Recurrent Training: Prescribed Programme or Data-Driven Programme

Salle de briefing vide, tableau blanc nu

The simulator session has opened for years on the same engine failure at takeoff, followed by the same go-around and the same classic approach. Meanwhile, the fleet’s most recent serious events have involved a stabilized approach caught too late, an automation mode misread by the crew, and a decision taken well after the moment when it would have mattered. Nothing in the syllabus links these two lists. That is exactly what separates two logics of recurrent training: guaranteeing that content has been covered, or guaranteeing that crews train for what actually breaks.

Where the content comes from: the text or the data

A recurrent programme always answers two distinct questions: what must be worked on, and how is it to be demonstrated. The classic regime settles the first question in the text. The rule, expanded through the authority’s approval, sets out subjects and a cadence: training and proficiency checking, line checking, safety and survival equipment, human factors woven through the whole, all of it required to cover, over a multi-year cycle, the major system failures and their associated procedures. On this basis the operator builds a delivery plan: who does what, when, and in which simulator device.

The data-driven regime moves the boundary one step further. The text stops naming the subjects and starts describing a process instead: collect operational data, analyse it, derive the period’s themes from it, work them in the session, measure the effect, start again. What is approved is no longer a content but a method, together with the conditions that make it credible. The operator inherits the choice of content, and with it the burden of proving that the choice was the right one.

The dividing line is not geographical, contrary to what the usual presentation suggests. The European body of air operations rules carries the classic regime and, alongside it, the Alternative Training and Qualification Programme, then Evidence-Based Training. US federal regulation carries the subparts devoted to crew member training and qualification and, alongside them, the Advanced Qualification Program, born from a special rule that has since become a subpart in its own right. On both sides, the data-driven route is optional, approved case by case, and reversible. What is actually being compared is therefore not Europe and the United States, but two ways of assigning responsibility for content.

Open chain and closed loop Two sequences compared. At the top, prescribed content runs from the applicable text to the mandatory items, then to a uniform session and a check whose outcome is binary; the final arrow leads back nowhere. At the bottom, data-driven content runs from operational data to analysis, then to an adapted session and to a competency grading that becomes an input again: a return line closes the loop. prescribed content text applicable items mandatory session uniform check pass or fail no feedback data-driven content data operational analysis and selected themes session adapted grading by competency grading becomes data again

The whole difference lies in one arrow: the one that comes back.

In the prescribed regime, the check’s outcome does not alter the content of the next session. In the other, the grading is itself an input, which makes its quality just as critical as that of flight data.

The prescribed programme follows a fixed list of training topics, while the data-driven programme adjusts based on real flight data and incidents. The first ensures everyone covers the same material, but may not address current risks. The second focuses on real-world issues but requires careful data analysis and instructor consistency.

An airline operating Airbus A320s notices repeated unstable approaches in their flight data. In a prescribed programme, crews would still train for standard engine failures. In a data-driven programme, the simulator session would focus on approach stability, using scenarios based on actual flight data, and adjust training based on crew performance.

What prescription guarantees — and what it does not

It first guarantees coverage. Over the cycle, every crew will have seen the flight-control failure, the depressurisation, the evacuation, the cargo-hold fire, regardless of what these events actually represent in the fleet’s real life. It also guarantees comparability: two pilots from two different operators have worked the same content and demonstrated it within the same tolerances, which makes the record legible to a recruiter, a foreign authority, or a licence-conversion service. Lastly, it guarantees low-cost oversight: checking coverage is a tick-box exercise, well within reach of a documentary audit. For an operator with neither the volume nor the analytical means, this is a real protection, not a stopgap.

What it does not guarantee is relevance. The list is updated at the pace of amendments, not at the pace of the fleet, the network, or the automation actually fitted in the flight deck. Between two revisions the gap widens without anyone making a mistake: everyone correctly applies a programme that is ageing. The second flaw is more insidious. Content known in advance ends up being learned as such. The crew knows what is coming, prepares for it, passes it; the check then measures preparation rather than capability. And the most useful information — the margin that remained, the number of retries, the competency that gave way first — is recorded nowhere. The outcome is binary, and binary data does not feed back.

