Skip to content
Loading
.marisi

How to Make Your Website GEO Ready

How to Make Your Website GEO Ready hero image

The Search Landscape Has Shifted

ChatGPT now has 900 million weekly active users. Google AI Overviews appear on somewhere between 20% and 50% of all queries depending on the category. Zero-click searches sit at roughly 58.5% of sessions. That means more than half of all searches end without a single visit to any website.

The traffic model that underpinned a decade of SEO strategy is fracturing. Users get synthesised answers directly from AI. If your content is not part of what gets synthesised, you are invisible, regardless of your ranking.

GEO (Generative Engine Optimisation) is the practice of structuring and signalling your content so that AI platforms retrieve, understand, and cite it in their responses. This post is a practical audit guide. No fluff, no vague "produce quality content" advice. Just concrete steps you can act on today.

How AI Search Actually Works

Before optimising for a system, understand what it does.

Modern AI search engines run a multi-step pipeline when a user submits a query:

  1. Query fan-out. The model expands the user's query into multiple sub-queries covering different angles, related concepts, and likely intents.
  2. Source retrieval and evaluation. A retrieval layer fetches candidate pages. These are ranked by a combination of traditional signals (authority, freshness, relevance) and AI-legibility signals (structure, directness, clarity).
  3. Synthesis and citation. The model reads the retrieved content, extracts the most useful fragments, and assembles an answer. It cites sources where it can attribute claims.

The key insight: a page ranked #5 can beat the #1 result if it gives a cleaner, more direct answer to the synthesised sub-query. The AI is not reading your page the way a human skims it. It is extracting discrete, attributable chunks of information. Your job is to make those chunks obvious and reliable.

GEO vs SEO: What Actually Changes

Good SEO is the floor for GEO. There is no shortcut around crawlability, authority, and relevance. The overlap is massive.

The difference is emphasis.

| Signal | SEO priority | GEO priority | |---|---|---| | Keyword placement | High | Medium | | Backlink authority | High | High | | Page speed | High | High | | Direct answer structure | Medium | Very high | | Schema markup | Medium | High | | E-E-A-T signals | Medium | Very high | | Content freshness | Medium | High | | Semantic HTML | Medium | High | | FAQ / definition blocks | Low | High |

SEO optimises for ranking in a list. GEO optimises for being extracted from a list and reassembled into an answer. The mechanics diverge most sharply in content structure and authority signalling.

The GEO Technical Checklist

This is the core of the post. Work through each section and treat it as a real audit.

1. Content Structure

AI systems reward content that answers questions directly and unambiguously. Buried answers, long preambles, and vague prose all reduce citation likelihood.

Do this:

  • Lead with the answer. State it in the first one or two sentences of a section, then elaborate.
  • Use H2 and H3 headings that mirror real questions. "What is GEO?" is better than "Overview".
  • Write explicit, citable definitions. "GEO is the practice of..." is extractable. "In today's digital landscape, many brands are exploring..." is not.
  • Keep paragraphs short. AI chunking algorithms work on paragraph-level blocks. Dense walls of text reduce extraction precision.
  • Use numbered steps for processes. Sequential structure is unambiguous and easy to attribute.

Avoid:

  • Long introductions before the actual answer
  • Hedging every claim into vagueness
  • Hiding key data or conclusions at the bottom of a long article

Pro tip: Read your H2s and H3s in isolation. If they could stand as standalone questions in a FAQ, you are on the right track.

2. Schema Markup

Structured data gives AI systems explicit, machine-readable context about your content, authorship, and organisation. It reduces ambiguity.

Implement these schema types as a baseline:

  • Article or BlogPosting with datePublished, dateModified, author, and publisher
  • Person for author profiles with name, url, jobTitle, and sameAs links to LinkedIn, GitHub, etc.
  • Organization with name, url, logo, and sameAs links to social profiles
  • FAQPage for any FAQ sections

Here is a minimal Article + Person example in JSON-LD:

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "BlogPosting",
  "headline": "How to Make Your Website GEO Ready",
  "datePublished": "2026-04-28",
  "dateModified": "2026-04-28",
  "author": {
    "@type": "Person",
    "name": "Your Name",
    "url": "https://yoursite.com/about",
    "sameAs": [
      "https://linkedin.com/in/yourprofile",
      "https://github.com/yourhandle"
    ]
  },
  "publisher": {
    "@type": "Organization",
    "name": "Your Site",
    "url": "https://yoursite.com",
    "logo": {
      "@type": "ImageObject",
      "url": "https://yoursite.com/logo.png"
    }
  }
}
</script>

Audit your existing schema with the Detailed SEO Extension for Chrome or Google's Rich Results Test. Missing or malformed schema is a quick win.

