How to Run a Free AI SEO Audit Using Claude Code (And Why Traditional SEO Audits Are Missing the Point)

Half your future customers will find you through an AI, not a search engine. Is your website ready for that?

I run an agency. We build websites, manage brands, produce content. And for years, our SEO process looked like everyone else's: run a Screaming Frog crawl, check Search Console, pull an Ahrefs report, fix meta tags, build backlinks, wait. It worked. Clients ranked. Everyone was happy.

Then sometime around mid-2025, I started noticing something in our clients' analytics. A new traffic source was growing fast. Not Google organic. Not social. It was referral traffic from ChatGPT, Perplexity, and Google's AI Overviews. People were asking AI a question and clicking through to our clients' sites — or more often, getting their answer directly from the AI without clicking at all.

That second part is the one that should concern you.

When someone asks Perplexity "best marketing agency in DC" and your competitor gets cited in the AI-generated answer, it doesn't matter that you rank #3 on Google. You're invisible in the conversation that actually happened.

This is GEO — Generative Engine Optimization — and it's the biggest shift in search since mobile. Here's how we audit for it now, and how you can do the same thing for free.

What Is GEO and Why Should You Care Right Now

GEO stands for Generative Engine Optimization. It's the practice of optimizing your website and brand presence so that AI-powered search engines — ChatGPT, Perplexity, Google AI Overviews, Gemini, Bing Copilot — cite, reference, and recommend you when answering user questions.

Traditional SEO asks: "How do I rank on page one of Google?" GEO asks: "When an AI answers a question about my industry, does it mention my brand?"

The numbers make the urgency clear:

That last number is the one I keep coming back to. Only 23% of marketers are even thinking about this. If you start now, you're not playing catch-up — you're establishing a lead.

And here's a data point from Ahrefs' December 2025 study that changed how I think about authority: brand mentions are 3x more powerful than backlinks for AI citation. Three times. All those years we spent obsessing over link building, and it turns out that for AI search, simply being talked about — on Reddit, on YouTube, in industry publications — matters more than any individual backlink.

Why Traditional SEO Audits Miss the GEO Picture

I still run traditional SEO audits. They're not useless. But they're incomplete in a way that's becoming dangerous. Here's what a standard SEO audit checks: page speed, meta tags, header hierarchy, broken links, mobile responsiveness, backlink profile, keyword rankings. All valuable. None of it tells you whether an AI can even access your site.

Most traditional audits don't check:

• Whether your robots.txt is blocking AI crawlers like GPTBot, ClaudeBot, or PerplexityBot
• Whether you have an llms.txt file (the emerging standard for AI-readable site descriptions)
• Whether your content is structured in a way that AI systems can confidently quote
• Whether your brand exists on the platforms AI models actually pull from — Reddit, Wikipedia, YouTube, industry forums
• Whether your schema markup is detailed enough for AI systems to understand your entity relationships

I audited our own agency site with a traditional SEO tool last month and got a 91/100. Looked great. Then I ran a GEO audit and scored 68. The gap between those two numbers is exactly the blind spot I'm talking about.

How to Run a GEO Audit With Claude Code

Claude Code is Anthropic's command-line interface for Claude. It's built for developers, but you don't need to be one to run a GEO audit. The tool includes a set of built-in GEO analysis skills that run 5 parallel agents against your site and produce a scored report. Here's how to do it.

Step 1: Install Claude Code

If you already have Node.js installed, it's one command:

npm install -g @anthropic-ai/claude-code

Then launch it:

claude

You'll need a Claude account. Once you're authenticated, you're ready.

Step 2: Run the GEO Audit

Inside the Claude Code prompt, type:

/geo audit https://yoursite.com

That's it. The tool spins up 5 parallel analysis agents that each examine a different dimension of your site's AI readiness. It typically takes 3-5 minutes to complete a full audit. You'll see progress output as each agent finishes its work:

# Running GEO audit for https://yoursite.com...

[Agent 1/5] AI Visibility & Citability  ........... done
[Agent 2/5] Platform Analysis  .................... done
[Agent 3/5] Technical Infrastructure  ............. done
[Agent 4/5] Content Quality & E-E-A-T  ........... done
[Agent 5/5] Schema & Structured Data  ............ done

────────────────────────────────────────
COMPOSITE GEO SCORE: 68 / 100
────────────────────────────────────────

There are also individual commands if you want to drill into specific areas:

/geo citability https://yoursite.com    # How quotable is your content for AI?
/geo crawlers https://yoursite.com      # Which AI bots can access your site?
/geo schema https://yoursite.com        # Audit your structured data markup
/geo llmstxt https://yoursite.com       # Check or generate your llms.txt
/geo content https://yoursite.com       # E-E-A-T and content quality scoring

Step 3: Understanding the Results

The audit produces a composite score from 0 to 100, broken down across 6 weighted categories. Here's what each measures and how heavily it factors into your final score:

A score above 80 means you're well-positioned for AI search. Between 60-80, you have the foundations but significant gaps. Below 60, AI search engines are probably ignoring you entirely.

Step 4: Quick Wins to Improve Your Score

The audit report includes prioritized recommendations. But in my experience running these across dozens of sites, the biggest point gains come from a handful of changes that take less than an hour:

1. Add an llms.txt file to your root domain. Claude Code can generate one for you: /geo llmstxt https://yoursite.com
2. Unblock AI crawlers in robots.txt. Many sites accidentally block GPTBot and ClaudeBot.
3. Add FAQ sections to your key pages with direct, quotable answers.
4. Expand your schema markup beyond the basics — add Organization, Person, FAQPage, and HowTo schemas.

