Building a GEO Audit: The Framework We Use for Every Client

Most brands have no idea how AI describes them. Ask ChatGPT, Perplexity, or Google's AI Overviews a question your buyers ask every day-"What's the best [your category] tool?"-and your company may not appear at all. Or worse, the AI gets it wrong: outdated pricing, a competitor's messaging, a description you wouldn't recognize.

Most brands obsess over Google rankings yet have no visibility into how AI engines perceive and present their brand. This blind spot is not minor. ChatGPT reached 800 million weekly active users by October 2025, doubling from 400 million in just eight months.

Every day, over 1 billion prompts are sent to ChatGPT, and more than 71% of Americans already use AI search to research purchases or evaluate brands. The shift from keyword rankings to AI citations is no longer emerging-it's the operating environment. A GEO audit provides the baseline. It tells you where you're visible, where you're absent, and where the AI is actively sending your buyers to a competitor. We've run this framework for dozens of clients since mid-2025, and what follows is the exact structure we use-the five-layer audit that turns AI invisibility into a prioritized 90-day action plan.

Why Traditional SEO Audits Miss What Matters Now

SEO audits were engineered for a different system. They measure crawlability, keyword rankings, backlink profiles, and page speed-all signals designed for an index that returns ten blue links. SEO audits focus on rankings and crawl/index factors, while GEO audits focus on extractability, trust/entity consistency, and off-site proof that influences AI answers.

The distinction runs deeper than metrics. AI engines synthesize information while search engines list URLs. Citation authority replaces backlinks. Conversational queries surpass rigid keyword phrases. Visibility scores become more critical than organic ranking. And the math is harsher: LLMs only cite 2–7 domains on average per response, far fewer than Google's 10 blue links. If you're not in that tight citation window, you don't exist. The measurement gap is equally stark. According to McKinsey's 2025 CMO survey, only 16% of brands today systematically track AI search performance. That means the vast majority of marketing teams are optimizing for a channel they aren't measuring-while their competitors quietly capture citation share. A GEO audit closes that gap before it becomes a revenue problem.

Layer 1: AI Visibility Benchmarking-Where You Stand Today

Every audit starts with a simple question: when buyers ask AI platforms about your category, does your brand appear?

Running the Prompt Library

Before you optimize, get a baseline. Manually test conversational, high-intent queries related to your products and industry in generative AI tools like ChatGPT, Gemini, and Perplexity. Document every instance where your brand is cited, a competitor is mentioned, or the information is incorrect.

We build a prompt library of 30–50 queries that mirror how real buyers search. These aren't keyword strings. Buyers do not type keywords into ChatGPT. They ask natural language questions like "What's the best revenue operations platform for a mid-market SaaS company?" or "Compare [Brand A] vs [Brand B] for enterprise deployment." We segment prompts by funnel stage-awareness, consideration, and decision-then run each across ChatGPT, Perplexity, Claude, Gemini, and Google AI Overviews.

Scoring What You Find

For each prompt, we track four things: whether the brand is mentioned, whether it's cited with a link, the position in the response, and the accuracy of the description. Citation rate measures how often your domain or page is cited as a source in AI answers. Mention rate measures how often your brand name appears in those answers. A page can be cited without the brand being named, which means your content influences the answer but your brand gets no recognition.

This produces a brand visibility score we can track month over month. How often does your brand appear when buyers ask relevant questions? If you test 20 prompts and your brand appears in 12 responses, you have a 60% brand visibility score. This is the foundational AI visibility metric. Tools like Otterly.AI, Profound, Peec AI, and Semrush's AIO module can automate parts of this process at scale. But the first audit should always include manual testing-automated tools miss nuance in how AI platforms frame your brand.

Layer 2: Technical Crawlability-Can AI Even Find You?

Visibility problems often start with a locked door. Many websites are blocking AI crawlers without realizing it-and then wondering why competitors show up in ChatGPT while they don't.

Robots.txt and AI Crawler Access

The first technical check is straightforward: does your robots.txt allow AI crawlers to access your content? If you want your content to show up in AI-powered search results, get referenced by ChatGPT, or appear in Google's AI Overviews, you need to roll out the red carpet for AI crawlers. Many websites are accidentally blocking these helpful bots without even realizing it.

