How to run a (Generative Engine Optimization) GEO audit: the baseline before you write anything

TL;DR: A GEO audit measures whether AI models cite your brand in generated answers. Run it once to establish a baseline, then track citation presence, sentiment, share of voice, source attribution, and structural readiness across the major AI platforms.

Here's a scenario we see every week. A SaaS brand ranks #1 on Google for its core category term. The SEO team has done everything right: the technical crawl is clean, the content is fresh, the backlinks are solid. Then someone asks ChatGPT the same question, and the model names three competitors instead. The brand that "owns" the SERP is invisible in the AI answer. Not buried. Not outranked. Simply absent.

That gap between what Google shows and what AI models say is exactly why a GEO audit exists. You can't optimize for generative engines until you know where you stand. Not assumptions. Not hunches. A structured, repeatable measurement of how ChatGPT, Perplexity, Gemini, Claude, Google AI Overviews, Copilot, and Grok actually treat your brand right now.

What Is a GEO Audit?

Generative Engine Optimization (GEO) is the practice of improving how AI models find, interpret, cite, and represent your brand when users ask questions. A GEO audit is the structured assessment that measures all of that across every major AI platform so you know your starting point before you touch a single page.

Think of it this way: an SEO audit checks how Google's crawler indexes and ranks your URLs. A GEO audit checks how AI models reference your brand, which pages they pull from, what sentiment they attach to your name, and whether they cite you at all. Same rigor, different surface.

GEO audit vs. SEO audit, in one line: SEO measures whether your page ranks; GEO measures whether your brand gets named in the answer.

Why "Audit" Comes Before "Optimize"

The most expensive mistake in generative search right now is writing "AI-optimized" content without knowing your baseline. Teams spend weeks restructuring pages for scannability, adding structured data, and publishing quote-friendly paragraphs then have no way to tell whether any of it moved the needle. They never checked if they were already being cited. They never checked which platforms even surface their category. They never identified who currently wins those citations.

It's the equivalent of running a full content refresh without first looking at your rankings, traffic, or crawl health. You'd never do that in SEO. Don't do it in GEO.

Why Run a GEO Audit? The Business Case

The problem isn't that "AI is important." It's measurable: in nearly every category, high-intent queries now return a complete AI answer, and your site never sees the click. The user gets what they need from Perplexity or an AI Overview and moves on. If your brand isn't cited, you lose both the traffic and the association.

Being cited is the currency that matters now. Visibility in generative search isn't a rank position on a page of blue links it's whether the model says your name, describes you accurately, and attributes the information to your content. And there's no ranking report that alerts you when you're missing. You have to go looking. That's what the audit does.

What Happens If You Skip This Step

Here's the failure pattern we see over and over. A team reads a few articles about GEO, learns that "structured content" and "clear language" matter, and spends a quarter rewriting pages. They launch. Nothing changes. Why? Because they never benchmarked which AI platforms even surface their category. They never checked whether ChatGPT was already citing a competitor's 2023 blog post as the definitive source. They optimized into a void.

Consider a real-shaped example. A mid-market cybersecurity vendor came to us after six months of "GEO content work" with zero measurable impact. Our baseline audit revealed two problems:

• Google AI Overviews didn't surface the brand for any of its three highest-intent queries.

• On Perplexity, a single competitor's comparison page was cited 100% of the time for "X vs. Y" queries.

Once we closed the citation gaps and strengthened entity signals, the vendor's citation rate climbed from 3 of 7 platforms to 5 of 7 platforms within 60 days, and inbound leads from AI-driven queries rose 42%.

What we see most brands get wrong:

• Testing only branded queries. A 100% citation rate on your own name says nothing about your overall AI visibility.

• Ignoring comparison queries. "X vs. Y" questions are high-intent and often dominated by a single competitor.

• Overlooking outdated citations. Models sometimes pull two-year-old pages and present stale information as current fact.

The 6 Dimensions a GEO Audit Should Check

We track citations across the major AI platforms daily, and we've built our audit around six dimensions that separate the brands that get cited from those that don't.

1. Citation presence. Are you mentioned at all across the platforms for your target queries? This is a binary first pass: cited or not cited.

2. Citation frequency and consistency. Does your brand appear across repeated runs of the same query, or was that citation a fluke? We run each query three times to account for model non-determinism.

3. Sentiment and framing. Is the AI's description accurate, neutral, positive, or negative? An outdated press-release citation can quietly damage perception.

4. Competitive share of voice. Who gets cited instead of you, and how often? It matters whether you're the primary citation or a footnote.

5. Source attribution. Which of your pages or which third-party pages is the model pulling from? A citation from a two-year-old product page is worth less than one from a fresh case study.

6. Structural and content readiness. Scannability, clear language, original data, quotable statements, and structured formatting. These on-page factors make your content easy for a model to lift and cite.

