Search engines rank pages; AI engines retrieve and cite them. Here are the five questions we grade a site on to measure how ready it is for LLMs, vector retrieval and AI answers.
By Hanova Editorial
Search engines rank pages. AI engines retrieve and cite them. That difference is why a site can rank well on Google and still go unmentioned inside ChatGPT, Gemini or Perplexity. Ranking measures how a page competes for a click. AI-readiness measures something else: how easily a model can find your content, understand it, pull the right piece of it into an answer, and trust you enough to name you.
When we audit a site for generative engine optimisation, we grade it on five questions. Here is what each one means, how to tell where you stand, and which to fix first.
Before a model can cite you, the pages it looks for have to exist: a clear About, real service pages, an FAQ, a genuine blog, case studies, and team or author pages. These are the documents an AI reaches for when it assembles an answer about your company. Most sites are thinner here than they think. The work is unglamorous and high-leverage: publish the pages, and make each one a complete, self-contained answer rather than a teaser.
A model should not have to infer what your page is. That comes down to markup. Every page needs complete metadata (title, description, canonical, Open Graph), valid semantic HTML, and structured data in JSON-LD that states plainly what the page is and who published it. Clean, machine-readable markup is the cheapest win in the whole exercise, and it is the one most sites skip everywhere except the homepage.
This is where AI-readiness departs hardest from SEO. Models do not read a page top to bottom. They split it into chunks, convert each chunk into a vector (a numeric fingerprint of its meaning), and pull back the chunks that match a question. So the unit that matters is not the page, it is the passage. Content does well here when it is broken into clearly headed sections, each of which makes sense on its own, with FAQ and how-to structure the retriever can lean on. Write in retrievable units, because a self-contained passage is what an AI lifts and attributes.
A model rewards a site that is internally coherent. That means recognisable entities (your products, services, people and ideas), clear relationships between them, and topics organised into a hierarchy instead of a flat pile of posts. Consistent naming, topic clusters built around pillar pages, internal cross-links, and Organization and Person schema all push it up. The payoff is that the model treats you as a single entity it can reason about, not a scatter of unrelated URLs.
Finally, would a model put its name next to yours? Three things decide it: a consistent identity (canonical URLs and metadata that all agree on who you are), demonstrated expertise (named authors with real credentials, sources and citations), and visible trust infrastructure (contact, privacy and terms pages, plus social proof). This is E-E-A-T, expressed in a form a machine can check. The work is to stop shipping anonymous, source-free content and start signing it.
You do not fix all five at once. In an audit, two move the needle fastest. Machine readability first, because structured data and clean markup are quick and every model relies on them. Then retrievability, because rewriting content into self-contained, well-headed chunks is what turns a page from paraphrased to quoted. Strong foundations, semantic coherence and trust then compound on top.
Grading yourself is more useful than a checklist, because it tells you where you stand, lets you rank the work, and gives you something to move quarter over quarter. The platforms will keep changing, but the shape holds. Be findable, be machine-readable, be retrievable in pieces, be coherent, and be trustworthy. Score yourself honestly on those five, fix the weakest two first, and you stop optimising for a ranking you can watch and start optimising for an answer you cannot.
Eighteen months of measured experiments across ChatGPT, Gemini and Perplexity, what triggers a brand citation, what gets paraphrased away, and how to structure content so the LLMs name you.
