A practical, copy-paste guide to the four infrastructure fixes that determine whether ChatGPT, Perplexity, and Google AI Overviews cite your site or skip it.
I audited 360 domains against an AI-search-readiness framework. The average score was 54.1 out of 100. Most sites fail not because of content quality, but because of four infrastructure issues that take 30 minutes to fix.
AI search is a selection problem, not a ranking problem. A model receives a query, decides which sources to consult, extracts a passage, and either cites you or paraphrases without attribution. There is no page two.
Getting selected requires four things to work in sequence: Access — the AI crawler can reach your content Orientation — it can find what matters Understanding — it can parse what you are Quotability — it can extract a self-contained passage
Fix access before schema. Fix schema before content. Wrong order = wasted work. I saw teams rewriting content for "AI optimization" while their robots.txt blocked GPTBot.
There are at least 11 AI crawlers actively indexing the web. Each has a distinct user-agent token. Most sites block them by accident — a CMS, a security plugin, or a boilerplate template added Disallow: / and nobody reviewed it.
If you see Disallow: / under User-agent: , you're blocking everything. If you see Disallow: / with no user-agent specified, same thing.
Here's a minimal robots.txt that explicitly allows the major AI crawlers while keeping private paths locked down:
One more thing on access: if your site is a SPA that returns an empty on the initial HTML response, AI crawlers see nothing. GPTBot and PerplexityBot do not execute JavaScript reliably.
If that returns nothing, your content is invisible. You need SSR, SSG, or a prerendering layer. This is not a 5-minute fix, but you need to know about it now.
llms.txt is a plain-text file at the root of your site that gives AI crawlers a structured summary of your content. Think of it as a sitemap that a model can actually read.
In my audit, 54.2% of sites had a llms.txt file, but only 26.9% had a complete one. A broken llms.txt is worse than none — it sends a model a map with missing streets.
Key rules: The # line is your site title The > line is a one-sentence summary of what you do ## sections group links Each link is - Title: Description — the description matters, it gives the model context about what's at that URL
Don't list every blog post. List the pages that answer "what is this site?" and "what does it do?"
Schema is how you tell a model what your entities are: who you are, what your site does, what questions your pages answer.
In the audit, 75.6% of sites had some schema, but only 52.2% had Organization schema and 18.1% had FAQ schema. The gap between "has some schema" and "has the schema types that matter for AI citation" is where most sites lose ground.
The @graph structure lets you declare multiple entities in one block and cross-reference them with @id. This is how you tell a model "this organization publishes this website."
If you have FAQ pages or pages that answer questions, add this on the relevant page: