Technical SEO Factors Affecting Your AI Visibility Score: What Means for ChatGPT

Your site might rank on Google and still be completely invisible to ChatGPT, Perplexity, and Google AI Overviews — because LLMs read sites differently from search engines, and most Web3 sites are not built for it. We built a free tool that checks every technical factor that affects whether AI systems can read, classify, and cite your site:

run it here:https://victoriaolsina.com/is-your-website-invisible-to-ai/ 

and get your score in 30 seconds.

This is the first of four interconnected layers in an LLM SEO strategy. This post covers the Technical layer — how your site’s architecture and structure affect whether AI systems can even read and understand your content. The other three layers are: Content (what types of information AI actually reuses), Authority (what signals models trust when choosing sources), and Reinforcement (how distribution and corroboration affect model confidence). All four layers work together.

This post explains what every factor in that score actually means, why it matters for LLM visibility specifically, and what to do when it is failing.

Understanding your technical foundation is the first step toward AI visibility. If you’re unsure whether your site is optimized for how AI systems crawl and understand Web3 content, you can run a quick audit using our free AI visibility checker. If you need professional help analyzing and fixing these technical blockers, our LLM SEO service includes a comprehensive technical assessment that covers every factor affecting AI system visibility.

These ten checks diagnose individual signals. For the complete architecture behind them, use the technical SEO framework for Web3 and blockchain sites.

Key points from the video

• Content inside JavaScript. Models read the HTML before scripts run, so half the page loads as empty code.
• Multiple H1 tags. One animation split a headline into nine H1s, so the model saw nine page topics and trusted none.
• Weak title tags. A title reading Home then your brand name tells a machine nothing about what you do.
• No schema markup. Without it the model guesses your category instead of knowing it, and guessing kills citations.
• No llms.txt. The file guides AI crawlers to your best pages, and almost no competitor has one yet.
• robots.txt blocking AI bots. Allow Googlebot, block GPTBot, and you vanish from ChatGPT by design.
• Score under 50 and the model never reads you, so it never recommends you.

Technical SEO guides for each AI visibility factor

Use these deeper guides to fix the issue behind each checker result:

• JavaScript SEO for Web3 dApps: fix client-side rendering, wallet-gated explanations and hidden content.
• Indexing and crawl management for Web3: audit crawl coverage, sitemaps, duplicate URLs and crawl budget.
• llms.txt for Web3: create the file that points AI systems to canonical pages, documentation and definitions.
• Title tags and meta descriptions: align search snippets, page intent and metadata.
• Redirection management for Web3: protect authority during migrations, rebrands and URL changes.
• SEO audit checklist: run the wider technical and on-page review around the AI visibility test.

What the AI Visibility Checker Actually Tests

The tool checks ten technical factors that determine whether AI systems can access, read, understand, and trust your site. These are not the same factors that affect Google rankings. A site can score 90 on traditional SEO and 30 on AI visibility — and most Web3 sites do exactly that.

LLM SEO is not about ranking signals. It is about machine readability — and the two are almost entirely different disciplines.

When you run the checker, you get a score out of 100 and a list of risk factors. Each one maps to a specific structural problem that prevents AI systems from extracting, classifying, or citing your content. The sections below explain each factor in the order the tool surfaces them.

Run the tool now before reading on. Your score will make everything below concrete: 

Factor 1: robots.txt and Crawl Access

robots.txt tells AI crawlers whether they are allowed to access your site at all. If key pages are blocked — intentionally or by accident — LLMs cannot index them, which means they cannot cite them, extract from them, or use them to understand your brand.

A site that blocks AI crawlers is invisible to LLMs by definition, regardless of how good the content is.

Most Web3 sites have a robots.txt file. Fewer have checked what it actually allows. The issue is rarely an explicit block on all crawlers — it is subtler: blocking the /docs subdomain, the /blog path, or allowing Googlebot but not GPTBot, PerplexityBot, or ClaudeBot.

• Does robots.txt exist and load at yourdomain.com/robots.txt?
• Are key content paths — homepage, product pages, blog, docs — accessible to all crawlers?
• Are major LLM crawlers explicitly allowed, or only search engine bots?

The fix is usually a one-line addition to robots.txt. But you need to check it first.

Factor 2: Sitemap

A sitemap tells crawlers which pages exist and should be indexed. Without one, crawlers discover pages by following links — which means orphaned pages, recent content, and docs sections are often missed entirely.

Missing or incomplete sitemaps are one of the most common and most fixable LLM visibility failures on Web3 sites.

The failure pattern in Web3 is almost always one of three things: the sitemap exists but points to the wrong domain, it exists but excludes the blog or docs, or it was set up once and never updated as the site structure changed.

• Does a sitemap exist at yourdomain.com/sitemap.xml?
• Does it include all content you want AI systems to find?
• Is it submitted to Google Search Console and referenced in robots.txt?