The key user agents to explicitly allow include GPTBot and ChatGPT-User (OpenAI), ClaudeBot (Anthropic), Google-Extended (Google's AI features), and PerplexityBot. When tested, ChatGPT-User fetched the robots file and stopped crawling when it was disallowed, with no follow-up crawls from other user agents or third-party bots. OpenAI respects the directive. Block GPTBot, and you disappear from ChatGPT. But robots.txt isn't the only barrier. Watch out for server-side bot detection (some WAFs block AI crawlers based on user-agent strings), rate limiting that returns 429 errors, and JavaScript-rendered content that some AI crawlers won't see. We've found Cloudflare's Bot Fight Mode blocking legitimate AI crawlers on multiple client sites. Since July 2025, Cloudflare prevents AI bots from crawling customer websites by default. If you're on Cloudflare and haven't explicitly whitelisted AI bots, you're almost certainly invisible to some platforms.

Structured Data That Helps AI Understand You

Schema markup serves as a direct translation layer for AI systems. Google's Search team confirmed in April 2025 that structured data gives an advantage in search results, and Microsoft's Fabrice Canel confirmed in March 2025 that schema markup helps Microsoft's LLMs understand content for Copilot.

The priority schema types for any GEO audit include Organization (brand entity identity), Article (content attribution and freshness), FAQPage (question-answer extraction), Person (author credentials), and Product or Service schemas where applicable. Use JSON-LD format, place your code properly, and validate everything.

One nuance that matters: A December 2024 study found no correlation between schema markup coverage and citation rates in isolation. Sites with comprehensive schema didn't consistently outperform sites with minimal or no schema markup. LLM systems appear to prioritize relevance, topical authority, and semantic clarity over whether content has structured markup. Schema is necessary infrastructure, not a silver bullet. It helps AI engines accurately interpret and attribute your content-but only when that content is worth citing in the first place.

Layer 3: Content Extractability-Is Your Content Citable?

AI doesn't rank your page. It extracts from it. AI systems don't "rank" your page-they extract from it. That means the question isn't whether your content is "good" in a human editorial sense. The question is whether AI can pull specific, verifiable claims and present them as trustworthy answers.

Structural Requirements for Citation

The core principles of effective GEO content include structuring content with direct answers in the first 40–60 words, maintaining fact density with statistics every 150–200 words, citing authoritative sources throughout, and implementing proper schema markup. This isn't a stylistic preference-it's how retrieval-augmented generation works. AI systems break pages into passages and evaluate each for relevance, clarity, and factual density. We audit the top 10–20 pages on each client's site against these structural criteria. Pages that bury the answer beneath three paragraphs of preamble fail extraction. Pages that make claims without supporting data get skipped in favor of competitors who provide specifics. Tables, comparison charts, and FAQ sections consistently outperform narrative prose for AI citation because they present information in structured, comparable formats.

Freshness as a Citation Signal

Recency plays a direct role in citation selection. Half of all citations come from content published within the last 11 months. Generative AI models primarily rely on non-paid sources for information, with earned media alone accounting for 82% of all citations. We check every audited page for visible "Last updated" timestamps, current statistics, and dateModified properties in Article schema. GEO content should be updated every 7–14 days based on observed citation decay patterns. Content without freshness signals loses citation priority after approximately 14 days.

A page published in 2024 with no updates will lose to a 2026 article on the same topic, regardless of domain authority. We flag every cornerstone page older than 90 days as a priority refresh.

Layer 4: Entity Clarity and Off-Site Authority

This is where most GEO audits-and most GEO advice-fall short. Teams spend all their energy on on-site optimization and ignore the single largest factor in AI citation: off-site authority and entity consistency.

The Earned Media Imperative

The research here is unambiguous. Researchers from the University of Toronto ran large-scale controlled experiments across multiple AI search platforms and found AI search exhibits a "systematic and overwhelming bias towards Earned media - third-party, authoritative sources - over Brand-owned and Social content." This isn't a marginal effect. Muck Rack analyzed more than one million AI citations and found 82% came from earned media sources and 94% from non-paid sources.

You can't buy eyeballs in LLMs the way you can with PPC in search engines. Up to 90% of citations that drive brand visibility in LLMs can come from earned media. This has profound implications for audit prioritization. A client with pristine on-page SEO but zero third-party coverage in trusted publications will consistently lose citations to a competitor with mediocre technical SEO but strong press coverage and community presence. We audit earned media footprint by tracking brand mentions across high-authority publications, industry outlets, review sites, and community platforms. ChatGPT cites most frequently from Wikipedia, Reddit, and Forbes. Google AI Overviews cites most from Reddit, YouTube, and Quora. Perplexity cites most from Reddit, YouTube, and Gartner. Community platforms where real users have unscripted conversations dominate every citation source list.