Pro tip: The sameAs property on both Person and Organization is particularly valuable. It lets AI systems cross-reference your identity across the web, which strengthens trust signals.

3. Crawlability

If an AI crawler cannot reach your content, nothing else matters.

Check your robots.txt first.

Common AI crawlers use these user-agent strings:

GPTBot          # OpenAI / ChatGPT
PerplexityBot   # Perplexity
Claude-Web      # Anthropic
Google-Extended # Google AI (Bard/Gemini training)
Applebot        # Apple Intelligence

An overly aggressive robots.txt blocks traffic you may actually want. Here is an example that allows all the above:

User-agent: GPTBot
Allow: /

User-agent: PerplexityBot
Allow: /

User-agent: Claude-Web
Allow: /

User-agent: Google-Extended
Allow: /

User-agent: Applebot
Allow: /

If you use Cloudflare, audit your bot management rules. "Bot Fight Mode" and certain firewall rules can inadvertently block AI crawlers. Check your Cloudflare Security logs for blocked requests from the user-agents above.

Server-side rendering matters.

If your key content is rendered client-side via JavaScript, some crawlers will not see it. Use SSR or static generation for any content you want indexed and cited. Next.js App Router with server components handles this correctly by default. Verify with a "view source" check: if the content is in the raw HTML, crawlers see it. If it is not, you have a problem.

Do not lock valuable content behind logins. AI systems cannot authenticate. If your best content is paywalled or login-gated, it will not be cited.

Consider an llms.txt file.

The llms.txt convention (inspired by robots.txt) is an emerging standard that gives AI systems a structured summary of your site and its key content. It lives at https://yoursite.com/llms.txt and looks like this:

# yoursite.com

> A developer portfolio and technical blog focused on frontend engineering and AI tooling.

## Blog

- [How to Make Your Website GEO Ready](https://yoursite.com/blog/how-to-make-your-website-geo-ready): Practical GEO audit guide for developers.
- [Build an MCP Server with Python](https://yoursite.com/blog/build-an-mcp-server-with-python): Step-by-step guide to building Model Context Protocol servers.

## About

- [About](https://yoursite.com/about): Background, skills, and contact information.

Not every AI platform reads llms.txt yet, but adoption is growing and the implementation cost is near zero.

Pro tip: Generate your llms.txt programmatically from your sitemap or content metadata. That way it stays in sync automatically.

4. Performance

Fast pages are crawled more reliably and ranked more favourably by every system that evaluates them, AI or otherwise.

The baseline targets:

  • LCP (Largest Contentful Paint): under 2.5 seconds
  • INP (Interaction to Next Paint): under 200ms
  • CLS (Cumulative Layout Shift): under 0.1
  • TTFB (Time to First Byte): under 800ms

Performance matters for GEO because AI search evaluation pipelines draw from the same signals as Google's mobile-first index. A slow page that frequently times out during crawl will have sparse or stale indexed content.

Keep your JavaScript bundle lean. Avoid render-blocking third-party scripts on content pages. Use next/image for automatic optimisation if you are on Next.js. Serve static or ISR pages where possible.

Pro tip: Run a Lighthouse audit on your most important content pages and fix anything in the red. A 40-point performance score is not just a UX problem; it is a crawl problem.

5. Authority Signals (E-E-A-T)

Google formalised E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness) as a quality signal years ago. AI systems have independently converged on similar heuristics. Content from known, credible sources gets cited more.

Build your authority signals deliberately:

  • Author profiles. Every post should have a named author with a linked bio. The bio should state relevant credentials, experience, and expertise. Link to your GitHub, LinkedIn, published work.
  • Original data and research. Posts that contain original statistics, benchmarks, or case studies are cited more than posts that only aggregate existing information. If you ran an experiment, publish the numbers.
  • Expert commentary. Quote practitioners with named credentials. "According to [Name], senior engineer at [Company]..." is more citeable than an uncredited assertion.
  • Cross-web presence. Mentions on trusted platforms (dev.to, GitHub, Hacker News, industry publications) reinforce your authority profile. AI systems do not just read your site; they read the web around it.
  • Outbound links to authoritative sources. Citing credible references (research papers, official docs, well-regarded publications) signals that your content is grounded.

Pro tip: Write an /about or /authors/[name] page with explicit expertise signals and link every post back to it. Schema sameAs links from that page to your other profiles compound the effect.

