AirIndex Market Readiness Score.
The AirIndex Market Readiness Score (AIS) measures how real, and how near, commercial Advanced Air Mobility is in a market — across four co-equal pillars (regulatory authority, demand, landing capacity, and operator commitment) and two modulators (momentum and operability). It is not a forecast. It reflects current, documented conditions, and it is published in full to enable independent validation.
v2.0 rebalances the model away from v1.3's regulatory-weighted, permission-and-equipment framing toward whether a market is permitted, demanded, landable, and being committed to — all four at once.
What the Score Measures
The AirIndex Market Readiness Score is a 0–100 composite index that answers one question: how real, and how near, is commercial eVTOL operation in this metropolitan area right now? v2.0 evaluates each market against six factors — four co-equal pillars and two modulators.
The pillars are conjunctive: a market is most ready when permission, demand, landable infrastructure, and operator commitment are present together. No single pillar can carry a market to the top tier on its own. This is the central change from v1.3, which allowed a regulatory-heavy market to score highly on permission alone.
The score is evidence-based and reproducible. Every factor is computed from public records and AirIndex's own audited engines — FAA databases, state and federal filings, the conversion-feasibility model, the Operator Intelligence Database, and station weather climatology. It is not an investment recommendation or a demand forecast.
Why v2.0 — What Changed
v1.3 asked whether a market was permitted and equipped: it weighted the regulatory family at roughly 45% of the score and treated infrastructure and operators as binary present/absent flags. That model was reproducible and defensible, but it had two structural blind spots — it never scored demand (whether the flights had anywhere to go) and it never scored operability (how many days you could actually fly). It also let permission dominate the result.
v2.0 keeps everything v1.3 measured but rebalances and extends it:
- The regulatory family is consolidated into one graded pillar and pulled from ~45% to 18%.
- Demand and operability are added as first-class terms.
- Landing capacity is computed from convertible stock, not asserted infrastructure presence.
- Operator commitment is graded by depth and raised in weight — the truest “it's happening” signal.
- Momentum grades the rate of change, excluding operator activity already counted elsewhere.
v1.3 remains the archived prior version. Scores published under v1.3 keep their original methodology hash; v2.0 is a distinct, separately hashed version, and the cutover date is a recorded version event. This is the same immutability discipline that governs every published AirIndex number.
v2.1: How Activity Is Counted
v2.1 took effect on 2026-09-15. The weights, pillars, tiers, and formula are unchanged from v2.0. What changed is how activity is counted for the momentum modulator and for the operator commitment reading in markets with no operator on record.
- One story counts once. The same article often reaches AirIndex several times, through different news feeds, syndicated copies, and repeat collection. Momentum now counts distinct stories, not every copy.
- Activity is dated by the source. A filing or article counts toward the 90 day window by its own date, not by the date AirIndex processed it, so older records reviewed late no longer register as new activity.
- Rejected events do not count. An event AirIndex reviewed and rejected as a false positive no longer adds to momentum.
- Operator news counts once per story. In markets with no operator on record, the operator reading rises with distinct operator expansion stories, not repeat coverage of the same one.
On the cutover date every market was republished under v2.1, because a methodology change is itself a recorded event. Most markets moved down by a few points as repeated and outdated activity fell away. v2.0 remains the archived prior version with its own hash (b03a6aadbadd…).
v2.2: What an Unmeasurable Pillar Means
v2.2 took effect on 2026-09-16. The weights, pillars, tiers, formula, and the way activity is counted are all unchanged from v2.1. One thing changed: what happens when a pillar cannot be read at all.
Through v2.1, a market whose landing capacity could not be assessed scored zero on that pillar. That is the same number a market earns when every landing site it has was assessed and every one was found unsuitable. Those are opposite findings, and the score could not tell them apart. A reader had no way to know which one they were looking at.
Under v2.2 an unreadable pillar makes the whole score unpublishable. The market holds its previous published score, and the gap is recorded rather than filled. AirIndex does not publish a score containing a pillar nobody measured, in either direction: not a zero that reads as a finding, and not a middling value that reads as a measurement.
No market moved at this cutover, and no market was republished. Every tracked market has landing stock and an assessable share today, so the condition does not currently apply to any of them. Scores published under v2.1 remain correct under v2.2, because v2.2 changes nothing about how a readable market is scored, and they keep their v2.1 stamp until each market next has a material move. The version exists because the rule was wrong before it was needed, not after. v2.1 remains the archived prior version with its own hash (e82069aaa6a6…).
Closed Stock Is Counted, and Here Is Why
Some of the landing sites in a market are carried by the federal record as closed indefinitely. They are counted in the convertible stock above, and they are not quietly counted: this section exists so the choice is on the record rather than buried in a calculation.
