How the Citable Score™ Works: Our Methodology

The Citable Score™ is a 0-100 metric that measures how likely AI search engines are to find, understand, and recommend your business. It evaluates six weighted categories across seven AI platforms, built on 36+ research citations from peer-reviewed studies and large-scale industry analyses.

We publish the full framework because transparency is a trust signal — both for human readers and for the AI systems that evaluate our content. If we’re going to measure AI citability, our own methodology should be citable.

This post summarizes how the scoring works. For the complete evidence base with all research citations, see our full methodology page.

Why a New Metric Exists

Traditional SEO metrics — domain authority, keyword rankings, organic traffic — are poor predictors of AI visibility. Ahrefs found that only 12% of URLs cited by AI search engines rank in Google’s top 10 for the same query. Web mentions correlate with AI brand visibility at 0.664, three times stronger than the correlation for total backlinks (0.218).

This means a business can have excellent Google rankings and be completely invisible to ChatGPT, Perplexity, and Gemini. The Citable Score™ was built to measure what existing metrics miss.

Source: Ahrefs, 2025

The Six Scoring Categories

Each category maps to specific stages of the AI citation pipeline — the retrieval-augmented generation (RAG) process that determines which sources AI systems cite. Weights reflect the strength of empirical evidence, the mechanistic role of each signal, and patterns observed across our audits.

AI Citability & Visibility — 25%

The highest-weighted category measures whether AI systems can find and extract citable content from your website. This includes content structure (120-180 word sections, answer-first format), brand mentions across AI-indexed platforms, AI crawler access, and machine-readable content discovery files.

The Georgia Tech GEO paper — the first controlled experiment in generative engine optimization — tested nine strategies across 10,000 queries. Adding source citations improved visibility by 40%, statistics by 38%, and quotations by 35%. Content citability is the foundational GEO signal.

Source: Aggarwal et al., ACM SIGKDD, 2024; SE Ranking, 2025

Entity Recognition & Brand Authority — 18%

AI systems resolve brands as entities by cross-referencing Wikipedia, social platforms, review sites, and online mentions. Wikipedia accounts for 47.9% of ChatGPT’s top-10 most-cited sources. Reddit occupies 27% of ChatGPT’s internal search slots but appears in only 0.35% of visible citations — 99% of its influence is invisible.

We measure presence across Wikipedia, LinkedIn, YouTube, Reddit, G2/Capterra (for SaaS), Google Business Profile (for local), and other platforms relevant to each business type. A brand that doesn’t exist as an entity across these platforms is structurally disadvantaged.

Source: Qwairy, 2025; Discovered Labs, 2025; Ahrefs, 2025

Content Quality & E-E-A-T — 20%

Experience, Expertise, Authoritativeness, and Trustworthiness signals correlate strongly with AI citation. BrightEdge reported that 96% of AI Overview content comes from sources with verified E-E-A-T signals, with a correlation of r=0.81.

We evaluate author attribution, first-hand experience indicators, content depth, source citations within content, and whether content provides complete answers. Amazon AGI’s research found that AI systems cite wrong sources 41% of the time — clearly authoritative content has the best chance of correct attribution.

Source: BrightEdge, 2025; Sorodoc et al., ACL 2025

Technical Foundations — 15%

Server-side rendering, Core Web Vitals, crawlability, and recency signals form the technical baseline. Content that isn’t technically accessible cannot enter the AI citation pipeline at any stage.

The most significant finding in this category: publication dates change AI reranking outcomes by up to 95 positions, even when content relevance is identical. This is a peer-reviewed, controlled experimental finding from Waseda University — one of the strongest pieces of evidence in the entire GEO research landscape.

Source: Fang et al., SIGIR-AP 2025; BrightEdge, 2025

Schema & Structured Data — 7%

JSON-LD structured data removes ambiguity about what your business is and how it relates to other entities. Microsoft confirmed at SMX Munich 2025 that schema markup directly helps their LLMs understand content — the first official platform confirmation.

GPT-4 accuracy improved from 16% to 54% when content used structured data — a 3.4x improvement. Sites implementing FAQ blocks with schema saw a 44% increase in AI search citations. We evaluate Organization schema, sameAs links, Article markup, FAQPage, BreadcrumbList, and industry-specific schema types.