Factor 3: llms.txt

llms.txt is an emerging convention — a plain text file at yourdomain.com/llms.txt that acts as a direct guide for AI crawlers. It tells LLMs which pages are most important, what content to prioritise, and how to understand the site’s structure.

llms.txt is the fastest-growing LLM SEO signal right now — and most Web3 sites have not implemented it yet, which makes it a low-effort, high-impact win.

Think of it as robots.txt for AI systems, but instead of blocking access, it actively guides understanding. A well-structured llms.txt can tell AI crawlers: here is our canonical product explainer, here is our risk documentation, here is the author responsible for this content.

The format is simple: a plain text file with structured headings and links. It does not require significant developer work beyond creating the file and uploading it to the root directory. Most sites do not have one. That is a meaningful differentiation opportunity right now, before it becomes standard practice.

Factor 4: Title Tag

The title tag is one of the first signals a retrieval system sees when deciding whether a page is relevant to a query. A vague or brand-only title gives machines nothing to work with. A descriptive, specific title tells the model exactly what the page covers before it reads a single word of content.

Title tags like “Home | Brand” or “Platform | Brand” actively reduce LLM visibility because they communicate no meaning — machines skip pages they cannot classify from the title alone.

The failure pattern is nearly universal in Web3. The Gnosis homepage title — “Gnosis — The Future of Finance” at 30 characters — tells an LLM nothing about what Gnosis actually does, what category it belongs to, or who it serves.

The rule: every important page title should contain the category, the primary function, and ideally the audience — without exceeding 60 characters.

Factor 5: Meta Description

The meta description is the page’s pitch to both humans and machines. For LLMs, it functions as a concise summary of what the page covers — which shapes whether the model uses the page as a source and how it describes the content in generated answers.

A specific, factual, constraint-aware meta description significantly increases the chance of accurate AI citations — a generic tagline does the opposite.

• States what the product is in plain language
• Includes the primary use case or user benefit
• Mentions one specific, verifiable fact or differentiator
• Avoids promotional language like “leading,” “best,” or “revolutionary”

The Gnosis meta description scores well in the tool — 156 characters, page-specific, non-generic. That is the standard. Most Web3 sites are not there yet.

Learn about ChatGPT search behavior.

Factor 6: H1 Tag Structure

Every page should have exactly one H1 tag. The H1 is the primary heading that tells machines what this specific page is about. Multiple H1s create ambiguity — the model cannot determine which one represents the page’s actual topic, so it hedges or ignores the page entirely.

Multiple H1 tags are one of the most common Web3 technical failures — and one of the most invisible to human visitors, which is exactly why it persists.

The Gnosis example from the tool shows this clearly: 9 H1 tags on the homepage, each containing a single word — OWN, YOUR, MONEY — because the site uses a Framer animation that splits the headline into individual elements, each tagged as H1.. Understanding ChatGPT search behavior helps optimize LLM visibility.

This is the React and Framer problem. Developers build visual effects by splitting text across multiple elements. The result looks like a single animated headline to a visitor. To an LLM reading raw HTML, it looks like nine competing page topics with no coherent meaning.

The fix: a single, descriptive H1 that states what the page is. Keep the animation — but use CSS and span elements rather than repeated H1 tags. This is covered in depth in the React and hidden content post.

Factor 7: Script Tag Ratio

The ratio of content to script tags tells you how much of your page is actual information versus code. A high script ratio — typically above 50% — signals that the page relies heavily on JavaScript to render, which means LLMs reading raw HTML may find very little actual content.

A page that is 52% scripts and 48% content is asking LLMs to read a page that is more than half empty in its raw state.

This is the single most common technical failure in Web3. The entire sector builds in React, Vue, or Framer — frameworks that render content client-side. LLMs do not execute JavaScript. They read the HTML served before any scripts run. If your content lives inside JavaScript components, it is invisible to the LLM.

The solution is server-side rendering or static generation for key pages. This ensures the HTML served to crawlers contains actual text content, not just script tags and empty div elements. The JavaScript SEO for Web3 post covers this in full.

Factor 8: Crawlable Word Count

Crawlable words are the words that exist in raw HTML — the text LLMs can read without executing JavaScript. A low crawlable word count means the model has very little information to work with when constructing an answer about your brand.

547 crawlable words on a homepage is not enough for an LLM to confidently classify and recommend a product — the minimum for meaningful AI visibility is 300 to 500 words of substantive, structured content.

The crawlable word count is a direct measure of how much information an LLM has access to. If that content is mostly navigation labels, footer text, and button copy, the model is working with almost nothing. The goal is substantive, structured content that answers: what is this, who is it for, how does it work, and who should not use it.

Factor 9: Structured Data (Schema Markup)

Schema markup is a vocabulary of HTML tags that labels your content for machines. It tells LLMs explicitly: this is an organisation, this is a product, this is a FAQ. Without schema, machines must infer all of this from surrounding text — which leads to misclassification, hedging, and reduced citation confidence.

