Guides

How AI answer engines read websites - written from what this tool keeps finding on real sites.

A growing share of questions never reach a results page. Somebody asks an assistant, the assistant reads a handful of sources, and it answers. Whether your site is one of those sources is decided by a different set of rules than the ones that decide where you rank, and most sites have never been checked against them.

These guides cover what this auditor keeps finding. They are written in the order the problems actually matter, because the first one on the list makes every other one irrelevant if you have it.

Start with rendering

The single most common finding, and the most severe, is that a site's content does not exist in the HTML the server sends. It is built in the browser by JavaScript. Google runs JavaScript and sees the finished page; most AI crawlers do not, and see close to nothing.

This is why a site can hold a strong position in search and be entirely absent from an assistant's answer about its own subject. The two systems are reading different documents at the same address. No amount of structured data or heading discipline compensates for content that is not there to read, which is why the rendering guide comes before the rest.

Then access, then legibility

Once your content is genuinely in the HTML, two questions remain. Are the AI crawlers allowed to fetch it - a robots.txt rule written years ago can be excluding you from systems that did not exist when it was written. And once fetched, can a model tell what the page says, who wrote it, and what it is about.

That second question is where structured data, entity clarity, heading structure and a named author come in. None of them is exotic. They are mostly old advice that never stopped being right, with a new reason to follow it.

The five guides

What these are not

They are not a growth-hacking checklist, and none of them promises a position in anything. Nobody outside the companies running these models knows how citations are selected, and anybody claiming otherwise is guessing with more confidence than the evidence supports.

What is knowable is the mechanical part: whether your content is in the HTML, whether the crawlers are allowed in, whether the page declares what it is. Those are facts about your site, they can be measured, and they are what this tool measures. The rest is content quality, which no auditor can check for you.

You can run the audit on your own site from the home page. It takes about a minute, needs no sign-up, and reports what it found rather than a grade.