Source: Search Engine Land, 2025; Data World, 2025; BrightEdge, 2025

Platform Readiness — 8%

Each AI platform has a distinct citation fingerprint. 86% of top-cited sources are unique to one platform — only 7 websites appear in the top 50 for all three of ChatGPT, Perplexity, and Google AI Overviews. We score readiness for each of the seven platforms individually.

This category carries the lowest weight because platform-specific patterns shift rapidly. Reddit’s citation share on ChatGPT went from ~60% to ~10% in six weeks. We update platform scoring as new data emerges.

Source: Ahrefs, 2025; Semrush, 2025

The Seven Platforms We Measure

Each audit evaluates visibility across Google AI Overviews, ChatGPT, Perplexity, Gemini, Bing Copilot, Meta AI, and Apple Intelligence. Each platform has different source preferences:

Google AI Overviews draw 54.5% of citations from organically-ranked pages. YouTube dominates at 9.5% citation share.

ChatGPT relies heavily on Wikipedia (16.3% of all citations) and has volatile Reddit influence.

Perplexity sources 46.7% of citations from Reddit and favors real-time web results.

Gemini has the highest first-party citation rate at 52.15% — it favors your own brand content more than any other platform.

Bing Copilot draws from the Bing index and benefits from IndexNow and LinkedIn presence.

Meta AI leverages Facebook and Instagram social graphs.

Apple Intelligence prioritizes Apple Business Connect and Yelp for local queries.

Score Tiers

Scores map to five tiers that indicate overall AI search readiness:

90-100 (Leading). AI search engines can reliably find, understand, and cite your business. Strong entity presence, structured content, and multi-platform optimization.

75-89 (Optimized). Solid foundation with specific gaps in one or two categories. Targeted improvements can push into Leading territory.

60-74 (Fair). Basic visibility exists but significant gaps in content structure, schema, or brand authority limit citation probability.

40-59 (Poor). Most AI platforms struggle to find or correctly identify your business. Multiple categories need attention. This is where the majority of businesses score today.

0-39 (Critical). Near-invisible to AI search. Often caused by blocked crawlers, missing schema, or zero entity presence across platforms.

Early audit data shows the average score is 49/100 — firmly in the Poor tier.

Confidence Levels

Not all data is equally reliable. Each scored dimension carries a confidence level:

High confidence — based on direct measurement or API data. Examples: Lighthouse Core Web Vitals scores, confirmed social profiles, schema validation.

Medium confidence — based on proxy signals or partial data. Examples: estimated performance from framework signals, web-search-based brand detection.

Low confidence — based on inference or unverified sources. Examples: API errors, assumptions about platform behavior.

Confidence levels are reported in every audit so you know which findings are rock-solid and which are directional.

What the Score Is Not

The Citable Score™ does not predict specific AI responses to specific queries. It does not guarantee citation by any platform. And it does not replace traditional SEO measurement. It measures the structural signals that make citation possible — the prerequisites that determine whether your business is eligible to be cited, not whether it will be cited for any individual query.

For the complete evidence base with all 36 research citations, see our full methodology page.

Frequently Asked Questions

How often are the scoring weights updated?

Weights are reviewed quarterly as new research emerges. The current allocation reflects evidence available through March 2026. Significant changes are announced on this blog.

Can I compare my score to competitors?

Yes. Each audit measures the same six categories and seven platforms, making scores directly comparable across businesses and industries.

Why is Platform Readiness weighted at 8%?

Platform-specific citation patterns shift rapidly — Reddit’s ChatGPT citation share dropped from 60% to 10% in six weeks. Structural signals like content quality, schema, and brand authority are more stable and predictive. We weight accordingly.

What data sources does the audit use?

Each audit combines automated analysis (Lighthouse, schema validation, crawler checks) with AI agent probes across all seven platforms. Brand presence is verified through direct platform checks. See our methodology page for the complete research foundation.

Is the Citable Score™ the same as a GEO score?

The Citable Score™ is our specific implementation of GEO measurement. “GEO score” is a general term. Our framework is built on 36+ peer-reviewed citations and measures six specific categories. Learn more about GEO.