliance and safety data, with aircraft visible through the windows, highlighting the dual functions of aviation safety management.

Every approved organisation runs one management system with two functions inside it, and the two are not versions of each other. One asks whether the organisation did what the rule and its own manual require. The other asks what could hurt it, and whether the defences it believes in are actually holding. They share an accountable manager, a set of data and often the same corridor — and when they are quietly merged, the failure is always in the same direction: the hazards that break no rule stop being recorded at all.

One system, two functions, one accountable manager

The European construction is explicit. Part-ORO for air operators, Part-145 for maintenance organisations, and the equivalent provisions for continuing airworthiness and air navigation service providers all require a management system, and all describe it as containing both a safety management component and a compliance monitoring function. The accountable manager holds the whole thing, has the authority and the money, and cannot delegate the accountability. Below that sit a safety manager and a compliance monitoring manager, and the rules allow a small organisation to combine those roles subject to the authority’s agreement — a concession on resourcing, not a statement that the functions are the same.

The international foundation is split across two places, which is part of why the confusion persists. Safety management lives in ICAO Annex 19, with its four components: safety policy and objectives, safety risk management, safety assurance, and safety promotion. The compliance side has no single annex — it is the sum of every requirement in Annexes 1, 6, 8 and the rest, and the state’s obligation to ensure that its operators meet them.

The US arrangement is different in shape and worth knowing if you work across both. Safety management sits in 14 CFR part 5; the internal audit tradition arrives through the internal evaluation programme and, on the maintenance side, through the continuing analysis and surveillance system required of part 121 certificate holders. The functions exist in both systems. The way they are packaged does not transfer, so do not assume a European organigram satisfies a US certificate or the reverse.

The two questions

Strip away the vocabulary and each function is a single question asked repeatedly. Everything else — the audit programme, the hazard register, the indicators, the review board — is machinery for asking it.

The two functions side by side
Compliance monitoring Safety management
The question Did we do what the rule and our own manual require? What could hurt us, how badly, and are our defences working?
Direction in time Backward — evidence of what happened Forward — what has not happened yet
Nature of the answer Binary: conforming or not Probabilistic: a risk with a likelihood, a severity and a set of controls
Reference against which it is judged The requirement and the approved procedure The organisation’s own risk tolerability criteria
Main inputs Audit and inspection, records, sampled evidence Occurrence reports, flight data, operational monitoring, change proposals, external data
Output A finding, a root cause, a corrective action, a closure date A hazard, an assessed risk, a mitigation, an owner, a performance indicator
What ends the matter Evidence that the non-conformity no longer exists Evidence that the risk sits inside tolerability and stays there
Who may not perform it Anyone auditing their own work or their own area Nobody is excluded — but the data has to arrive freely

Compliance monitoring checks if rules are followed, while safety management looks ahead to prevent future risks. They use different tools and answers, but both are essential for safety.

An airline finds a missing signature on a training record. Compliance flags this as a non-conformity. But if this happens often, safety management steps in to check if crews are properly qualified, even if the rule isn’t technically broken.

The last two rows carry most of the practical difference. A compliance finding closes when the gap is gone. A safety risk never closes in the same sense: it is held at a level, monitored by an indicator, and reopened when the indicator moves. An organisation whose safety register is full of items marked « closed » is usually running a compliance register under a different name.

Where a finding actually lands

The clean test is to take real observations and ask which function owns each. Most fall to one side only, and the ones that fall to both do so for different reasons and produce different work.

Eight observations and the function that owns them
Observation Compliance Safety Why
A recurrent training record is missing the instructor’s signature Yes Only if the pattern is systemic A records non-conformity. It becomes a safety matter when it stops being possible to know who is qualified.
A crew flew with an expired proficiency check Yes, and near the top of the scale Yes Both a serious non-conformity and evidence that a barrier the operator relies on failed silently.
Rosters are inside every flight time limit, and crews report arriving fatigued on a specific rotation Nothing to find Yes, and urgently The limits are an outer boundary, not a fitness certificate. Nothing is breached and the risk is real.
The same minimum equipment list item has been deferred eleven times in three months, each deferral applied correctly Clean every time Yes Perfect compliance is precisely what conceals it. Only a trend view sees the reliability problem.
A legal but ambiguous chart depiction at one aerodrome produces repeated level busts Nothing to find Yes No requirement is breached by anyone. The hazard is in the interaction, not in the conformity.
A new ground handling contractor at an outstation, contract compliant, no oversight data yet Compliant on paper Yes — a change to be assessed before it bites Management of change exists for exactly this: a decision whose risk profile is unknown, not non-conforming.
An approved procedure that is routinely impossible to follow in the time available Yes — every application of it is a non-conformity Yes The compliance finding is real; the safety question is why the procedure was written that way and what crews do instead.
A stand whose taxi time to the runway consumes most of the holdover time in snow Nothing to find Yes Compliant procedure, local geometry, seasonal exposure. No audit programme is built to see it.

The hazards that break no rule

Four rows in that table have nothing for compliance monitoring to find, and they are not exotic. They are the ordinary content of an operational safety register: fatigue inside the limits, deferral trends, chart and procedure ambiguity, contractor changes, aerodrome peculiarities, seasonal exposure, workload concentration at particular points in the flight.

The reason they exist is structural. A regulation is written to be enforceable across a whole industry, which means it must be general, stable and verifiable from evidence. That combination guarantees it cannot describe every dangerous configuration of a specific operation on a specific day. The gap is not a defect in the rules — it is the necessary consequence of what rules are, and it is the entire reason a second function was invented and made mandatory alongside the first.

