How AI search works —
and what we do about it.
Understanding the system is the first step to winning in it. Here's what happens when someone asks ChatGPT, Perplexity, or Google a question about your industry — and how we make sure you're part of the answer.
What happens between the question and the answer.
When someone types a question into ChatGPT or Perplexity, the AI doesn't generate an answer from memory alone. It follows a process called Retrieval-Augmented Generation — RAG for short. Three steps, every time:
Retrieve
The AI searches the web (or its index) for content relevant to the question. Different platforms search different sources. ChatGPT uses Bing. Perplexity has its own index of 200 billion+ URLs. Google AI Overviews pull from their existing search results.
Synthesize
The AI reads the retrieved content and writes a response, combining information from multiple sources into a single coherent answer.
Cite
The AI attributes its answer to specific sources. These citations are the new visibility. If your site is cited, you exist in the answer. If it's not, you don't.
Each platform runs this process differently. ChatGPT leans on Wikipedia and established authority sites. Perplexity favors Reddit and content updated within the last 30 days. Google AI Overviews mostly cite pages from their existing top 10 organic results. Only 11% of websites cited by ChatGPT also get cited by Perplexity for the same topic.
One optimization strategy doesn't cover all platforms. That's why we measure each one independently.
The GEO Audit: a full diagnostic across 7 AI platforms.
Before we recommend anything, we measure everything. A GEO Audit tests how your site actually performs across ChatGPT, Perplexity, Google AI Overviews, Gemini, Copilot, Claude, and Grok — using 70 distinct scoring criteria tuned to each platform's source selection logic.
The result is a GEO Score from 0 to 100, broken into five dimensions. Each dimension reflects a different aspect of what AI platforms evaluate when deciding who to cite.
Brand & Entity
30% of your scoreThis is the single strongest predictor of AI citation. AI platforms don't just evaluate your website — they check whether your brand exists across the platforms they trust. We scan YouTube, Reddit, Wikipedia, LinkedIn, GitHub, and G2 — each weighted by its measured correlation with AI citations.
Here's a finding that surprises most people: an Ahrefs study of 75,000 brands found that YouTube presence correlates 2.8x more strongly with AI visibility than your backlink profile. YouTube correlation: 0.737. Domain Rating correlation: 0.266. If your GEO strategy is focused on backlinks, it's focused on the wrong signal.
Content Quality
25% of your scoreExperience, Expertise, Authoritativeness, Trustworthiness. Google's quality guidelines have used this framework for years, and AI platforms have adopted similar signals. We evaluate whether your content shows genuine expertise — author credentials, original research, source citations, topical depth — not just keyword coverage.
Content freshness is the only signal with causal evidence across multiple AI platforms. A SIGIR study confirmed that recently updated content gets preferential citation across 7 different LLMs. We also analyze content gaps: topics your competitors cover that you don't, and FAQ opportunities that would make your content more citable.
AI Citability
23% of your scoreThis is the core question: can AI extract a clear, quotable answer from your content? We don't score this with a vague "looks good." We analyze every content passage on 20+ individual signals — definition patterns, answer placement, sentence structure, statistical density, self-containment. Each signal adds or deducts specific points, so you see exactly what's working and what isn't.
The research behind this is specific. Georgia Tech found that definition patterns increase citation rates by 2.1x. Princeton showed that statistics boost AI citation by 40%. IIT Delhi found that authoritative quotations increase citation by 115%. We built these findings into measurable scoring criteria.
AI Discoverability
12% of your scoreCan AI crawlers actually find and access your content? This dimension covers the mechanical prerequisites for citation: server-side rendering (6 out of 8 AI crawlers can't execute JavaScript), crawler access permissions for 12 AI bots classified into three strategic tiers, sitemap health, and schema markup quality.
Schema markup has a counterintuitive finding: rich, specific schema helps, but generic schema actually hurts. Sites with generic markup score 41.6% versus 59.8% for sites with no schema at all. We also check for llms.txt — a protocol proposed in September 2024 that tells AI systems exactly what your site is about. Fewer than 5% of websites have one.
Technical Foundation
10% of your scoreTechnical factors are necessary conditions, not differentiators. Once a site passes basic technical hygiene, additional improvements show diminishing citation returns. Page speed has a weak negative correlation with AI visibility (ρ=-0.12 to -0.18) — it's a floor constraint, not a growth lever.
We check crawlability, security headers, internal linking structure, and page speed. Severe failures (no HTTPS, CSR-only rendering) are citation killers. But if your technical foundations are sound, the other four dimensions are where the real leverage lives.
Five dimensions. One GEO Score.
Does your brand exist on the platforms AI trusts? The strongest predictor of AI citation.
Does your content demonstrate genuine expertise, freshness, and authority?
Can AI extract clear, quotable answers from your content?
Can AI crawlers find, access, and parse your site?
Does your site pass baseline technical checks? A floor constraint, not a differentiator.