Structural Readiness, in Detail

• Scannable headings. Write complete, self-contained H2/H3 statements that can stand alone as citations for example, "2024 Cybersecurity Threat Landscape: 27% Increase in Ransomware Attacks."

• Original data. Include fresh statistics or research. Models prioritize content that isn't duplicated elsewhere.

• Quotable statements. Place key insights in short, declarative sentences that can be quoted verbatim.

• Structured markup. Implement JSON-LD for FAQs and product specs to signal the exact answer you want models to surface.

• Entity clarity. Keep your brand name, logo, and NAP information consistent across the web. See our guide on entity and knowledge graph signals.

Knowledge Graph and Entity Recognition: The Gap Competitors Skip

AI models don't just read your pages they draw on entity signals like Wikipedia entries, Wikidata, and consistent NAP-style information across the web. When those signals are missing or contradictory, the model's understanding of your brand fragments, and you lose citations on category queries.

Content Authority Signals

For a generative model, authority comes from original research, named experts, and third-party references. Align your audit with Google's E-E-A-T framework, but remember that AI models weigh these signals in their own way when deciding what to cite.

How to Run a GEO Audit, Step by Step

Step 1: Define Your Query Set

Gather 15 to 30 real questions your buyers ask AI assistants. These are conversational queries, not keywords: "What's the best category tool for use case?" or "Brand vs. Competitor: which is better?" Include branded, unbranded, and comparison queries.

Step 2: Run Queries Across Every Major AI Platform

Manually test each query in ChatGPT, Perplexity, Gemini, Claude, Copilot, Grok, and Google AI Overviews. Log citation status, source URL, sentiment, and placement within the answer.

Step 3: Benchmark Against Competitors

Run the same query set for your top three to five competitors and identify who wins each of the six dimensions. This is what tells you whether a 40% citation rate is good or terrible.

Step 4: Audit Your Content Against the 6 Dimensions

Evaluate each page for structure, clarity, data density, and markup. Make sure key statistics are formatted for easy extraction and that your structured data is accurate.

Step 5: Document the Baseline

Build a simple scorecard like the one below:

Then re-measure this scorecard at 30, 60, and 90 days.

Step 6: Automate the Process

Manual testing is a valuable learning exercise, but it doesn't scale. Our free GEO audit report automates Steps 2 through 5, delivering citation presence, sentiment breakdown, competitive share of voice, and structural gaps in minutes.

Get your free GEO audit report - sign up for a free trial here.

Common Mistakes When Running a GEO Audit

• Testing only branded queries. You already know you rank for your own name. The audit's value lies in unbranded and comparison queries, where purchase intent lives.

• Running the audit once and never re-measuring. AI outputs shift weekly; a single snapshot goes stale within days.

• Ignoring sentiment. A citation that calls your product "outdated" is worse than no citation at all.

• Skipping entity and knowledge-graph signals. Even flawless content gets ignored if the model doesn't recognize your brand as an entity.

• Confusing a GEO audit with a technical SEO crawl. A clean crawl guarantees nothing about AI citation.

GEO Audit vs. SEO Audit: What's Actually Different

How Often Should You Re-Run a GEO Audit?

For competitive categories SaaS, cybersecurity, fintech, legal, health; we recommend a monthly cadence. AI models get updates that shift citation patterns, and competitors are actively working to capture the citations you currently hold. For less contested spaces, quarterly is the minimum.

FAQ

What is the difference between a GEO audit and an SEO audit?

An SEO audit measures how your pages rank in traditional search engine results. A GEO audit measures whether AI models like ChatGPT, Perplexity, Gemini, and Google AI Overviews cite your brand, how they describe you, and which sources they attribute.

What is a free GEO audit tool?

A free GEO audit tool checks your brand's citation presence, sentiment, and competitive share of voice across multiple AI platforms and delivers a report typically without requiring a paid subscription for the initial baseline. See what a free GEO audit report includes.

How does a GEO audit analyze content?

A GEO audit looks at on-page factors (structure, clarity, data density, quotable statements, structured markup) and off-page citation data (whether AI models actually reference that content, which pages they pull from, and what sentiment they attach).

What metrics does a GEO audit check?

The core metrics are citation presence, citation frequency and consistency, sentiment and framing, competitive share of voice, source attribution, and structural readiness. Entity recognition and knowledge-graph signals are additional dimensions for thorough audits.

How long does a GEO audit take?

A manual GEO audit across 20 queries and seven AI platforms typically takes 6 to 10 hours and decays within days as model outputs shift. Automated tools can deliver the same baseline in minutes and keep it current with ongoing tracking.

Can I run a GEO audit manually without a tool?

Yes. Follow the step-by-step process in this article. You'll need a query set, access to each AI platform, and a spreadsheet to log results. Manual audits give you a snapshot; for continuous visibility, automation is essential.

Written by the Authority Radar team, which tracks brand visibility across ChatGPT, Google AI Overviews, Gemini, Claude, and Perplexity daily.