Intelligence Framework · v1.1
How ChatRank measures AI visibility.
A documented framework for what we measure, how we measure it, what the numbers mean — and what they don't. Built for CMOs who need defensible data and analysts who need to stress-test the methodology.
Up to 4
AI Platforms Tracked
0–100
Composite Score Range
Our Philosophy
Measurement as discipline.
ChatRank was built on a simple assumption: the most useful thing we can do for a B2B CMO is tell them what is actually happening in AI engines — not what we wish were happening. That means publishing our methodology, acknowledging its limitations, and refusing to imply precision we haven't earned.
AI visibility measurement is an observability discipline, not a ranking system. We surface directional signals that tell you where you stand, where your competitors are winning, and what is structurally driving the gap. We do not manufacture certainty. We build frameworks for understanding a system that is, by design, probabilistic.
"A single AI query is an anecdote. A structured query set, run consistently across platforms and over time, is intelligence."
— ChatRank Methodology, v1.1Section 1 · Context
The shift to AI-assisted buying.
B2B buyer research has moved through three distinct phases in the last two decades. Each phase required a different visibility strategy. Most B2B marketing teams are optimized for Phase 2. Buyers are already in Phase 3.
Phase 01
Search Engines Ranked Pages
Google returned ten blue links. Buyers clicked, compared, and decided. SEO optimized for crawler signals: keywords, backlinks, domain authority.
Still ActivePhase 02
Review Platforms Ranked Vendors
G2, Capterra, and Gartner Peer Insights became the research layer. Buyers validated vendor shortlists through peer reviews and analyst ratings.
Still ActivePhase 03
AI Systems Synthesize Recommendations
AI engines answer buyer questions directly — citing whoever they judge to be the category authority. The buyer never sees the alternatives. That is a market signal being sent without your input.
Now · AcceleratingGartner projects that traditional search engine volume will drop 25% by 2026 due to AI chatbots and virtual agents. G2's 2026 Buyer Behavior Report found that 51% of B2B software buyers now start research with an AI chatbot — more often than Google. Forrester's State of Business Buying identifies generative AI as the single most cited research source in B2B purchasing, ahead of vendor websites and sales representatives.
Section 2 · The Five Signals
What ChatRank measures.
Every ChatRank engagement — free or paid — measures five proprietary signals. Each signal is defined, scored, and interpreted consistently. Expand any signal to see exactly what it measures, why it matters, how the calculation works, and — critically — what it does not mean.
- What It Measures
- The percentage of AI-generated answers in a defined category where your brand is mentioned at least once.
- Why It Matters
- This is the foundational visibility metric. If AI engines don't mention your brand when buyers ask category questions, you are absent from the consideration set — regardless of your SEO performance.
- How We Calculate It
- Measured across four buyer-intent queries per AI engine. Brand name detection is case-normalized and whitespace-collapsed. A mention counts once per response regardless of how many times it appears.
- Score Interpretation
- 0–10%: Brand is not yet established as a category authority in AI engines. 10–30%: Emerging presence. 30–60%: Competitive. 60%+: Category leader.
What This Signal Does NOT Mean
AI Share of Voice measures presence, not quality of mention. A brand mentioned once in a caveat scores the same as a brand recommended first. Answer Position is the signal that captures mention quality.
- What It Measures
- The percentage of AI-generated answers in which your brand's own content is cited as a source — distinct from brand mention.
- Why It Matters
- Citation is the stronger signal. When an AI engine cites your content, it is treating your brand as a primary source, not a secondary reference. Citation is what drives recommendation authority over time.
- How We Calculate It
- Measured primarily via Perplexity, which returns explicit source citations with every response. Also fed by the brand-anchored probe query across all platforms. A citation requires the AI engine to reference a specific URL or publication, not merely name the brand.
- Score Interpretation
- Citation Density below 5% indicates content is not structured for AI retrieval. Above 20% indicates strong content authority.
What This Signal Does NOT Mean
Citation Density is most reliably measured on Perplexity, which is web-grounded and returns explicit citations. Claude and Gemini answer from training data and do not return citable sources in the same way — their contribution to this metric is structural, not link-based.
Citation Accuracy Disclosure
Citation mapping in AI-generated answers is a probabilistic process. AI engines may surface a brand mention without citing a specific URL, or may attribute a claim to an adjacent source. ChatRank's Citation Rate measures observed brand presence and sourcing patterns across a standardized query set. Individual citation accuracy within any given response is not guaranteed by the underlying AI platforms.
- What It Measures
- The average position at which your brand is first mentioned within AI-generated responses. Position 1 means your brand appears in the first sentence or recommendation. Position 5+ means you appear late or not at all.
