The 93-factor AEO taxonomy is AIVZ's comprehensive measurement framework for AI visibility. No competitor has published or operationalized anything comparable. Every factor carries a confidence label. Every score is reproducible from the same factor evidence.
The 93-factor taxonomy didn't start at 93. It started at 27 — the original "Core" factor set ported from AIVZ's WordPress plugin foundation in early 2025. That Core set covered the highest-impact AEO signals available at the time, and it was sufficient to produce a meaningful initial score.
The taxonomy expanded through five implementation phases as the AEO discipline matured. Each phase added factors only after they cleared three tests.
The current count is 93. It will be more next year and the year after — AEO is a young discipline; the measurable surface keeps expanding. The number is not the point. The point is that every factor we measure is calibrated, documented, validated, and inspectable.
Detectable through deterministic rules, validated LLM judgments, or platform APIs we have stable access to.
Carries a confidence label that honestly reflects how proven its citation impact is.
Correlates with observed AI citation outcomes against real platforms before shipping.
Categories map onto the three-layer AI Visibility Stack but are not isomorphic with it — some categories span multiple layers.
What it measures: Whether AI bots can physically reach and render your content. The foundational category — every other category presupposes this one works.
robots.txt AI bot permissions (GPTBot, ClaudeBot, PerplexityBot, GoogleOther, Bingbot)llms.txt declarationUnintentional robots.txt blocks that pre-date AI crawler awareness; aggressive WAF rules that challenge AI user agents; JS-rendered content with no SSR fallback.
llms.txt, AI-bot-specific user agent identification).What it measures: Whether AI systems can parse the structured metadata you publish — JSON-LD, Schema.org types, machine-readable feeds, semantic HTML.
sameAs to authoritative profilesmainEntity, nested entity references, cross-page consistencyNo JSON-LD at all; JSON-LD with errors that pass display tests but fail semantic parsing; Organization schema present but no sameAs linking; schema-content drift where the structured data describes content that isn't actually on the page.
What it measures: Whether AI systems can extract clean, citable answer blocks from your prose. The "structural quality of writing" category — most factors here are about how content is organized, not what it says.
Answers buried in paragraph 4 or later; headings that are decorative rather than question-aligned; comparison content in prose rather than tables; statistics without source attribution; no FAQ structure on pages where Q&A format would extract better.
What it measures: Whether AI systems recognize the entities (people, organizations, places, products, concepts) on your pages — and whether those entities are grounded in authoritative knowledge graphs.
No Wikidata entity for the company; inconsistent entity naming across the site; ambiguous entity references that AI can't disambiguate confidently; missing cross-page entity links.
What it measures: Whether AI systems trust the source — author credentials, publication history, freshness, original research, factual accuracy, YMYL handling.
Anonymous content (no named author); author bios that don't establish credentials; stale content with last-updated dates from years ago; YMYL content without reviewer credentials; no editorial policy or corrections policy.
What it measures: Whether external sources validate your authority — citations, references, mentions, links from authoritative domains. The AuthorityGraph engine surface.
Strong site, weak external authority — domain hasn't been cited or referenced enough by authoritative sources; author authority disconnected from organizational authority; over-concentration in low-authority sources.
What it measures: Whether your content matches the way users actually ask questions — conversational alignment, topical depth, intent matching, query-to-answer correspondence.
Content written in marketing language that doesn't match how users phrase their questions; thin coverage on pages claiming topical depth; answer format mismatched to the question.
What it measures: Per-platform readiness for the major AI answer surfaces — ChatGPT, Google AI Overviews, Perplexity, Gemini, Microsoft Copilot, voice assistants. Captures readiness that doesn't generalize across platforms.
Composite score in the AI Extractable tier but Invisible to AI on Voice (no speakable schema); strong on Google-derived surfaces but weak on Perplexity (live-crawl issues); IndexNow not adopted, slowing Bing/Copilot indexing.
What it measures: Whether you can track AEO outcomes over time — AI crawler analytics, citation simulation, score history, alert systems, change detection. This category isn't about being visible; it's about measuring visibility over time.
No crawler analytics — you can't tell whether AI bots are even reaching your content; no citation tracking — you can't tell whether the score improvements are correlating with citation improvements; no historical record.
