Fly What You Can Verify

01

Evidence is what the crew can observe and test

A reassuring statement is not evidence merely because it sounds confident. “The aircraft is configured” is a conclusion; the actual configuration indications, checklist status, flight-path response and any relevant warnings are the evidence used to support it. Likewise, “the automation has it” is an assumption until the crew confirms the active modes, selected targets and resulting trajectory.

Useful evidence has three qualities: it comes from a source suitable for the question, it is current enough for the phase of flight and it can be checked against another indication or expected result. A single display can be accurate while the underlying data entered into it is wrong. Two displays driven by the same erroneous source may agree without providing an independent check.

The crew should distinguish an observation from an interpretation: “vertical speed is increasing” describes a trend; “the autopilot is malfunctioning” assigns a cause. Keeping that distinction visible prevents an early theory from directing attention away from competing explanations.

02

Start each phase with a predicted aircraft state

Monitoring is most effective when pilots know what should be seen next. Before a level-off, the crew expects an appropriate change in vertical guidance and a reduction in vertical speed. After a thrust selection, it expects the commanded thrust and the aircraft's acceleration to agree with the intended operation. On final approach, it expects a converging lateral path, a controlled vertical path, suitable energy and the required configuration.

This prediction gives meaning to indications. A deviation is identified earlier when the crew knows both the target and the expected trend. Monitoring only after an alert or a large excursion leaves less time to recover.

“Recheck” completes the loop. A crew may select a new mode or speed and still fail to obtain the intended result. The corrective action becomes evidence-based only when its effect on the actual aircraft state is verified.

03

A briefing should produce testable expectations

A strong departure or approach briefing identifies the highest-consequence threats and the indications that will reveal whether they are being managed. It describes the intended runway and entry, flight path, automation use, terrain or weather constraints, likely workload peaks and the conditions that require a change of plan.

For example, “the runway is short and wet” becomes operational when the crew confirms the actual runway and intersection, the current runway condition, the performance basis, the intended landing point and the approved decision criteria. “There may be a tailwind” becomes useful when the crew agrees which current wind reports and aircraft indications will be checked, and what consequence a change would have for landing performance.

The briefing should also invite the other pilot to challenge a mismatch. A shared plan can create shared expectation bias; explicit cross-check roles help prevent both pilots from accepting the same incorrect assumption.

04

Takeoff: verify the real runway and aircraft

Takeoff performance depends on the conditions actually present, not just those stored in a calculation. Runway entry, available distance, aircraft mass, configuration, thrust setting, environmental inputs and limitations must match the approved performance data. The runway sign and chart, the cleared route, the performance calculation and the aircraft setup are separate parts of one safety case.

An intersection departure illustrates the value of independent evidence. A crew may have correctly calculated full-length performance but then accept an intersection clearance. The runway in front of the aircraft is shorter than the runway in the calculation. A deliberate comparison of clearance, physical entry point and performance basis exposes the mismatch before thrust is applied.

During the roll, the pilots monitor the actual response through the indications and callouts specified by their SOP. Acceleration, thrust, airspeed agreement, configuration messages and runway cues can reveal that the aircraft is not behaving as expected. These observations must feed the operator's approved rejected-takeoff decision logic; they do not justify inventing a new speed threshold during the roll.

05

Automation: test the mode against the path

A selected value is an intention. An active Flight Mode Annunciator (FMA) mode shows what guidance is engaged. The aircraft's actual path shows the result. These are three different pieces of information, and none should silently substitute for another.

After a mode change, the crew should apply the operator's callout and cross-check standard, then observe whether the aircraft responds in the expected direction. If an altitude capture is expected but vertical speed remains high, the discrepancy deserves attention while there is still margin. If a lateral path diverges after a route modification, re-reading the flight management entry alone is insufficient: the trajectory must be confirmed.

Automation can reduce workload, but only while its behavior remains understandable and suited to the task. When the current mode, target or path cannot be verified promptly, pilots use the approved level of automation and flight-path control that restores a clear, manageable aircraft state.

06

Anomalies: separate fact from explanation

During an abnormal event, the first cue may be incomplete or ambiguous. An airspeed disagreement, navigation shift, unexpected alert or cabin-altitude trend does not immediately reveal its cause. The crew gains time by protecting the flight path, identifying what each source actually reports and using the applicable approved procedure.

Consider a sudden speed split. The evidence may include differing airspeed indications, pitch and thrust, ground speed, aircraft attitude, flight-path trend and alert status. Some of these sources may share dependencies or be affected by the same conditions; none should be promoted to “truth” solely because it appears plausible. The crew must follow the aircraft-specific unreliable-airspeed procedure and use the references it authorizes.

Say what is observed before saying why it happened. “Left and right indicated speeds disagree; pitch and thrust are stable” gives both pilots a common starting point. “It is a blocked probe” may be a hypothesis, but it must not shortcut the approved diagnosis or draw attention away from maintaining control.