The reasoning is that the landing pillar asks what could be brought to an eVTOL operating standard, not what is operating today. A closed pad with sound geometry is often the strongest conversion candidate a market has, because the structure already exists and there are no incumbent flight operations to work around. Several of the closed sites in the tracked markets are hospital rooftops of exactly that kind. Excluding them would lower a market's score for holding the asset the pillar is built to find.
The counterargument is real and we are not hiding from it: a closed pad is not available capacity today, and a reader sizing a market on what can fly this year will weigh it differently. That is why the closure is published per facility rather than folded silently into one number. Every facility record states its status, so a reader can open any site in a market and see whether it is operational, closed indefinitely, or closed permanently, and reach their own conclusion.
Closure here means the federal record states it. A site missing from a capture is not closed and is never treated as such. A site the record closes permanently is a different fact from one closed indefinitely, and the two are not merged.
The Four Pillars and Two Modulators
Regulatory authority is the legal permission to operate — enacted state legislation, the municipal regulatory posture, and vertiport zoning, combined into one graded pillar. In v1.3 the regulatory family carried roughly 45% of the total score. v2 pulls it to one of four co-equal pillars (18 of 100). The thesis: regulation enables a market, it does not by itself make one real. A statute with no demand, no landable infrastructure, and no operator committing capital is permission to do nothing. Regulation still matters — a market cannot operate commercially without it — but it is weighted as a prerequisite among equals, not the dominant term.
Graded composite: enacted/actively-moving/none legislation (45% of the pillar), friendly/neutral/restrictive posture (30%), and vertiport zoning present/absent (25%). The sub-weights concentrate on the factor with the longest lead time and highest capital-gating effect — legislation — while still crediting posture and zoning.
State legislature records, governor's office and state DOT publications, municipal zoning codes and council minutes, FAA community-engagement records.
Demand asks why anyone would fly here. A market can be fully permitted and fully equipped and still have no trip base to support commercial operations. v2 introduces demand as a co-equal pillar because the absence of a demand term was the largest blind spot in v1.3 — it scored permission and infrastructure but never whether the flights had anywhere to go. Demand is the clearest separator between markets that look ready on paper and markets where an operator can actually fill aircraft.
Composite proxy: metro-area population (60%) and medevac/hospital landing-facility density (40%), each normalized across the tracked set with a 90th-percentile cap so a single very large metro cannot crush the rest of the field. This is the one factor still carried as a proxy — population and medevac density stand in for modeled origin-destination demand pending a dedicated demand model.
U.S. Census metropolitan-area population, FAA NASR 5010 facility records (hospital/medevac landing facilities), AirIndex facility classification.
Landing capacity measures whether a market has somewhere for eVTOL aircraft to actually land — not in aggregate heliport count, but in convertible stock: how much of the market's real, assessed landing infrastructure could be brought to an eVTOL operating standard. This pillar is computed directly from AirIndex's conversion-feasibility engine over the market's FAA-registered heliports, not asserted. A market can be permitted and demanded and still have nowhere viable to put a vertiport; this pillar makes that constraint visible and scored.
Convertible share = feasible / (feasible + constrained + infeasible) across the market's assessed heliports, per the AirIndex conversion-feasibility model (load-bearing FATO gate, dimensional standard FAA EB-105A). Stock the federal record carries as closed indefinitely is counted, because the question this pillar asks is what could be brought to an eVTOL standard, and a closed pad with sound geometry can be. A market with no assessable stock is not scored zero and not imputed: the pillar is unreadable, and under v2.2 the market holds its previous published score until the stock can be assessed.
FAA NASR 5010 heliport records, AirIndex FacilityDimensions, AirIndex conversion-feasibility engine, RiskIndex facility audit.
Operator commitment is the truest single signal that commercial eVTOL is actually happening in a market, because operators commit capital to the markets they have judged ready. v2 raises operator weight relative to v1.3 and grades it by depth rather than scoring it as a binary presence flag. A market with a committed operating presence scores categorically higher than one with a single announced intention. Where a market has no curated operator presence but carries genuine operator-expansion signal, a capped interest floor credits that activity without letting it reach the level of a verified commitment.
Depth = sum of OID deployment-stage scores in-market (commercial operations weighted highest, announced lowest, exited zero), with full credit reached at a stage-score sum reflecting one operating presence plus breadth. For signal-rich markets with no curated presence, an operator-interest floor (noise-filtered, de-skewed, and ceilinged below full commitment) substitutes. The factor takes the greater of curated depth and the interest floor.
AirIndex Operator Intelligence Database (OID), operator disclosures and SEC filings, classified operator-market-expansion signals.
Momentum is the rate of change — how much non-operator activity (regulatory, infrastructure, and institutional events) a market has generated in the last 90 days, counted once per distinct story and dated by the source. It is a modulator, not a pillar: it does not establish readiness on its own, but it distinguishes a market actively moving from one that reached its current state and stalled. Momentum deliberately excludes operator-expansion signals, which are already counted under Operator Commitment, so the two terms do not double-count the same activity.