6. Content Freshness

An article from 2023 competing against one from 2026 on the same topic is at a structural disadvantage. AI systems weight recency, especially for topics where the facts change.

Practical steps:

  • Add a visible "Last updated" date to cornerstone content and update the dateModified in your schema when you refresh it.
  • Do a content audit twice a year. Identify your highest-traffic or most-linked posts and update the statistics, examples, and tool references.
  • Avoid dates in URLs if you want articles to stay evergreen without canonical confusion.
  • When you update a post substantially, add a brief "Updated [Month Year]" note at the top with a summary of what changed.

Pro tip: A 15-minute refresh of an existing post's statistics and examples often outperforms a brand-new post on the same topic in terms of GEO impact. The existing page already has authority; you are just making it current.

7. Accessibility and Semantic HTML

This one is underrated. Semantic HTML is not just for screen readers. AI models parse structure the same way assistive technologies do. A well-structured document is more legible to both.

Checklist:

  • Use heading levels correctly. One h1 per page, logical h2/h3 hierarchy, no skipping levels.
  • Use <article>, <section>, <nav>, <main>, <aside>, <footer> appropriately.
  • Write descriptive alt text for all images. If an image contains data or key information, that context exists nowhere else in the DOM without alt text.
  • Use <table> with <thead>, <tbody>, and scope attributes for tabular data.
  • Avoid CSS-only content (text via content: in pseudo-elements) for anything meaningful.

Clean semantic HTML reduces the noise an AI has to parse to find the signal. It is not glamorous work, but it compounds with everything else on this list.

Pro tip: Run axe or the Lighthouse accessibility audit on your key pages. Accessibility violations frequently correlate with semantic HTML problems that hurt AI legibility.

The UX and GEO Venn Diagram

Here is the practical reality: almost everything that makes a site better for users also makes it better for AI.

  • Clear structure helps users scan and helps AI chunk.
  • Fast load times reduce bounce and improve crawl reliability.
  • Logical navigation aids discoverability for both humans and spiders.
  • Readable prose is good for both a tired developer at 11pm and an LLM trying to extract a claim.
  • Semantic HTML powers screen readers and AI parsers alike.

The Venn diagram between good UX and good GEO is nearly a circle. Optimising one rarely conflicts with the other. If you are ever unsure whether a change is worth making for GEO, ask whether it also improves the experience for a real user. If the answer is yes, do it.

Measuring GEO Impact

Measurement is the least mature part of the GEO stack right now. Honest acknowledgment of that saves frustration.

What you can measure today:

  • AI referral traffic in GA4. Check Acquisition > Traffic acquisition and filter for sources containing perplexity.ai, chatgpt.com, bing.com (Copilot), and you.com. This is incomplete but growing.
  • Manual citation audits. Query your key topics directly in ChatGPT, Perplexity, Claude, and Google AI Overviews. Note whether your site is cited. Do this monthly for your most important content.
  • Share-of-voice tools. Semrush has started tracking AI visibility. Geoptie and similar tools are purpose-built for AI citation tracking. Coverage is improving quarter by quarter.

What is still hard:

  • ChatGPT's browsing citations are not fully auditable externally.
  • Google AI Overviews do not expose a consistent citation API.
  • Attribution across paraphrased content (where your words are used but not directly linked) is nearly impossible to track.

Set up the GA4 referral monitoring now. Build the manual audit habit. The tooling will catch up.

Summary: Your GEO Audit in One Table

| Area | Key action | Quick win? | |---|---|---| | Content structure | Lead with answers, question-based headings | Yes | | Schema markup | Add Article, Person, Organization, FAQ schema | Yes | | Crawlability | Audit robots.txt, check Cloudflare rules, add llms.txt | Yes | | Performance | Hit Core Web Vitals targets, reduce JS on content pages | Medium | | Authority signals | Author bios, original data, cross-web presence | Medium | | Content freshness | Update dateModified, refresh statistics annually | Yes | | Semantic HTML | Correct heading hierarchy, landmark elements, alt text | Yes |

Where to Start

If you want one concrete action from this post: run a "view source" check on your three most important pages, a Lighthouse audit, and a manual query in Perplexity and ChatGPT for your primary topic. That thirty-minute audit will surface more actionable issues than any amount of strategic reading.

GEO is not a separate discipline you bolt on. It is what happens when you do the fundamentals of web development well: fast, accessible, well-structured, credible content that is easy for any system to read. AI just raised the bar on what "easy to read" means.

Fix the basics. Keep them current. Build genuine expertise signals. The citations will follow.