No schema markup is one of the four highest-impact LLM visibility failures — it is the difference between a machine guessing what you are and a machine knowing with certainty.

• Organization on the homepage: name, URL, logo, social links
• Product on product pages: name, description, category
• FAQPage on any page with questions and answers
• Article on blog posts: headline, author, publish date
• Person on author pages: name, role, credentials

The Gnosis homepage has zero schema blocks. Every LLM that encounters it must infer the company’s category and function from unstructured text alone. The result is exactly what you would expect: hedged, imprecise descriptions in AI-generated answers. The full schema implementation guide is in the appendix of Mastering AI Search for Crypto & Web3 Brands.

Factor 10: Metadata Alignment

Metadata alignment means your title tag, H1, and meta description all describe the same thing using consistent language. When they tell different stories, LLMs receive conflicting signals and cannot settle on a confident classification.

Metadata misalignment is a trust signal failure — when a machine sees three different descriptions of the same page, it treats the page as unreliable and reduces citation confidence accordingly.

• Title: “Gnosis — The Future of Finance”
• H1: “OWN YOUR MONEY” (split across 9 H1 tags)
• Meta description: “A new generation of collectively-owned financial products…”

Three different framings, none of which clearly state what Gnosis is. The model has no stable anchor. The fix is a single, consistent definition of what the product is, used in the title, H1, and meta description — which is exactly what the entity SEO and canonical definition framework addresses.

What Your Score Actually Means

A score below 50 means fundamental structural problems that prevent LLMs from reading and classifying your site. These are eligibility failures — the site is not in the game yet.

A score of 50 to 70 means partially readable with significant gaps. LLMs can find some content but will hedge, misdescribe, or underrepresent the brand in AI-generated answers.

A score above 70 means the technical foundation is solid. The focus shifts to the content, authority, and reinforcement layers of the four-layer GEO framework — which determines whether AI systems actively recommend you, not just find you.

The technical layer is Layer 1. Fixing it is necessary. It is not sufficient.

Run the free checker to see exactly which factors are failing on your site: victoriaolsina.com/is-your-website-invisible-to-ai/

Download Mastering AI Search for Crypto & Web3 Brands to understand what to fix first and in what order — the book covers all four layers of the framework in full.

If you want a done-for-you audit and implementation plan, book a free 45-minute strategy call with Victoria.

Frequently Asked Questions

What is an AI visibility score and how is it calculated?

An AI visibility score measures how well a website is structured for LLM readability — covering technical factors like crawl access, schema markup, JavaScript rendering, H1 structure, and metadata alignment. Each factor is weighted by its impact on whether AI systems can extract, classify, and cite the site’s content. A score of 100 means no structural barriers to AI readability. Most Web3 sites score between 40 and 70.

An AI visibility score is a measure of machine readability, not search ranking — a site can rank on Google and score under 50 on AI visibility.

Why does JavaScript affect ChatGPT visibility?

LLMs do not execute JavaScript — they read raw HTML. If content is rendered by React, Vue, or Framer components, it does not exist in the HTML that AI crawlers receive. The page appears empty to the model, regardless of how it looks in a browser. Server-side rendering or static generation for key pages solves this by ensuring content appears in the HTML before any scripts run.

JavaScript-rendered content is invisible to LLMs — without server-side rendering, your content simply does not exist to AI search systems.

What is llms.txt and why does it matter for AI search?

llms.txt is a plain text file placed at your domain root that guides AI crawlers to your most important content — functioning like a table of contents for LLMs. It is an emerging standard with rapidly growing adoption. Most Web3 sites have not implemented it yet, making it a low-effort, high-impact differentiation opportunity in 2026.

llms.txt is the fastest-growing LLM SEO signal right now — implementing it takes less than an hour and most competitors have not done it yet.

How many H1 tags should a page have?

Exactly one. The H1 tells machines what a specific page is about. Multiple H1 tags — a common result of React and Framer animation patterns — create conflicting signals that models cannot resolve. The fix is a single descriptive H1 containing the page’s core topic, with animations handled via CSS and span elements rather than repeated heading tags.

One H1 per page, describing what the page is — not a tagline, not a brand name, not an animated headline split across nine tags.

What schema types matter most for Web3 brands?

The four non-negotiables are Organization (homepage), Product (product pages), FAQPage (any Q&A content), and Article (blog posts). A fifth — Person schema on author pages — significantly increases trust signals in high-risk domains like finance and crypto. Without schema, LLMs must infer your category and function from unstructured text, which produces imprecise or hedged descriptions in AI-generated answers.

Schema markup converts LLM guesswork into confirmed classification — it is the single highest-impact one-time technical fix for AI visibility.

Your technical foundation determines whether AI systems can even begin to understand your brand. If you’ve found gaps in your AI visibility score or need help implementing these technical fixes, a Web3 SEO specialist trained in technical architecture can audit your site structure and create a roadmap for improvement. See featured perspective on AI search visibility.