- Why It Matters
- Where you appear matters as much as whether you appear. A brand recommended first in an AI response has meaningfully more buyer influence than one mentioned sixth in a comparison list. Answer Position captures the quality dimension that AI Share of Voice cannot.
- How We Calculate It
- Scored on a log scale. Position 1 scores 100. Position 2 scores approximately 75. Position 3 scores approximately 50. Position 5 or greater scores near zero. Responses where the brand is not mentioned receive a null position, not a zero — this distinction matters for averaging.
- Score Interpretation
- An average Answer Position below 3.0 indicates competitive positioning. Above 3.0 indicates the brand is being surfaced as a late consideration or alternative, not a primary recommendation.
What This Signal Does NOT Mean
Answer Position is sensitive to query phrasing. A direct brand comparison query will produce a different position than an unprompted category query. ChatRank scores use only unprompted buyer-intent queries to ensure position scores reflect organic AI citation behavior, not prompted brand defense.
- What It Measures
- A ratio expressing how frequently your top competitor appears in AI-generated category answers relative to your brand. A score of 9× means the top competitor appears nine times more often than you.
- Why It Matters
- Displacement is the urgency metric. It translates an abstract visibility gap into a concrete competitive reality. When a buyer asks AI about your category and your competitor answers 9 times before you answer once, that is a pipeline signal — not a content problem.
- How We Calculate It
- The top competitor is identified automatically from the brands most frequently mentioned across all unprompted query responses. The ratio is computed as competitor mention count divided by your brand mention count. When your brand appears zero times in unprompted queries, the ratio is returned as null — 'brand was not mentioned' — rather than as a fabricated ratio. This is an intentional design decision.
- Score Interpretation
- A displacement ratio of 1–2× is competitive parity. 3–5× indicates a meaningful gap requiring systematic intervention. Above 5× indicates the competitor has established category authority and your brand is not in the active AI consideration set.
What This Signal Does NOT Mean
Competitive Displacement identifies who is winning, not why. The why — entity architecture gaps, content structure, citation source quality — requires the paid engagement audit with human analysis.
- What It Measures
- The month-over-month rate of change in AI Share of Voice. A positive Visibility Velocity means your AEO interventions are compounding in AI engine understanding of your brand.
- Why It Matters
- Velocity is the leading indicator. It shows whether your investments are working before the AI Share of Voice number moves significantly. Positive velocity on a low-SOV brand means you are on the right trajectory — the compounding has begun.
- How We Calculate It
- Baseline is established on the first ChatRank audit. Velocity is calculated from the second audit onward. Because a single-point measurement cannot produce a trend, Velocity carries zero weight on first-audit scores and the remaining four signals are reweighted proportionally.
- Score Interpretation
- Any positive Velocity in the first 30 days indicates schema and structured data changes are surfacing. Sustained positive Velocity over 90 days indicates entity authority is compounding. Flat or negative Velocity on a low-SOV brand requires strategy review.
What This Signal Does NOT Mean
Velocity is excluded from the free automated audit score. It requires at least two measurement periods and is reported monthly in retainer engagements only.
Section 3 · The ChatRank Score
The composite score.
The ChatRank Score is a 0–100 composite weighted across all five signals. Weighting reflects the relative pipeline impact of each signal based on current AI engine citation behavior research.
Competitive Displacement20%
Why AI Share of Voice Leads at 30%
Share of Voice is the foundational signal. A brand that doesn't appear in AI category answers cannot benefit from Answer Position, Citation Density, or competitive framing. Presence precedes influence.
Why Velocity Is Excluded from First Audits
Velocity requires two measurement periods to compute. On a first audit, Velocity carries 0% weight and the remaining four signals are reweighted proportionally to sum to 100%. Reporting a velocity of zero on a first audit would be misleading — it is simply unmeasured.
Score Range Interpretation
0–20: Brand is not yet established in AI engine knowledge. 20–40: Emerging presence. 40–60: Competitive position. 60–80: Category authority. 80+: Dominant AI visibility.
Section 4 · How We Measure
The query framework.
ChatRank uses buyer-intent queries exclusively — the questions real buyers ask AI engines when researching a purchase. We do not use informational queries ("what is CRM software?") because they produce encyclopedic responses that do not reflect purchase-stage AI behavior. The distinction matters: the same brand can score very differently on informational versus buyer-intent queries. We score the queries that send pipeline signals.