Entity Consistency Across the Web

Entity clarity is the degree to which AI systems can unambiguously identify your brand, understand its category, differentiate it from similar entities, and accurately describe its core offerings. When your brand identity is inconsistent across the web, AI answers become inconsistent-or your brand gets dropped entirely. Our entity audit checks five things:

• Knowledge Graph presence: Does a Knowledge Panel appear when you search your brand?

If it does not, the foundational signals are missing. Establishing Knowledge Graph presence requires consistent NAP data across business directories, a claimed Google Business Profile, and structured data on your website.

• NAP consistency:

AI search engines use NAP for entity verification, cross-referencing multiple sources before deciding to recommend a business. We check Foursquare, Yelp, Google Business Profile, Apple Maps, Bing Places, LinkedIn, and Crunchbase for exact matches. - Brand description alignment: Write one canonical 2–3 sentence company description. Post it verbatim on your website, LinkedIn, Crunchbase, G2, and anywhere else your brand appears. Consistency signals to AI that all these profiles refer to the same entity.

• Wikipedia and Wikidata entries:

Brands with Wikipedia articles receive significantly higher citation rates from AI engines because Wikipedia is a primary training data source for large language models.

• sameAs schema properties: Linking your brand entity to authoritative external profiles through sameAs attributes helps AI systems resolve your identity across sources.

One client's entity audit revealed 14 platforms with inconsistent brand descriptions. Three listed outdated service offerings, two had incorrect founding years, and five used different category classifications. After standardizing entity information across all platforms, AI engines' descriptions of the brand became measurably more accurate.

Layer 5: Competitive Citation Gap Analysis

No audit is complete without understanding where competitors are winning. We run the same prompt library against the top 3–5 competitors and map every citation gap.

What to Measure Against Competitors

Share of voice is the core competitive metric. Share of voice in AI search answers a simple question: when buyers ask AI systems for help, which brands show up most often? Unlike traditional SEO competitor tracking, this analysis focuses on brand mentions inside AI answers rather than keyword rankings.

We calculate it as: Share of Voice (%) = (Your Brand's Mentions ÷ Total Mentions of All Brands) × 100 across the full prompt library. A brand appearing in 15 out of 100 category prompts holds a 15% share. If the top competitor appears in 45, the gap is visible-and quantifiable. Beyond share of voice, we analyze why competitors get cited. Competitors who have more reviews, directory dominance, case results visibility, or media coverage will appear more frequently in AI responses regardless of on-page quality. The competitive audit reveals whether the gap is technical (they have better schema), structural (their content is more extractable), or reputational (they have stronger earned media).

Building the Priority Matrix

Each finding from all five layers feeds into a prioritized action matrix. We categorize every issue as Critical (blocking citations entirely-like robots.txt blocks or missing entity signals), High (weakening citation probability-like outdated content or missing schema), or Medium (limiting optimization-like incomplete FAQ sections or thin third-party coverage). The client walks away with three deliverables: a visibility baseline they can measure against quarterly, a ranked list of fixes with estimated effort and impact, and a 90-day roadmap that sequences technical fixes first, content optimization second, and earned media campaigns third.

Turning the Audit Into a Repeatable System

A one-time audit is useful. A repeatable system is transformative. The audit framework described here should run quarterly, with continuous prompt monitoring in between.

AI citation patterns can shift by over 30% in a single month due to model updates and competitor content changes. Weekly monitoring of core commercial prompts catches drops before they cascade into pipeline impact. Tools like Otterly.AI for prompt-level tracking, Profound for citation analysis at scale, or Semrush's AIO module for integrated SEO+GEO reporting make this operationally feasible. The measurement stack for ongoing tracking should include citation rate and mention rate across platforms, AI share of voice versus competitors, brand sentiment in AI responses, AI referral traffic in GA4, and conversion rates from AI-sourced visitors. Research from Ahrefs shows that AI search traffic converts at a rate 23 times higher than conventional search engine visits-AI search accounts for just 0.5% of total website visits, yet these visitors generated 12.1% of all signups. Small absolute numbers, enormous conversion advantage. GEO auditing isn't a one-time project and it isn't optional. The brands investing in systematic AI visibility measurement right now are building compounding advantages. Treating GEO as a one-time content tweak is the biggest mistake you can make. In reality, GEO demands the same ongoing discipline as SEO. Every model update reshuffles citation patterns. Every competitor's press release or product launch shifts share of voice. The framework described here gives you the instrumentation to detect those shifts, diagnose their causes, and respond before your buyers notice the change. The audit is the starting line. What matters is what you build after it.