Each dimension is scored independently. Your total GEO Score is the weighted average across all five.
What does a GEO audit measure?
A GEO audit evaluates your website's visibility across 7 AI search platforms — ChatGPT, Perplexity, Google AI Overviews, Gemini, Copilot, Claude, and Grok — using 70+ scoring criteria organized into 5 dimensions: Brand & Entity (30%), Content Quality (25%), AI Citability (23%), AI Discoverability (12%), and Technical Foundation (10%). Each dimension produces a score from 0 to 100, weighted to produce a composite GEO Score. According to Ahrefs' analysis of 75,000 brands, branded web mentions correlate roughly 3× more strongly with AI citations than traditional backlinks — which is why Brand & Entity carries the highest weight. The audit tests your content against each platform's actual citation behavior, because only about 11% of domains cited by one AI platform also get cited by another.
How is a GEO audit different from an SEO audit?
An SEO audit measures how well your site ranks on Google. A GEO audit measures whether AI search engines cite you in their answers — a fundamentally different question. Traditional SEO metrics like Domain Authority and keyword rankings tell you nothing about AI visibility. A site ranking first on Google may still be invisible to ChatGPT if its content isn't structured for passage extraction, or invisible to Perplexity if it hasn't been updated recently. The Aivarize GEO audit scores five dimensions that don't exist in any SEO dashboard: entity recognition across knowledge graphs, passage-level citability, AI crawler access permissions, and platform-specific source preferences. Research from Profound's analysis of 680 million citations found that 40–60% of cited domains change monthly for identical queries — GEO visibility is volatile and requires ongoing measurement, not a one-time check.
Targeted fixes based on what the audit finds. Not a generic checklist.
Every site has different problems. A B2B SaaS company might score well on content but have every AI crawler blocked. A media company might have great technical foundations but zero brand presence on YouTube. That's why we don't sell pre-packaged bundles.
Your audit determines which fixes you need. Here are the seven we offer:
AI Crawler Access
Your robots.txt and meta tags configured so the right AI crawlers can read your site while training-only bots are blocked. A small change with immediate impact. Sites that unblock GPTBot and PerplexityBot become eligible for citation overnight.
Schema Markup
Complete JSON-LD structured data, ready to deploy. We have 12 templates covering Organization, Local Business, Article, E-Commerce, SaaS, FAQ, Professional Service, and more — each built with AI-specific properties like speakable markup and comprehensive sameAs links.
Content Optimization
The biggest lever. We rewrite and restructure your highest-impact pages for AI citability: clear definition patterns, answer-first structure, statistical density, self-contained passages. This directly impacts 48% of your GEO Score (AI Citability 23% + Content Quality 25%). We also map content gaps, fix internal linking issues, and optimize images.
Brand Authority Strategy
A roadmap for building presence on the platforms AI trusts. YouTube channel strategy, Reddit engagement plan, Wikipedia notability assessment, LinkedIn optimization, cross-platform sameAs link consolidation. We prioritize by correlation with AI citations, not vanity metrics.
Platform-Specific Optimization
Custom action plans for the 2–3 AI platforms where you're weakest. What works for Google AI Overviews (question-based headings, FAQ sections, table formatting) is different from what works for ChatGPT (Wikipedia entity work, Bing visibility) or Perplexity (Reddit presence, 30-day content freshness).
Technical SEO
Security headers, meta tag corrections, sitemap health, redirect cleanup, mobile optimization, and SSR assessment. If your site needs a server-side rendering migration, we scope that as a separate project.
llms.txt Creation
A complete llms.txt file for your site root, with automatic page categorization, per-page descriptions, and a deployment guide. One of the fastest ways to improve AI discoverability.
GEO isn't a one-time fix. AI responses change every month.
40–60% of the sources AI platforms cite shift every month. A competitor publishes new content. A platform updates its ranking algorithm. Your carefully optimized page gets replaced by someone else's.
That's why ongoing monitoring isn't optional — it's how you protect the gains you've made.
Monitoring Retainer
Monthly re-audits across all 7 AI platforms, competitor tracking, AI share-of-voice measurement (what percentage of AI responses in your category mention you), professional reporting, and a monthly advisory call. You get the data and recommendations. Your team implements.
Growth Retainer
Everything in Monitoring, plus we do the work. Five pages optimized per month, three new content briefs, schema and llms.txt maintenance, and a monthly strategy call. For teams that want GEO handled, not just tracked.
"Your GEO score went from 38 to 67, your AI citation share grew from 5% to 22%, and AI referral traffic generated 14 leads this month." — That's the kind of monthly reporting we deliver.
Start with a score.
Everything begins with understanding where you stand. Our free GEO Quick Scan gives you a score from 0 to 100 in about 30 seconds — your biggest strength, your biggest gap, and three quick wins.
No signup. No sales call. Just a clear picture of where you stand.
Want the full picture? Learn about the GEO Audit →