Free Audit — 5 Queries
Q1
Best [category] for [target buyer]
Unprompted · Feeds SOV
Q2
[Category] comparison and reviews
Unprompted · Feeds SOV
Q3
Top [category] platforms 2026
Unprompted · Feeds SOV
Q4
[Primary use case] software recommendations
Unprompted · Feeds SOV
Q5
Tell me about [company] — what do they do, who are their main competitors, and where are they cited in [category]?
Brand Probe · Feeds Citation Density
Free Audit
Directional Signal
4 unprompted queries + 1 brand probe across OpenAI and Perplexity. Produces statistically sufficient directional signal for category visibility gaps. Results delivered in 60–90 seconds. Automated, no human review.
Paid Engagement Audit
Verified Intelligence
20–30 buyer-intent queries across all four AI platforms including named competitor comparisons. Human-reviewed findings. Results in 5 business days. The full query set reduces response variance and enables statistically meaningful competitive displacement mapping.
Retainer Measurement
Longitudinal Trend
Monthly re-audit using consistent query set. Directional trend over time is more reliable than any single measurement point. Visibility Velocity — the leading indicator — is only computable across multiple measurement periods. This is the product that turns data into strategic intelligence.
Section 5 · Measurement Integrity
The non-determinism acknowledgment.
Large language models are probabilistic by design. The same prompt submitted twice can return different responses. Any methodology that does not address this directly is either unsophisticated or dishonest. Here is how ChatRank handles it.
Free Audit
Directional, Not Statistical
The free audit uses 4 unprompted queries to produce a directional signal. We do not claim statistical certainty from 4 queries. We claim — and consistently deliver — a directional read on whether your brand is visible, absent, or competitive in AI engine category answers. A brand scoring 8% on a free audit is not precisely 8%. It is demonstrably not at 40%, which is what matters for a CMO deciding whether a gap exists.
Paid Audit
Reduced Variance
Paid engagement audits use 20–30 queries per platform. A larger query set meaningfully reduces the variance that a single-query anecdote cannot. Named competitor comparisons, multiple query types, and cross-platform measurement produce findings that hold up to analytical scrutiny. Human review validates AI findings against raw responses before delivery.
Retainer
Trend Beats Point-in-Time
Longitudinal trend movement over time is more reliable than any single measurement. A brand whose Visibility Velocity is positive across three consecutive monthly audits is genuinely compounding AI visibility — regardless of point-in-time score fluctuation. The retainer product is designed to surface the trend, not to manufacture false precision from individual measurements.
Read the full argument: Is AEO Built on Sand? →
Section 6 · Platform Coverage
What we track — and what we don't.
We track every AI platform we can measure reliably. We do not promise coverage of platforms where reliable measurement is technically impossible. That distinction is not a limitation to apologize for — it is a commitment to delivering data you can actually trust.
On Training Data vs Real-Time Signals
Perplexity and OpenAI are real-time web-grounded — content changes surface in their responses within days. Claude and Gemini answer from training data — a brand that improved its AEO recently may not see that reflected until the next model training run, which is unpredictable. ChatRank labels these distinctly in all paid reports: Real-Time Web Signal vs Training Data Signal. Both are valid signals of different things. Neither is withheld from the analysis.
Section 7 · Research Grounding
The evidence base.
ChatRank's methodology is grounded in publicly available research and documentation from the organizations building the AI systems we measure. These organizations validate the market transition — not ChatRank specifically. We cite them because the research supports the strategic case for AI visibility, not because they have reviewed or endorsed our methodology.
Google · E-E-A-T Framework
Google's Quality Evaluator Guidelines define Experience, Expertise, Authoritativeness, and Trustworthiness as the four criteria for content quality. E-E-A-T is a search-quality rater guideline, not an AI citation algorithm — but it describes signals that are directionally aligned with what generative engines reward when selecting sources. ChatRank's entity and content recommendations are designed to satisfy both.
Google Search Quality Guidelines →Gartner · Search Volume Prediction
Gartner predicts traditional search engine volume will drop 25% by 2026 due to AI chatbots and virtual agents. The market transition ChatRank is designed to help B2B companies navigate is documented at the analyst level, not just claimed by a vendor.
Gartner Press Release, Feb 2024 →G2 · Buyer Behavior Report 2026
G2's 2026 Buyer Behavior Report found that 51% of B2B software buyers now start research with an AI chatbot — more often than Google. 69% chose a different vendor than planned based on AI guidance. One third purchased from a vendor they had never previously encountered.
G2 Buyer Behavior Report →Forrester · State of Business Buying 2026
Forrester's State of Business Buying identifies generative AI as the single most cited research source in B2B purchasing decisions — ahead of vendor websites and sales representatives. The buyer journey has shifted before most marketing teams have adapted to it.