Those four changes alone typically move a score by 10-15 points.

What We Found Auditing Our Own Sites

I believe in eating your own cooking. Before we started offering GEO audits to clients, we ran the tool against our own three properties. Here are the real numbers:

Those scores stung. We're an agency that builds websites for a living, and we were sitting in the "significant gaps" range on our own properties. The traditional SEO scores for the same sites were all above 85.

That gap — between your SEO score and your GEO score — is what most businesses don't know they have. And it's growing wider every month as AI search takes more market share from traditional search.

We've since made changes to all three sites and pushed our scores up. The process of fixing our own properties is what convinced us to build GEO auditing into our client work.

The 5 Most Common GEO Issues We See

After running GEO audits on our own sites and a batch of client properties, clear patterns emerged. These five issues show up on almost every site we check.

1. No llms.txt File

This is the single most common miss. The llms.txt file is an emerging standard — think of it as robots.txt for AI models. It tells AI systems what your site is about, what pages matter most, and how to accurately describe your brand. Almost nobody has one yet. Adding it takes 10 minutes and immediately gives AI models a clearer picture of your site.

/geo llmstxt https://yoursite.com

# Output: a complete llms.txt file ready to upload to your root domain

2. AI Crawlers Blocked in robots.txt

Many sites are accidentally — or intentionally — blocking AI crawlers. Some web hosting platforms add these blocks by default. Some CMS plugins include them. The result is that GPTBot, ClaudeBot, and PerplexityBot can't read your site, which means AI search engines have to rely on secondhand information about you instead of your actual content.

Run /geo crawlers https://yoursite.com to see exactly which AI bots can and can't access your site. The output looks like this:

AI Crawler Access Map
─────────────────────
GPTBot (OpenAI)      ALLOWED
ClaudeBot (Anthropic) BLOCKED  <-- fix this
PerplexityBot        ALLOWED
Google-Extended      ALLOWED
Bytespider (TikTok)   BLOCKED
CCBot (Common Crawl) ALLOWED

3. Missing or Minimal Schema Markup

Most sites have basic schema — maybe an Organization type, maybe some breadcrumb markup. That's table stakes for regular SEO. For GEO, you need richer structured data: FAQPage, HowTo, Person (for team members and authors), Article with full author attribution, LocalBusiness with service areas. AI models use schema to understand entity relationships, and the more specific you are, the more confidently they can cite you.

4. No Brand Presence on AI Source Platforms

AI models don't just read your website. They synthesize answers from across the internet. If your brand doesn't exist on Reddit, YouTube, LinkedIn, Wikipedia, or industry-specific forums, you're invisible to a huge portion of the AI's knowledge base. This is where the Ahrefs finding about brand mentions being 3x more powerful than backlinks really hits home. You need to be talked about in places AI models actually index.

5. Content Not Structured for Citation

This one is subtle but important. AI systems prefer to cite content that contains direct, specific answers. If your blog posts are 2,000 words of meandering narrative without a single clear, quotable statement, AI won't cite you — even if the overall content is good. The fix: add FAQ sections, write clear topic sentences, include specific numbers and data points, and structure content so that any paragraph could stand alone as an answer.

Quick Wins Anyone Can Do Today

You don't need to hire an agency (though, obviously, I think you should) to make meaningful progress on GEO. Here are concrete actions you can take this afternoon:

1. Check your robots.txt right now. Go to yoursite.com/robots.txt and search for "GPTBot," "ClaudeBot," and "PerplexityBot." If any are blocked, unblock them unless you have a specific reason not to.
2. Create an llms.txt file. At minimum, it should include your company name, what you do, your primary URL, and your most important pages. Upload it to your root domain.
3. Add FAQ schema to your top 5 pages. Write 3-5 frequently asked questions per page with clear, direct answers. Mark them up with FAQPage JSON-LD.
4. Google your brand name + "Reddit." If nothing shows up, your brand is invisible to one of the biggest sources AI models reference. Start participating authentically in relevant subreddits.
5. Add author bios with credentials to all blog content. AI models weigh expertise signals heavily. A byline with "John Smith, 15 years in supply chain management" carries weight that "Posted by Admin" never will.
6. Rewrite your About page as an entity definition. Instead of marketing copy, write a clear, factual description of what your organization does, when it was founded, who leads it, and what it's known for. AI models use About pages to build entity understanding.
7. Run the free audit. Install Claude Code and run /geo audit on your site. You'll have a prioritized action list in under 5 minutes.

The Bigger Picture: SEO Is Not Dead, But It's Not Enough

I want to be clear: I'm not saying burn your SEO playbook. Good SEO fundamentals — fast load times, clean HTML, strong content, proper schema — are the foundation that GEO builds on. A site with terrible SEO isn't going to score well on GEO either.

But if your entire search strategy is still "rank on Google," you're optimizing for a world that's shrinking. AI-referred traffic grew 527% in early 2025. Gartner projects that by 2028, organic search traffic will decline 25% as AI answers replace click-through searches. The brands that invest in GEO now will own the AI search results the same way early SEO adopters owned page one of Google in 2010.

With only 23% of marketers even aware of GEO, the window for first-mover advantage is wide open. It won't stay that way.

Frequently Asked Questions

Last updated: March 28, 2026