What a data-driven programme produces

It begins with an analytical exercise that does not look like pedagogy. Recorded flight data, event reports, line observations, technical reports and grades from previous sessions are cross-checked to extract the situations that actually degrade the conduct of flight at that particular operator. To this local material is added shared material: theme tables established by aircraft technology generation, which carry the accumulated experience of a worldwide fleet and spare each operator from rediscovering alone what others have already paid to learn.

The second change is the unit of measurement. What is assessed is no longer a manoeuvre but a set of observable competencies: application of procedures, communication, manual aircraft control, automation management, leadership and teamwork, problem-solving and decision-making, situation awareness, workload management — to which revisions of the framework have added knowledge application. Each competency is graded on a multi-level scale, one threshold of which marks the expected performance, based on observable behaviours described in advance. A session therefore no longer produces a verdict but a profile, and a profile is compared with that of the previous period.

The third change is the most visible one for the crew: the session adapts. An evaluation phase opens the session in the form of representative scenarios; what it reveals determines what will be worked on next, within the same session. Discovering a weakness stops being a deferred sanction and becomes immediate instruction instead. And the grading produced at the end of the session feeds back into the analysis: it is this grading, aggregated across the fleet, that will shift the following period’s themes.

The session does not run in the same order

The order of blocks within a recurrent session Two timelines compared. In the prescribed programme, the session runs through briefing, mandatory manoeuvres, check, then debriefing: the assessed part comes at the end and is used to decide pass or fail. In the data-driven programme, the evaluation phase comes first, decides the content of the manoeuvres that follow, then comes a line scenario and a debriefing. prescribed programme: content fixed before the session briefing mandatory manoeuvres check debriefing data-driven programme: content fixed during the session evaluation phase adapted manoeuvres scenario debriefing the content of what follows is decided here progress of the session

The evaluation changes place and, in the same move, changes function.

Placed at the end, it decides pass or fail; placed at the start, it dictates the rest of the session. The central block no longer has content fixed in advance, which shifts the burden onto the instructor’s preparation and onto the debriefing.

Moving the evaluation to the start of the session is not a scheduling adjustment, it is a change of function. Placed at the end, it certifies and no longer teaches anything: what is observed arrives too late to be worked on that same day. Placed at the start, it dictates. The practical consequence is that the content of the central block is no longer known when the crew walks through the door, which rules out targeted preparation and restores a share of unpredictability to the exercise.

The instructor’s role is changed even more than the crew’s. The instructor no longer runs through a written scenario ticking off tolerances: they observe, grade, choose from a pre-analysed catalogue the exercises that address the competency at issue, then conduct a debriefing that becomes the main teaching moment. This calls for different instructor training, a qualification that must be kept current, and a periodic check of grading concordance between instructors. Without this last piece, the chain breaks at the exact point where it produces its data.

From the manoeuvre to the competency: what the evaluation records

Two recurrent-training regimes, point by point
What is at stake Prescribed programme Data-driven programme
Origin of content the text, then the approval that expands it the operator’s data, supplemented by shared tables for the aircraft generation
What the authority checks that the items have been covered over the cycle that the selection and review method exists, works, and is actually applied
Unit assessed the manoeuvre, performed within tolerances the competency, observed within a scenario
Timing of the evaluation at the end, to certify at the start to diagnose, then continuously
Outcome of a session pass, or repeat a competency profile, which becomes the content of what follows
Time needed to correct the programme that of the text amendment that of the internal review cycle, applicable to the current period
Main cost keeping an ageing list relevant standardising instructors and maintaining the data chain
Comparability between operators strong, the reference framework is common weak, the profile can only be read together with the programme that produced it
Failure mode crews train very well for what no longer breaks only what has been instrumented is seen, and it is graded poorly

The right-hand column does not say that the data-driven programme is more lenient. It remains capable of saying no, and says it differently: a competency below threshold triggers targeted supplementary training, followed by re-assessment; if the profile does not recover, the consequence reverts to the familiar one, namely suspension of the exercise of privileges pending demonstration. The difference lies in the fact that the decision rests on a trajectory, not on a single day: a crew that fails a manoeuvre under a prescribed regime and retakes it in a follow-up session leaves with a record identical to their neighbour’s, whereas the profile keeps a trace of the episode.

