What the score is
The approach score is a graded quality measure (0–100) of how closely an aircraft flew a stabilized approach profile, based on public flight trace data, runway geometry, aircraft context, derived flight behavior, and weather context where available. A higher score means the aircraft stayed closer to the expected runway path, descent profile, speed profile, and alignment during final approach. A lower score means one or more parts of the approach showed meaningful deviation from that profile.
It is not a pilot grade, an accident finding, or a regulatory conclusion. It is not a replacement for airline FOQA/FDM systems, onboard avionics, ATC data, or a flight data recorder. It is an analytical score for review, comparison, debriefing, and trend spotting.
What it's grounded in
A stabilized approach is not just "the aircraft landed." A good final approach usually has several things happening at the same time: the aircraft is aligned with the runway, the ground track is not wandering significantly, the aircraft is on or near the expected glidepath, the descent rate is controlled, the speed is appropriate for the aircraft and phase of flight, and the aircraft is not making aggressive bank or turn corrections close to the runway.
The criteria come from the Flight Safety Foundation Approach-and-Landing Accident Reduction (ALAR) stabilized-approach framework (the industry standard that underpins most airline FOQA/FDM programs), plus common FAA, ICAO, airline, and training concepts. The framework emphasizes being stabilized by a defined gate, commonly 1,000 ft above touchdown in instrument conditions or transport-category operations and 500 ft in visual / light-aircraft contexts, along with:
- runway alignment
- localizer/glidepath-style deviation
- controlled descent rate
- speed discipline on final
- excessive sink rate avoidance
- wings-level or appropriately banked final approach
- stabilization gates close to landing
SkyMeter does not claim that every operator uses the same numbers, gates, or tolerances. Airlines, aircraft types, training programs, and safety departments can define their own criteria. SkyMeter uses a consistent public-data model so flights can be compared across airports, runways, aircraft types, and time periods.
The six dimensions
Each scored approach is evaluated across six dimensions. The score is not based on one thing: a flight can lose points for being high, fast, offset, misaligned, descending aggressively, or still correcting close to the runway.
- Centerline (25%). Lateral offset from the extended runway centerline. Was the aircraft lined up with the runway, or left or right of the approach path? This is similar to the localizer dimension of an instrument approach, but SkyMeter computes it from the flight trace and runway geometry rather than cockpit instrument data.
- Track alignment (20%). The aircraft's ground track versus the runway heading. An aircraft may be near the centerline but still tracking at an angle, as during turns to final, sidesteps, corrections, or crosswind approaches.
- Descent rate (15%). Whether vertical speed was consistent with a controlled approach. A steep or rapidly changing descent rate can reduce the score, especially close to the runway.
- Speed (15%). Whether the aircraft appeared within a reasonable final-approach speed band. When reliable airspeed is not directly available, SkyMeter estimates wind-aware speed using groundspeed and weather context; reference speeds may be estimated from aircraft type, category, or known performance data. The speed score should be read as an analytical estimate, not a certified Vref comparison.
- Bank / turn behavior (15%). Whether the aircraft was still making aggressive lateral corrections on final. SkyMeter derives turn behavior from the flight path as a proxy for whether the aircraft was wings-level or still maneuvering close to the runway, which helps identify late turns, overshoots, unstable alignment, and circling-like behavior.
- Glidepath (10%). Vertical position relative to the expected descent path for the runway (a standard or published-style approach path where available). The goal is to identify whether the aircraft was high, low, or reasonably close to the expected final approach profile.
Six altitude gates, weighted toward the runway
SkyMeter samples the approach at multiple altitude gates above the runway, typically 3,000 / 2,000 / 1,500 / 1,000 / 500 / 200 ft AGL. The exact gates available depend on the quality and continuity of the flight trace. Each gate receives its own dimension scores, which are then blended into the final approach score.
The gates are weighted toward the runway: a deviation at 200 ft matters more than the same deviation at 3,000 ft, because there is far less time and altitude left to correct it. This is intentional: a wandering setup that's tidied up by short final still scores well, and a clean setup that destabilizes near the ground is penalized heavily, which is the behavior the safety standard intends.
Score vs. the "unstable approach" flag
The 0–100 score and the "unstable approach" flag are related, but they are not the same thing, and we keep them separate:
- The 0–100 score is a graded quality measure across the whole approach. It answers: "How good was the approach profile overall?"
- The "unstable approach" flag is a binary event detection based on whether the aircraft appears to violate stabilized-approach criteria at the stabilization gate (1,000 ft transport, 500 ft light GA). It answers: "Did the approach appear unstable at the key stabilization point?"
As a rule of thumb, a composite score of about 70 marks the boundary between stabilized-quality flying and an approach with meaningful deviations. Usually the two agree; when they don't, it is often because a single gate or single dimension crossed a threshold while the broader approach was otherwise acceptable. In those cases the score breakdown is usually more informative than the headline flag alone.
Reading a score
Use the score as a starting point, not the final answer. A high score usually means the approach was close to a stable, aligned, controlled profile. A medium score usually means the approach had noticeable deviations but may have been corrected before landing. A low score usually means the aircraft showed significant deviations in one or more important dimensions, especially closer to the runway.
The important part is the breakdown. On any flight's analysis page, the score breaks down into the six dimensions so you can see precisely where points were lost. A flight may score low because it was consistently high on glidepath; another because it turned late onto final and crossed the centerline; another may be well aligned but show excessive descent rate below 500 ft; another may be stable early but lose points because of short-final corrections. A score of 46 should not just mean "bad"; it should answer 46 because of what? The number is the summary; the breakdown is the explanation.
Honest limitations
SkyMeter does not currently score everything a pilot or airline safety department would consider. Public flight data usually does not include flap position, gear position, power setting, aircraft weight, exact landing configuration, cockpit-selected approach mode, ATC instructions, pilot intent, runway contamination, actual onboard indicated airspeed, or true flight data recorder values. So the score measures the parts of the approach that can be reasonably inferred from public trace data, runway geometry, aircraft context, and weather context.
- Configuration isn't scored. Gear and flap state aren't in the broadcast trace, so the configuration element of the ALAR criteria is not currently evaluated.
- It's derived from public data, not the FDR. Airspeed is frequently derived (from groundspeed and resolved wind) rather than broadcast, and reference speeds are estimated when the manufacturer value isn't published for the type. The analysis page flags estimated values.
- Sparse coverage is excluded, not guessed. Approaches are excluded from scoring when the trace is not good enough: missing data during final approach, signal loss before landing, large sample gaps, an uncertain runway match, unreliable altitude behavior, ambiguous aircraft identity or flight split, or an approach path too irregular for a fair runway-based score. When the data is too sparse, the better answer is no score, not a confident-looking number based on guesses.
- The scoring formula is ours. The criteria and tolerances are recognized standards; the specific dimension weights, gate set, and gate weighting are SkyMeter's own analytical model, analogous to how each commercial FOQA tool has its own scoring on top of the same shared criteria.
A pilot may remember a flight differently than SkyMeter scores it, and that does not automatically mean the pilot is wrong or the system is wrong. They may be looking at different things. The pilot may know ATC assigned a visual approach, that the aircraft was intentionally offset, that there was traffic spacing, a sidestep, or a strong crosswind correction, and that the approach was safe despite looking unusual from public data. SkyMeter analyzes the public-data trace; it does not know everything the crew knew. Use the score as an analytical prompt ("What does the data show?"), not as an accusation.
Related
See the broader methodology for the data sources behind every replay and known limitations, and how flight events are detected for how go-arounds, unstable approaches, stalls, runway events, and other detections work.