AEO is a young discipline. We don't pretend everything is equally proven, and we don't bury uncertainty in marketing copy.
| Label | What it means | Example factors |
|---|---|---|
| Established | Well-supported by web standards, platform documentation, or broadly accepted technical practice. | JSON-LD presence, robots.txt configuration, HTTPS, mobile usability, Schema.org core types |
| Strongly Inferred | Not always formally documented, but strongly supported by research or repeated industry observation. | Front-loaded answers, concise answer blocks, citation-formatting quality, entity density |
| Indirect / Correlated | Likely influences AI visibility indirectly through search prominence, authority, or trust. | Off-site authority signals, social presence, brand mention frequency, backlink profile |
| Emerging / Experimental | New or evolving factors not yet stable or universally adopted. | Speakable schema, IndexNow support, platform-specific freshness weighting, NavBoost-class signals |
Confidence labels move both ways. A factor classified as Emerging can be promoted to Strongly Inferred as evidence accumulates. A factor classified as Strongly Inferred can be demoted to Indirect / Correlated if the evidence base weakens. The label is a current assessment, not a permanent assignment.
Public changelogThe 93-factor taxonomy didn't ship at 93. It expanded through five phases as we validated factors against real citation outcomes.
| Phase | Factors Added | Cumulative Total | Focus |
|---|---|---|---|
| Phase 1 · Foundation | 27 | 27 | Highest-impact factors ported from the AIVZ WordPress plugin — the first measurable AEO signal set with validated citation correlation. |
| Phase 2 · On-Page Expansion | +28 | 55 | Comprehensive on-page coverage: schema completeness, content structure, entity grounding, internal authority signals. |
| Phase 3 · Authority + Platform | +15 | 70 | Off-site authority integration; per-platform readiness scoring for the six major AI surfaces. |
| Phase 4 · LLM + Advanced NLP | +15 | 85 | LLM-judged factors — semantic matching, topical depth, conversational alignment — with the 0.18 weight cap on LLM-derived sub-scores. |
| Phase 5 · Remaining Experimental | +8 | 93 | Emerging factors: IndexNow, machine-readable feed architecture, cross-page entity consistency, YMYL sensitivity, platform-specific freshness, NavBoost-class signals. |
Phase 5 includes factors classified as Emerging / Experimental. We measure them and surface them with the appropriate confidence label rather than excluding them — the alternative (waiting until they're "proven") means missing the leading edge of where AEO is heading.
Future phases will likely focus on multimodal content readiness as AI answer surfaces increasingly include multimodal output and citation.
The taxonomy is operationalized across the AIVZ subscription tiers. Free-tier scans cover the most-impactful subset. Paid tiers extend coverage in proportion to subscription scope.
| Tier | Factor Coverage |
|---|---|
| Free | Highest-impact subset — enough to produce a meaningful score and the top three fix recommendations. Sufficient for ad-hoc scans. |
| Pro | Comprehensive on-page coverage — all factors that can be evaluated from a single-page or single-domain crawl without external authority data. |
| Agency | On-page coverage plus off-site authority signals plus per-platform readiness. The full set needed for client-facing portfolio work. |
| Enterprise | Complete 93-factor coverage including emerging factors and custom-extension factors negotiated per engagement. |
Coverage at every tier is fully documented in the product — every score is paired with the factors that produced it. There's no "premium-tier-only" hidden methodology.
Tier scope, capabilities, and pricingWithin the 93-factor taxonomy, eleven specific factors carry disproportionate impact on whether AI systems select a page for citation. The Core 11 spans Categories 2, 3, 4, and 5.
The Core 11 is the highest-leverage starting point. The other 82 factors matter — that's why they're in the taxonomy — but the Core 11 is where the leverage concentrates.
Per-factor depth on the Core 11Get the composite score, layer breakdown, and category-by-category detail of which factors are passing and failing on your real content.
Run a free scanPer-factor depth on the eleven highest-leverage factors. The starting point if you're prioritizing AEO work.
Read the Citation Core 11The three-layer Stack — Access, Understanding, Extractability — and how factors aggregate into the composite score.
Read the AI Visibility StackOr — return to the methodology hub: Read the AEO methodology →
Run a free scan. Get your AI Visibility Score and per-category factor breakdown across the highest-impact subset — in under 60 seconds.