Forrester Research →Generative Engine Optimization Research
ChatRank's methodology extends a body of peer-reviewed academic research on generative engine optimization (GEO) — the discipline of measuring and improving how brands appear in AI-generated answers. Foundational research has identified that expert quotations, statistics, and authoritative source citations are the highest-impact content interventions for AI citation. ChatRank's signal framework — Answer Position, Citation Density, Competitive Displacement — operationalizes these findings into measurable business outcomes.
What We Do Not Claim
ChatRank's methodology does not derive from any proprietary or non-public source. We extend a published, peer-reviewed framework with additional signals — Competitive Displacement and Visibility Velocity — developed through real client engagements. We do not claim that any of the organizations whose research we cite have reviewed or endorsed our methodology. The mechanism is documented. The differentiation is the application.
Section 8 · Scope and Boundaries
What ChatRank is not.
Clarity about what a methodology does not do is as important as clarity about what it does. These are the boundaries of ChatRank's methodology — not apologetically, but precisely.
◎
Not an SEO Tool
ChatRank does not measure keyword rankings, domain authority, backlink profiles, or any signal used by Google's crawler-based ranking system. These are valid SEO metrics. They are not AI visibility metrics. A company can rank on page one of Google for its core keywords and score 4% on AI Share of Voice. Both things can be true simultaneously.
⊘
Not a Guarantee of AI Citation
No methodology can guarantee that an AI engine will cite a specific brand. AI engines update their models, retrieval behavior, and citation logic continuously and without notice. ChatRank monitors these changes and adapts its methodology accordingly. What ChatRank can guarantee is a systematic, documented, repeatable approach to building the entity authority and content architecture that AI engines consistently favor.
⌘
Not a Content Farm
ChatRank does not generate high-volume, low-quality content designed to flood AI training data. This approach — sometimes called AI bait — produces short-term citation spikes followed by model updates that penalize it. ChatRank's methodology builds entity authority through coherent, citable, structured content that reinforces brand positioning across trusted ecosystems.
↗
Not a Black Box Score
Every ChatRank score can be traced back to the queries that produced it, the platforms that were queried, the brand mentions that were counted, and the calculation that produced the final number. This document exists to ensure that any analyst, CMO, or CFO who wants to understand what the number means has everything they need to evaluate it independently.
On the difference between entity authority and AI bait
The distinction between building genuine AI visibility and gaming AI systems is a real one — and it matters for long-term results. Modern AI systems increasingly reward coherent entity authority across trusted ecosystems rather than isolated content volume. A brand that publishes one well-structured, accurately cited, FAQ-schema-marked piece of content that answers a buyer's question definitively will outperform a brand that publishes fifty thin posts designed to repeat keywords for AI indexing. ChatRank's methodology is built on the former principle. The content we write is designed to be genuinely useful to buyers — because that is exactly what AI engines are trained to recognize and cite.
Section 9 · Frequently Asked Questions
Questions this page should answer.
These are the questions a serious buyer will ask after seeing a ChatRank report. Answered directly.
What is AI Share of Voice?
▼
AI Share of Voice is the percentage of AI-generated answers in a defined category where your brand is mentioned at least once. A score of 8% means your brand appeared in 8 out of 100 AI-generated answers about your category.
How is Citation Density different from AI Share of Voice?
▼
AI Share of Voice measures whether your brand is mentioned. Citation Density measures whether your content is cited as a source. A brand can be mentioned frequently in AI responses without any of its own content being cited — typically because the AI engine has absorbed brand awareness from third-party sources rather than from the brand's own authoritative content. Citation is the stronger signal of content authority.
Why does my score sometimes differ from what I see when I ask ChatGPT myself?
▼
Because AI language models are probabilistic, not deterministic. The same question asked twice can return different results depending on the model's temperature settings, session context, and retrieval state. A single manual query is an anecdote. ChatRank scores are based on repeated, structured query sets run under controlled conditions — which is why they differ from casual observation, and why that difference is the point.
Why does score fluctuate between audits?
▼
Directional trend over time is more reliable than any single measurement. LLM outputs have inherent variance. A single audit score is a signal, not a census. This is why ChatRank retainer engagements track Visibility Velocity — the month-over-month trend is more meaningful than any individual score. If your score moves significantly in one direction over three consecutive audits, that is a real signal. A single-audit fluctuation within a narrow range is expected and normal.
Can two companies see different results from the same prompt?
▼
Yes. LLMs are non-deterministic by design. Two identical prompts submitted seconds apart can return different orderings, different brand mentions, and different citation sources. ChatRank controls for this by using multiple queries per platform, normalizing results across query runs, and reporting directional signal rather than precise rankings. The methodology is designed to surface real visibility gaps, not to imply false measurement precision.