The trap. A safety register whose entries all correspond to a regulatory requirement is not a safety register. It is an audit programme with different column headings. The proportion of hazards in your register that breach no rule is the single quickest diagnostic of whether the two functions are genuinely separate — and in a healthy operation that proportion is large.

Safety assurance is the joint, not the merger

Between the two functions sits safety assurance, the Annex 19 component that most often gets used to justify collapsing them. It does draw on audit output — but that is one input among several, and treating it as the whole is where the collapse begins.

Assurance asks whether the controls the organisation claims to rely on are working. Some of those controls are procedures, and audit is a good way to test whether they are followed. Others are technical, behavioural or organisational, and audit says nothing about them: whether a stabilised-approach gate actually produces go-arounds, whether a checklist is read or recited, whether the reporting system receives what it should. The evidence for those comes from flight data monitoring, from occurrence reports, from line observation, from talking to people.

So assurance consumes compliance data and is not made of it. Where an organisation gets this right, the audit programme is one feed into the safety review, alongside the report count, the flight data exceedances, the change log and the indicator set. Where it gets it wrong, the safety review reads out the audit results, everybody nods, and the meeting has learned nothing that was not already written in a procedure somewhere.

What a merged system deletes

Merging can happen in either direction, and both are damaging, but they fail differently.

Symptoms of a merged system, and what each one hides
Symptom Direction of the merger What disappears
Every hazard in the register cites a requirement Safety absorbed into compliance The whole class of hazards that break no rule
Risk assessments are completed by checking the procedure against the rule Safety absorbed into compliance Likelihood, severity and the actual state of the defences; the assessment becomes a conformity statement
Corrective actions are overwhelmingly « crew reminded », « notice issued », « procedure re-emphasised » Either Root cause. Communication closes a finding and changes nothing measurable
Occurrence reports are handled as findings, with actions attached to individuals Safety absorbed into compliance The reports themselves, within a few months — and with them the only view of what actually happens
A confirmed regulatory non-conformity is risk-assessed as tolerable and left open Compliance absorbed into safety The requirement. Risk tolerability is not an alternative to a rule and cannot authorise a departure from one
The audit schedule and the safety action group agenda are the same list Either Anything not written in a procedure, since nothing else is on the agenda
Indicators are all lagging counts of findings Safety absorbed into compliance Any warning that arrives before the event

The fourth row is the expensive one. An occurrence report handled as an audit finding attracts a corrective action, and corrective actions attach to whoever is nearest. Nothing in that has to be punitive in intent for reporting to fall away: it is enough that filing a report reliably produces work and attention for the filer. The European occurrence reporting framework and the just culture principles behind it exist to hold that door open, and they only work if the organisation keeps the report channel separate from the finding channel in practice, not merely in the manual. This is the point where operators most often discover that their written system and their real one diverged years ago, and where honest comparison between operators is more useful than another internal review — the sort of exchange the FL330 professional forums exist for.

Independence, and what it is not

Compliance monitoring must be independent of the activity it monitors. That is a narrow requirement and it is routinely over-read.

What it means: the person auditing flight operations cannot be the person who wrote or manages the flight operations procedure being tested. Auditing your own work produces evidence with no value, and the rule closes that loop.

What it does not mean: isolation from the operation. An audit function that never sits in a briefing, never rides a sector, never walks a hangar floor can only compare documents to documents. It will confirm that the manual matches the regulation and never notice that neither matches the day. The most useful compliance findings almost always come from someone who understood the work well enough to know where to look, and independence was preserved because they did not own that work — not because they were unfamiliar with it.

The same distinction applies to grading. When the authority raises findings it grades them, with the most serious category reserved for non-compliance that lowers safety or seriously hazards flight safety, and immediate action available against the approval. Internal findings do not have to use that scale, but they need some scale, and it must be about the requirement rather than about the risk — otherwise the two functions have merged inside the finding itself.

The right habit. When something is raised, ask two questions in sequence and never skip the second. First: does this breach a requirement or an approved procedure? Second, independently of the answer: does this tell us something about a risk we hold? A « no » to the first and a « yes » to the second is not a dead end — it is the most valuable input the system receives, and it needs somewhere to go.

Indicators that catch the drift early

Merger is gradual and it is comfortable, because a merged system produces tidy paperwork. A few measurements make the drift visible before an audit does.

Two men in high-visibility vests analyze data on a computer monitor in a control room.
Data-driven monitoring ensures compliance and safety in operations.
  • The share of open hazards with no regulatory reference. Near zero means the register is an audit list.
  • Reports per thousand sectors, split by source. A falling crew report rate alongside a stable finding count is the merger signature; so is a cabin reporting rate far below the flight deck’s.
  • The proportion of corrective actions consisting only of communication or training. High means root cause analysis is not happening on either side.
  • Repeat findings. The same non-conformity in successive audit cycles is a closed action that was never effective, and effectiveness checking is what should have caught it.
  • Change proposals that reached a risk assessment before implementation. Management of change is the function’s forward edge; if changes are only assessed after they have been running for a quarter, the system is reactive whatever the manual says.
  • Time from report to feedback to the reporter. Not a compliance metric at all, and the best single predictor of whether reports will keep arriving.

None of those require new software or a bigger department. They require the two functions to keep separate registers, separate closure criteria and separate meetings, so that the numbers above can be produced at all. An organisation that cannot generate them has already answered the question — the two functions are one, and the class of hazards that violates no rule is not being seen by anybody. Further operational material on how these functions are set up in practice sits on FL330.

Commentaires

Leave a Reply

Découvrez l’écosystème FL330

FL330, c’est toute l’aviation professionnelle — explorez nos autres espaces.