AI Visibility Audit — LLM / GEO Visibility Audit
People increasingly ask ChatGPT, Perplexity, Gemini and Copilot instead of Google: "where to invest", "who to choose", "what's more reliable". An AI answer names not 10 links but 1–3 brands. The audit shows whether you are one of them — and exactly what stops AI engines from recommending you.
Why this matters right now
Google AI Overviews already reach millions of users, ChatGPT Search is available to everyone, and Perplexity is one of the fastest-growing search engines. More and more buying decisions start with a question to an AI engine rather than a Google search. The problem: almost nobody measures AI visibility — classic reports simply don't cover it. A brand can rank at the top of Google and be completely absent from model answers, where competitors get named instead.
The audit closes this blind spot. It answers two questions with numbers, not gut feel: is your site ready to be cited by models, and are they actually citing it — and if not, who gets named instead of you.
Methodology: two measurement layers
It's built on the AgentBridge LLM Visibility methodology. It separates "readiness" from "fact" because they are different: a site can be technically ready yet have no mentions — and vice versa.
Is the infrastructure there for you to be known
Visibility Score 0–100 across 4 dimensions and 26 sub-factors. Entity, citation, content and technical readiness — on live data.
Do models actually mention you
A prompt set run across 6 AI platforms via Brand Radar. Real mentions, "cold" SOV, citation sources.
Where you sit on the Visibility × Performance map
The result maps to a quadrant: Invisible / Sleeping potential / Fragile leader / Leader — with a clear next step.
Layer 1 — Visibility Score: 4 dimensions
Each dimension is a weighted sum of sub-factors scored 0/25/50/75/100. Formula: Visibility = Entity×0.30 + Citation×0.25 + Content×0.25 + Technical×0.20.
Entity — 30%
Does AI treat the brand as a distinct entity: Google Knowledge Graph, Wikidata/Wikipedia, Organization schema + sameAs, directories, NAP consistency, founder profile.
Citation — 25%
Citability: brand demand (GSC), mentions in media and listicles, independent reviews, citation of your own data, backlink profile.
Content — 25%
Content extractability: chunk structure, factual density, FAQ/Q&A, freshness, comparison pages, proprietary terminology, facts page.
Technical — 20%
AI-bot access (robots/GPTBot), content without JS, schema coverage, sitemap + lastmod, Core Web Vitals, meta/OG/canonical, llms.txt and future-readiness.
Layer 2 — real mentions in AI (Performance)
A set of 60+ prompts of five types is run across 6 AI platforms in your region via Ahrefs Brand Radar. Each prompt is run several times: we compute a mention rate because models respond non-deterministically.
Prompt types
Brand, Category, JTBD, Comparison, Transactional — from "what is X" to "where should I invest".
"Cold" SOV
Share of voice on non-branded prompts only — an honest signal, unskewed by branded prompts.
Citation sources
Which domains models cite in your niche — those are the targets for outreach and getting into answers.
Baseline + trend
The first run is a starting point. The value is in the trend across repeated measurements, not a one-off number.
Visibility × Performance quadrant
Two numbers — readiness and fact — combine into a single map. It shows immediately what work is needed: build infrastructure, drive up mentions, or defend a position.
Invisible
Low readiness and low mentions. Needs a program from scratch: Entity + Citation.
Sleeping potential
Readiness is there, mentions are not. The lever: extractable content and outreach to cited domains.
Fragile leader
Mentions exist but the base is weak. Risk of losing position — strengthen entity and technical base.
Leader
High readiness and stable mentions. The job is to hold and scale.
What you get
- Full Layer 1 + Layer 2 report — Visibility Score across 4 dimensions and 26 sub-factors with evidence for each item.
- Competitor benchmark — the same sub-factors for niche competitors: where you lag and what's cheapest to copy.
- Cited-domain map and link-gap — where to push to appear in model answers (the domains AI cites).
- Executive summary — a 3-minute overview for a non-technical audience: quadrant, key gaps, priorities.
- 90-day roadmap — tasks by impact priority with dependencies and KPI checkpoints.
- Prompt-set + Brand Radar baseline — the prompt set and a starting point to track the trend (extended package).
Data from live sources, not guesswork
Scores aren't invented: every sub-factor is backed by a tool check. Each item in the report comes with a confidence map — where it's precisely measured vs. a reasoned estimate (proxy).
How the audit runs
Brief + access (GSC, GA4)
Layer 1 — Visibility Score
Competitor benchmark
Layer 2 — Brand Radar
Report + roadmap
Not sure whether AI engines see you? Send your domain — I'll tell you roughly which quadrant you're in and whether a full audit is worth it.
Get in touch →Packages and pricing
- Visibility Score (4 dimensions, 26 sub-factors)
- Benchmark of 3–5 competitors
- Technical AI check (schema, robots, CWV)
- 90-day roadmap
- Executive summary
- Everything in Basic
- Real-mention measurement in 6 AIs
- "Cold" SOV + citation sources
- Prompt-set (60+ prompts)
- Brand Radar baseline for the trend
- Cited-domain map / link-gap
- Everything in Standard
- Costed promotion plan
- Dev task prioritization
- Implementation support (GEO/AIO)
- Monthly re-measurement and report
Who it's for:
AI visibility audit vs. SEO audit
| Criterion | SEO audit | AI visibility audit |
|---|---|---|
| Core question | Rankings in Google results | Whether AI engines know and recommend you |
| Where we look | SERP, 10 links | ChatGPT, Perplexity, Gemini, Copilot, AI Overviews |
| Key signal | Links, keywords, technical | Entity + Citation + extractability |
| Metric | Rankings, traffic | Visibility Score + "cold" SOV |
| Fact measurement | — | Brand Radar across 6 platforms |
FAQ
How is an AI visibility audit different from an SEO audit?
An SEO audit checks rankings in classic Google results. An AI visibility audit answers a different question: do AI engines know you and recommend your brand in their answers. We measure readiness to be cited (Layer 1) and real mentions in model answers (Layer 2 via Brand Radar).
How is the Visibility Score calculated?
Visibility Score (0–100) is a weighted sum of four dimensions: Entity 30%, Citation 25%, Content 25%, Technical 20%. Inside are 26 sub-factors, each scored 0/25/50/75/100.
What is "cold" SOV?
Aggregate Share of Voice is inflated by branded prompts. "Cold" SOV is computed only on non-branded prompts (category, JTBD, comparison) — an honest signal of whether models recommend you when the user doesn't yet know about you.
Which AI platforms are measured?
Six surfaces: ChatGPT, Perplexity, Gemini, Microsoft Copilot, Google AI Overviews and Google AI Mode — in your region, with several runs per prompt (mention rate).
How long does it take and what do you need from me?
The basic audit takes 5–7 business days; the extended one with Brand Radar up to 2 weeks. I need read access to Google Search Console and GA4 plus a short brief about the business and competitors.
Does the audit guarantee I'll appear in ChatGPT answers?
No. The audit is a diagnostic and prioritized plan, not a ranking guarantee. Models are non-deterministic and citations drift. The value is in the trend and in working by order of impact. Implementation is covered by the GEO/AIO service.
Find out whether AI engines see you
The AI visibility audit is the entry point to GEO/AIO. Numbers and priorities first, then work by impact.
Order the audit →