What AI engines does ChatRank track?
▼
The free automated audit covers ChatGPT and Perplexity (sonar-pro). Paid engagements add Claude and Gemini. Models are role-assigned and pinned to dated snapshots: OpenAI gpt-5.2-2025-12-11 handles judgment and extraction, gpt-5.4-mini-2026-03-17 runs the volume buyer queries, Anthropic claude-haiku-4-5-20251001 handles structured parsing, and claude-sonnet-4-6 writes the report narrative. Claude and Gemini scores are labeled as Training Data Signals — they reflect historical AI training patterns, not real-time web positioning. Google AI Overview and Bing Copilot are not currently trackable; neither platform offers a public API for AI-generated answers and automated scraping is actively blocked.
How long does it take to see results?
▼
Structured data (FAQPage, HowTo, Article schema) improves how Perplexity parses and cites your content. Google has stated that schema markup is not required for and does not improve visibility in Google AI Overviews. Entity authority work — establishing your brand as a recognized, citable source across trusted ecosystems — typically takes 30–90 days to compound meaningfully in AI Share of Voice scores. ChatRank's 90-day engagements are designed to show measurable Visibility Velocity within the first 30 days and board-reportable AI Share of Voice movement within 90 days.
How is ChatRank different from my SEO agency?
▼
SEO agencies optimize for Google's crawler — a link-based ranking system that scores pages for keyword relevance and domain authority. ChatRank optimizes for LLM retrieval behavior — a fundamentally different mechanism. AI engines don't crawl and rank pages; they retrieve from compressed knowledge representations and cite sources based on entity authority, structured data clarity, and citation ecosystem strength. Ask your SEO agency what your AI Share of Voice is. If they don't know, that is the answer.
Is ChatRank software or a consulting firm?
▼
ChatRank is an advisory firm with a proprietary intelligence platform — not software that requires a separate strategist alongside it. The strategic advisory layer is part of what you engage when you work with ChatRank. The free audit is automated. Paid engagements include a human-reviewed audit, written citation-optimized content, schema implementation guides, a prioritized action roadmap, and advisory calls. The platform provides the measurement infrastructure. ChatRank provides the strategy and execution. They are one product, not two separate things.
Can AI visibility be manipulated?
▼
Short-term citation spikes can be manufactured by flooding AI training pipelines with low-quality, brand-saturated content — sometimes called AI bait. The results are real in the short term and self-defeating in the medium term. Modern AI systems are trained to recognize and discount thin, repetitive content. Model updates specifically target citation manipulation patterns. ChatRank's methodology is built on the opposite principle: coherent entity authority across trusted ecosystems, structured content that genuinely answers buyer questions, and schema markup that helps AI engines correctly parse and attribute your brand. These interventions compound over time. They are not gamed out by the next model update — they are reinforced by it.
Can competitors see my ChatRank report?
▼
No. Free audit reports are stored at a UUID-based URL with no public index. They are not discoverable via search engines and are delivered only to the email address submitted. Paid engagement reports are delivered directly and are never published without explicit written consent from the client.
Why doesn't my company show up in ChatGPT or Gemini yet?
▼
There are two different systems at work, and they move on very different timelines. Google Search indexes your website within days or weeks of publication. But ChatGPT, Gemini, and Claude are trained on snapshots of the web — not live data. When OpenAI or Google trains their next model, your content flows into it. Until then, your site can be perfectly indexed in Google Search and still invisible in AI-generated answers. This is not a failure. It is the normal gap between web publishing and LLM training cycles, which typically update every 6–12 months depending on the model provider. Perplexity is the exception — it runs live web search and can surface your content within 48–72 hours of publication, which is why it appears in ChatRank's free audit alongside ChatGPT and Claude.
How long does it take to see AEO results?
▼
Schema fixes and structured data can surface in Perplexity responses within days. Entity authority work — establishing your brand as a recognized, citable source — typically takes 30–90 days to compound meaningfully in AI Share of Voice scores on real-time platforms like Perplexity. Movement in ChatGPT and Gemini is tied to model training cycles, which update periodically — typically every 6–12 months. ChatRank's 90-day engagements are designed to show measurable Visibility Velocity within the first 30 days and board-reportable AI Share of Voice movement within 90 days. The brands establishing AI visibility signals today are the ones appearing in AI answers when the next training cycle completes.
The audit takes
60 seconds.
It's free.
Gartner predicts 25% of traditional search traffic disappears by 2026. Your AI visibility gap is either growing or closing right now. The audit tells you which.
GET MY FREE AUDIT →