07

Approach: recognize a shrinking safety margin

An approach can appear orderly while several margins deteriorate together: speed drifts high, the descent rate increases, configuration is late, the touchdown point moves down the runway or the wind changes. Each deviation may look recoverable on its own. Their combination may make a safe landing unlikely.

The crew should compare actual flight path, energy, configuration, runway condition and landing-performance assumptions with the operator's stabilization and go-around criteria. The purpose is to decide in time, while a normal escape path is still available. A late correction that temporarily centers an indication does not necessarily restore the full safety margin.

If the approach does not meet the applicable criteria, the approved response applies. A go-around is a planned operational option, not a judgment on a pilot's ability. A timely decision uses present evidence rather than hope that the next few seconds will repair the state.

08

PF and PM build independent checks

The Pilot Flying (PF) manages the flight path; the Pilot Monitoring (PM) monitors the aircraft state and system status, calls deviations and supports the operation under the operator's SOP. Both pilots need a shared understanding of the intended path, but they should preserve independent attention to the aircraft's actual behavior.

An effective monitoring call states the discrepancy clearly: “High on path,” “Speed increasing,” or the operator's prescribed callout. The next exchange confirms a corrective action and its effect. Long causal debates while the flight path is deteriorating can delay intervention. If the deviation persists, the crew applies the escalation and control-transfer rules in its SOP.

Monitoring is active work. It includes checking the clearance against the intended route, verifying an FMA change, watching the trend after a correction and speaking up when the evidence conflicts with the plan—even when the other pilot is senior or both pilots were previously confident.

09

Set decision thresholds before workload rises

Crews make better decisions when the conditions for continuation, modification or abandonment are identified before the critical moment. Examples include the approved stabilization gate, runway and performance constraints, fuel decision points, weather limits and the trigger to stop taxi when position is uncertain.

These thresholds come from regulations, aircraft performance, operator procedures and the actual situation. This article does not supply numbers. Its operational lesson is to make the relevant threshold visible in the briefing, monitor the evidence against it and execute the associated action when it is reached.

A threshold loses its protective value if the crew continually moves it to accommodate a worsening picture. When new information invalidates an assumption, the crew should update the plan explicitly rather than silently accepting more risk.

10

Worked scenario: an approach that stops matching the plan

Briefed plan. The crew has reviewed an instrument approach to a wet runway. The current landing calculation and the operator's stabilized-approach criteria support the intended landing. The crew agrees to monitor wind, flight-path capture, energy and the intended touchdown area.

New evidence. After a late vector, the aircraft is above the planned vertical path. Speed remains high while configuration is delayed. A fresh wind report differs from the briefing, and the runway remains wet. None of these facts alone explains the entire situation; together they reduce the available margin.

Cross-check and decision. PF and PM confirm the active mode and actual path, the latest wind and performance implication, then compare the current aircraft state with their approved stabilization criteria. If the criteria are not met at the required point—or the crew cannot establish an acceptable state—the go-around follows the operator's procedure. The crew then rebriefs with updated weather and performance information before another approach.

The key learning is the sequence. The crew does not wait for a runway warning or a dramatic deviation. It uses modest, accumulating indications to make a decision while options remain.

11

Debrief observable crew behavior

EBT is based on evidence about real operational threats and on observable crew competencies. Its most useful cockpit question is not “Did the crew finish the event?” but “What did the pilots notice, say and do as the situation changed?” An instructor might examine whether the pilots anticipated the threat, detected a mismatch, communicated it, managed workload, used the appropriate procedure and confirmed that their action worked.

For the approach scenario, a useful debrief identifies the actual cue, who noticed it, the callout, the agreed decision point and the aircraft state after the decision. “Good situational awareness” is too broad to guide improvement. “PM identified the high-energy trend, stated the deviation, and both pilots compared the state with the stabilized-approach gate” is observable and discussable.

The same discipline helps a crew learn from ordinary line operations. An unexpected mode, a corrected runway-entry mismatch or a timely go-around can reveal a successful defense worth preserving. Evidence supports learning when it describes behavior precisely and avoids guessing at motive.

12

Professional takeaways

1. Brief a prediction. Know the state and trend that should appear next.
2. Observe actual behavior. Distinguish selected targets and written calculations from what the aircraft and environment are showing.
3. Cross-check the source. Agreement between dependent indications may not be independent confirmation.
4. Call the discrepancy. State the observed mismatch clearly before committing to a cause.
5. Act at the approved threshold. Preserve margin by using established SOP and aircraft-specific criteria.
6. Verify the effect. A mode selection or corrective input is incomplete until the aircraft responds as intended.
7. Debrief behavior. Record the cue, crew response and resulting aircraft state to strengthen future performance.

Official references and further study

1. ICAO. Doc 9995-1 — Manual of Evidence-Based Training, Volume I: Aeroplane Pilots.
2. EASA. Easy Access Rules for Air Operations — ORO.FC.231 and associated AMC/GM on EBT and observable behaviors.
3. FAA. AC 120-71B — Standard Operating Procedures and Pilot Monitoring Duties for Flight Deck Crewmembers.
4. FAA. AC 120-123 — Flightpath Management.