This trace is also what legitimately worries crews. A richer record is a record more legible to others. Serious programmes address this issue head-on, by separating individual data — used for instruction and kept under restricted access — from aggregated, de-identified data, which alone feeds back into programme analysis. The point deserves to be checked in the operator’s documentation rather than assumed: it is a condition for the sincerity of the grading, and hence for the validity of the whole edifice.

The bedrock that does not move

A data-driven programme cannot select something that has never happened. This is a limitation of method, not a failure of execution: the rare, catastrophic event appears in no local statistics, and the only way to keep it in the programme is to prescribe it. This is why a bedrock remains whatever the chosen logic — safety and survival equipment, upset prevention and recovery, exercises tied to events one only ever wants to see in the simulator — and why the programme’s recognition as equivalent to a check, for licensing purposes, remains written into the approval.

The same reasoning produces a counter-intuitive effect that must be kept in mind before reading any theme table: a subject that has been well handled disappears from the data. It does not disappear because it has stopped being dangerous, but because the training is doing its job. A mechanical steering of themes by event frequency then ends up removing from the programme exactly what justified it, only to see the subject return a few years later. Serious methods therefore keep a baseline weighting alongside the situational weighting, and an operator unable to explain how it handles this case probably has no method at all.

Where each logic breaks down

What each source captures of the real job, and what it misses
Source What it captures Blind spot
Recorded flight data the measurable deviation from a parameter defined in advance the reasoning that led to the deviation, and everything no one thought to instrument
Event reports what a crew judged worth reporting, and knew how to put into words what ended well, and whatever has no box on the form
Line check the conduct of an ordinary flight, under observation the difficult day, which cannot be scheduled in advance
Simulator gradings competency under pressure, within a chosen scenario the scenario nobody wrote, and the severity gap between instructors
Technical reports and maintenance feedback the actual failure, its frequency and its context the way the crew actually handled it
Generation theme tables the accumulated experience of a worldwide fleet what is specific to the network, the airfields and the operator’s culture

The table makes the real risk of the data-driven regime visible: it does not see the job, it sees its sensors. Flight data only picks up deviations someone decided to instrument, reports only carry what someone dared to write down, and the line check runs on a chosen day. A fleet can therefore display a reassuring dashboard while training beside its real risk, with the added, dangerous conviction of having acted on evidence.


The trap.

In a data-driven programme, an instructor’s grading is not an end-of-session impression: it is the primary sensor. As soon as the internal culture pushes toward grading everyone at the expected level — out of kindness, to avoid conflict, or because a low grade triggers an administrative burden — the loop closes on noise. The operator then finds itself in the worst position: no imposed coverage left, and an analysis that measures nothing. A grade distribution where almost no one is ever below threshold is not a good sign; it is a sign of failure.

What the authority approves, and what it can withdraw

Which regime for which operator Two-axis matrix. On the horizontal axis, the richness of available operational data; on the vertical axis, the ability to standardise instructors. The prescribed programme occupies the box of scarce data and heterogeneous instructors; the full data-driven programme occupies the opposite box; a mixed programme suits cases where instructors are standardised but data is scarce; the box of abundant data graded by heterogeneous instructors produces an unusable loop. mixed programme themes drawn from generation tables full programme data-driven prescribed programme guaranteed coverage unusable loop non-comparable gradings instructors standardised instructors heterogeneous scarce data abundant data available operational data

It is not fleet size that decides, it is the ability to grade the same way.

Abundant data used by non-standardised instructors produce a profile that means nothing; scarce data with standardised instructors remain usable by drawing on shared tables.

Approval of a data-driven programme is not the validation of a syllabus. It bears on a system: data sources and their quality, the method for selecting themes, the qualification and standardisation of instructors and evaluators, the review cadence, the indicators reported to the authority, and the demonstration that the safety level is at least maintained. It comes with conditions, is monitored over time, and can be withdrawn. This last point has an operational consequence often forgotten when building the case: the operator must remain able to revert to the prescribed regime without interrupting operations, and must therefore keep a current classic programme, compatible simulator slots, and instructors capable of running an old-style check.


