To get cited in ChatGPT, Perplexity, and Google AI Overviews, your page has to be reachable by AI crawlers, easy for a model to parse, and clearly trustworthy. In practice that means allowing AI bots in robots.txt, adding an llms.txt and structured data, leading every page with a concise answer, and earning third-party mentions. Below is the exact playbook, plus a free way to score your site.
The short answer
AI search engines do not rank ten blue links. They read a handful of sources, synthesize one answer, and cite a few of them. You win by being one of the sources the model can find, parse, and trust. The mechanics differ from classic SEO, and most sites are not set up for them yet.
Why AI search is different from Google
Traditional SEO optimizes for a ranked list a human scrolls. AI search optimizes for extraction: a model fetches pages, pulls the cleanest answer, and attributes it. That rewards machine-readable structure over keyword density, and it punishes anything a crawler cannot reach or parse. The result is a new, largely uncontested signal set sitting on top of SEO.
What actually gets you cited
We scanned roughly 60 sites with the DigiJaws SEO Agent and the gaps were striking: only 46% had an llms.txt, just 7% had FAQ schema, 41% had a valid sitemap, and 29% were quietly blocking AI crawlers. Closing those gaps is most of the battle. Here is what matters, in order:
- Let AI crawlers in. Allow GPTBot, ClaudeBot, PerplexityBot, and Google-Extended in robots.txt. If they are blocked, you are invisible to those engines no matter how good your content is.
- Add an llms.txt. This standard file tells language models what your site is and where your key content lives. Fewer than half of sites have one, so it is an easy edge.
- Lead with the answer. Put a concise, direct answer in the first two sentences of every page. Models lift lead answers; burying it costs you the citation.
- Mark up your content. Organization, FAQ, and HowTo JSON-LD let a model label your content instead of guessing. FAQ schema in particular is almost unused (7% of sites), so it is low-hanging fruit.
- Use question-style headings and lists. Extractable structure (clear H2 questions, short lists, definition pairs) maps directly to how answers are assembled.
- Earn mentions. Generative engines weight what other credible sites say about you. Digital PR and citations still matter, arguably more than links alone.
- Keep it readable. 86% of pages we checked failed a basic readability bar. Short sentences and plain language help both people and models.
A step-by-step plan
- Audit your current AI-readiness so you know your baseline and which signals fail.
- Unblock AI crawlers in robots.txt and publish a sitemap.
- Add an llms.txt that points to your most important pages.
- Add Organization and FAQ JSON-LD to your key pages.
- Rewrite the top of each page to open with a direct answer.
- Build a few authoritative mentions through digital PR.
- Re-scan and track the score over time so regressions get caught.
How to check your site in 30 seconds
You do not have to guess. Run any URL through the free SEO Agent to get an AI-readiness score, a per-signal breakdown, and the exact fixes generated from your own site. Want the full method first? Read how the score works and what AI search readiness is.
FAQ
How do I get my site cited by ChatGPT?
Make sure GPTBot is allowed in robots.txt, add an llms.txt and structured data, lead pages with a concise answer, and earn credible mentions. Then verify with a free AI-readiness scan.
Does AI search replace SEO?
No. It sits on top of SEO. Many basics (HTTPS, clean titles, fast pages, structured data) help both, but AI search adds new signals like llms.txt and AI-crawler access.
What is the single biggest miss?
In our data, FAQ schema (missing on 93% of sites) and llms.txt (missing on 54%) are the most common, highest-leverage gaps.
