Audit my website for AI SEO: a simple step-by-step process

An AI SEO audit answers one practical question: can AI systems find, understand, cite, and recommend this website?

You do not need a giant process to start. You need a small number of checks in the right order.

Step 1: Confirm the site is reachable

Run:

curl -I https://example.com

Look for:

• 200 status
• no redirect loops
• no accidental noindex
• no authentication wall
• no country or bot blocking that affects public pages

If the homepage is unstable, fix that before touching content.

Step 2: Inspect robots.txt

Open:

https://example.com/robots.txt

Check whether you are blocking AI retrieval bots. Be especially careful with broad rules:

User-agent: *
Disallow: /

Also separate training crawlers from retrieval and search crawlers. Blocking training may be a business decision. Blocking retrieval usually hurts visibility in AI answers.

Step 3: Check your sitemap

Your sitemap should include the pages that explain your business, product, pricing, docs, comparisons, and core educational content.

Common problems:

• sitemap missing
• old staging URLs
• pages returning 404
• important pages excluded
• thousands of thin pages included

For AI search, the sitemap is a discovery signal. Keep it clean.

Step 4: Read the raw HTML

Use:

curl -L https://example.com

Search the output for your main message. If the important content is missing, your page may depend too much on client-side rendering.

Your initial HTML should make the page understandable.

Step 5: Validate structured data

Look for JSON-LD:

<script type="application/ld+json">

For most sites, start with:

• Organization
• WebSite
• WebPage
• Article or BlogPosting
• FAQPage when the page contains real FAQs
• Product, SoftwareApplication, or Service when applicable

Structured data will not rescue bad content, but it helps AI systems identify entities and facts.

Step 6: Add llms.txt

Create:

/llms.txt

List your best pages, not every page. The file should help an LLM quickly understand what matters on the site.

Step 7: Score answer readiness

Pick ten prompts your buyer might ask. For each prompt, ask whether you have a page that answers it directly.

If the answer is no, create a page. If the answer is yes, improve the first 200 words so the direct answer appears immediately.

Step 8: Check agent readiness

When the basics are fixed, check advanced discovery:

• API catalog
• MCP server card
• agent skills manifest
• OAuth discovery
• protected resource metadata
• markdown versions of docs

These are early signals, but they matter for sites that want agents to do more than read.

Run the audit automatically

NoticeMeAI runs the technical part of this audit for you and turns it into a score. Use it as the first pass, then write or improve the pages that match your highest-value prompts.