The right reflex.

Before discussing content, ask operations management three things: the cycle coverage table, showing what the programme claims to cover and over what period; the dated source of the last theme revision, which must point to an identifiable analysis rather than a meeting; and the result of the last instructor-concordance exercise. These three items reveal very quickly whether the loop is actually turning. If any one is missing, you are looking at a prescribed programme that has simply changed its vocabulary.

What to check for yourself

For a crew member, the most concrete question is not pedagogical but administrative: what is your record worth elsewhere. A competency profile does not travel like a passed proficiency check; what actually circulates is the recognition of the programme as equivalent to that check, and the resulting type rating. Before a change of operator, before a licence conversion, the question must therefore be asked early, to the receiving organisation rather than to one’s own, and the attestations must be obtained in the form it expects. This is also what separates a serious qualification plan from a mere intention: knowing

what your programme gets recognised

before you need it.

Two pilots in a cockpit, one pointing at a flight display while the other observes.
Training ensures pilots adapt to real-time flight scenarios effectively.

For an instructor, the point of vigilance has shifted to their own standardisation: grading has become a technical act that is trained, checked, and corrected. For a safety manager, the check comes down to one question: do the department that analyses events and the one that writes the themes read the same data, at the same pace? Two review cycles running out of sync produce two lists of priorities and cancel out the benefit of the method.

Grey areas remain that no one has closed: from what volume does a fleet produce enough data to steer its own content, how should generic tables be weighted against an atypical network, and by what rules is a competency record read from one operator to another, or from one regulatory system to another. On these three points, practice varies more than the texts do, and

instructor feedback shared between operators

often informs better than a comparative reading of the regulations.

This text describes mechanisms, not a specific situation: the applicable regulation, the exact terms of the approval issued to your operator, and its approved documentation always prevail over what is written here.

Questions fréquentes

Does a data-driven programme eliminate the proficiency check?

It does not eliminate it, it reorganises it. The evaluation stops being an event placed at the end of the session and becomes a phase at the start of the session plus continuous observation, and it is the approval issued to the operator that establishes that the whole is equivalent to the required check. The ability to conclude that a crew member is not up to standard remains intact, with targeted supplementary training and, if the profile does not recover, suspension of the exercise of privileges pending demonstration.

Can a small operator qualify for this type of programme?

Yes, through the mixed route, which places most of the theme-selection burden on tables established by aircraft generation and uses local data only as a supplement. What conditions approval is less the fleet size than the ability to standardise instructors and maintain a documentary review cycle. An operator unable to demonstrate grading concordance has no interest in leaving the prescribed regime, whatever the volume of its data.

What happens to my training record if I change operators?

A competency profile is not transferable as is: it only makes sense when read together with the programme and the scale that produced it. What actually circulates are the check attestations and the resulting type rating, in the form expected by the receiving organisation. Ask that organisation the question before leaving your operator, since obtaining attestations in a specific format is considerably simpler while you are still on staff.

Who decides the content of the session in an adaptive programme?

The evaluation phase decides it, and the instructor applies it within a written framework. The instructor chooses, from a catalogue of pre-analysed exercises, those that address the competency observed as weak, according to rules set by the approved programme. The instructor therefore does not improvise the content: what varies from one crew to another is the selection, not the catalogue.

Why keep mandatory items if data drives the content?

Because a data analysis cannot point to an event that has not yet occurred, and because a well-handled subject disappears from the statistics precisely thanks to the training that addresses it. A prescribed bedrock therefore covers the rare and catastrophic and prevents the programme from emptying itself of what justified it. It also meets a legal requirement, the recognition of the programme for licensing purposes.

How can I tell whether the loop is actually working at my operator?

Three items are enough to tell: the cycle coverage table, the dated source of the last theme revision, and the result of the last instructor-concordance exercise. If the revision points to an identifiable analysis and if the grading distribution is not concentrated on the single expected level, the loop is turning. Otherwise, you are looking at a prescribed programme that has changed its vocabulary.

Commentaires

Leave a Reply

Découvrez l’écosystème FL330

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