Count of classified, non-operator regulatory/infrastructure signals affecting the market over the trailing 90 days, normalized across the tracked set with a 90th-percentile cap. Operator-market-expansion events are excluded by construction.
AirIndex classification pipeline (Federal Register, LegiScan, regulations.gov, Congress.gov, municipal and press sources).
Operability is how many days you can actually fly — the share of flyable conditions derived from station weather climatology at the market's facilities. It is a modulator because weather does not determine whether a market is structurally ready, but it materially shapes the dispatch reliability an operator can expect. Two otherwise-equal markets are not equal if one loses substantially more days to instrument conditions, convective weather, or high wind. v2 makes that difference an explicit, climatology-grounded term rather than the v1.3 weather-infrastructure proxy.
Flyable share = 1 − the average weather-hazard score across the market's stations, computed from per-station IFR/LIFR/MVFR, convective, and high-wind climatology. Markets without station climatology are held at a neutral midpoint rather than scored as best- or worst-case.
Iowa Environmental Mesonet ASOS climatology, FAA NASR nearest-station mapping, AirIndex StationClimatology and weather-hazard model.
Weighting
Four co-equal pillars at 18 points each (72 of 100) and two modulators at 14 points each (28 of 100). The equal-pillar structure is the methodology's core claim: a market is ready when it is permitted, demanded, landable, and being committed to — and no one of those substitutes for another.
Sub-weights within the regulatory pillar and the normalization parameters are published in the methodology spec and available to licensed clients in full.
Readiness Tiers
Scores map to four tiers. The thresholds are unchanged from v1.3 (75/50/30), but their meaning is recalibrated: a top-tier score now requires strength across all four pillars at once, not dominance on any single one.
An ADVANCED market scores strongly across all four pillars at once: it is permitted, it has a real trip base, it has convertible landing capacity, and operators are committing depth — with momentum and operability not working against it. Readiness in v2 is conjunctive: a market reaches the top tier only when permission, demand, infrastructure, and operator commitment are simultaneously present. These are the markets where capital goes first.
A MODERATE market is strong on some pillars and visibly short on others — typically real demand and permission without the operator depth or convertible landing capacity to match, or the reverse. The gap is identifiable and closeable. These are active-transition markets where the missing pillar is the thing to watch.
An EARLY market has one or two pillars established — often demand or permission — but lacks the operator commitment and landing capacity that translate intent into near-term operations. Watch-list markets, not deployment markets.
A NASCENT market is weak across most pillars. Potential may exist on geography or population, but the structural prerequisites for commercial eVTOL — permission, demand, landable infrastructure, and operator commitment together — are not yet in place.
When the Public Score Updates
AirIndex recomputes every market's score continuously as new inputs arrive. The public number, however, is updated only when a change is material. A recomputed score is published when the four readiness pillars on their own move the score by at least two points or carry it into a different readiness tier, or when a documented high-confidence event is on record and the pillars it touched moved the score by at least a point. The pillar share is measured by itself: a large swing in the rolling modulators cannot ride through on a small pillar change, and a tier change that the modulators alone produce is held. Operator commitment counts only where operators are on record in the market; in markets without one, the operator reading comes from the volume of operator news, and a change there is treated like the rolling modulators. When a change does qualify, the full recomputed score is published. A movement carried by the modulators is held no matter its size, and the published number stays where it is until a real change moves it.
This rule exists because two of the score's inputs — activity momentum and flyable-day operability — are measured over rolling windows and normalized against the full set of tracked markets. As those windows advance, a market's score can drift by a point or more without anything having happened in that market, and a large enough drift can clear a fixed point threshold or a tier boundary on its own. So AirIndex requires a pillar event (permission, demand, landing capacity, or operator commitment) large enough to explain any published move that a cited driver does not already explain. AirIndex publishes events, not noise. A score that holds still is a deliberate statement that nothing material changed.
The full recomputation history is still recorded. Every staged score is preserved append-only as an immutable ScoreSnapshot; the publication policy governs only which of those snapshots is promoted to the public number, never what is retained. The policy itself is versioned and hash-anchored (market-publish-1.3, a4d3f893e7d5…) in the same registry as the scoring methodology, so the rule that decides what readers see is as tamper-evident and citable as the score it gates.
The Audited Record
Every score AirIndex publishes is preserved append-only. v2.0 scores are materialized as immutable ScoreSnapshot rows stamped with the methodology version and hash, so any historical score is a direct, tamper-evident lookup. v1.3 history is retained under its own version; the two series do not mix. A reader can navigate to any market's history page (e.g. /markets/phoenix/history) or query the same record through the JSON API.
The model is deterministic with published weights and a published spec hash (16694469d5095d54…). Anyone with the same source inputs should arrive at the same number — the definition of a reproducible, citable index.