AIDRank measures how AI models see your domain today. This page focuses on what you can actually influence: what future model versions and AI search tools browsing the web today learn about you.
There are really two different questions hiding inside "AI visibility." The first is what a specific model already knows from completed training. That part is frozen. Nothing you publish today reaches backward into a model that has already shipped, and that is exactly what your AIDRank score reflects. The second question is what the next model learns, and what AI systems that actively browse the live web find when they look you up today. That part is genuinely open, and it is what the rest of this page is about.
Not all AI visibility comes from training. A lot of it comes from AI reading your pages in real time.
What a model absorbed while it was being built. This is frozen the day training ends, and it is what AIDRank measures. No site update touches it until a new model version is trained from scratch.
Tools like ChatGPT Search, Perplexity, and Google AI Overviews actively fetch and read pages when answering a question. This updates continuously, and ordinary technical SEO feeds it directly.
A fast page, a working sitemap, clean and valid HTML, a site Google can actually crawl and index, and content that loads without errors on a phone: none of that is new advice, and none of it stopped mattering. It is simply doing double duty now. The same technical health that helps you rank also decides whether an AI system browsing live can reach your page, read it cleanly, and quote it with confidence.
AI systems do not read a page top to bottom like a person does. They cut it into passages and pull out whichever one answers the question being asked.
Open each section with the direct answer in plain sentences, before any background or scene-setting. A person skimming benefits from this exactly as much as a machine extracting a passage does, which is a good sign it is not a trick, just clearer writing.
"Overview" tells a machine nothing about what is under it. "How Domain Slots Work" tells it exactly when to use that section. Phrasing a heading as the actual question someone would type works well, because it is quite literally the query being matched against.
A paragraph that makes one clear claim and backs it with a sentence or two of support is far easier to lift cleanly than a paragraph weaving three ideas together with transitions. If you can split it, split it.
"It handles this automatically" only makes sense with the previous sentence attached. "AIDRank re-scans tracked domains automatically" stands on its own. A passage pulled out of context still needs to make sense, so avoid leaning on pronouns to carry meaning across sentences.
If you are comparing two or more things side by side, a small table beats a paragraph describing the same comparison in words. It is easier for a machine to extract cleanly, and honestly, it is easier for a human to scan too.
Schema.org markup describes your page to a machine in plain, unambiguous terms, sitting quietly in the code where a visitor never sees it.
Four types cover most of what matters. Article or BlogPosting states who wrote a piece and when, with a datePublished and a dateModified that should actually be true. Organization states who runs the site. Person links an author to real credentials. FAQPage marks up a block of questions and answers as a clean, structured unit rather than plain paragraphs a machine has to guess at.
Worth knowing before you add it: the expandable FAQ accordion that used to appear under some Google results is gone as of mid-2026. That specific visual reward is no longer there. FAQPage markup is still useful, though, because AI systems reading your page still treat it as a clean, pre-formatted set of questions and answers, exactly the shape they are built to extract and quote. You are not doing it for the old Google feature any more. You are doing it for the AI reading the page right now.
<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "FAQPage",
"mainEntity": [
{
"@type": "Question",
"name": "Is scanning free?",
"acceptedAnswer": {
"@type": "Answer",
"text": "Yes, any domain, with no account needed."
}
}
]
}
</script>AI crawlers are not one thing. Training bots and live-search bots are separate, and robots.txt lets you treat them differently.
Each AI company runs at least two kinds of bot, and they behave nothing alike. Training crawlers like GPTBot and ClaudeBot fetch content in bulk to feed a future model's training run. Search or retrieval crawlers like OAI-SearchBot and Claude-SearchBot fetch a page live, right when someone asks a question, to answer it and cite a source. Blocking one does not block the other, because they use separate names in your robots.txt file.
That means you get a real choice, not an all-or-nothing switch. You can allow the search bots so your pages stay eligible to be cited in live AI answers, while blocking the training bots if you would rather your content not feed a future model's training data. Neither choice is automatically correct. It is a policy decision about your own content, and it is worth making on purpose rather than copying a template blindly.
# AI search / retrieval bots — allowed, so pages can be cited live
User-agent: OAI-SearchBot
Allow: /
User-agent: Claude-SearchBot
Allow: /
User-agent: PerplexityBot
Allow: /
# AI training bots — blocked, so pages are not used to train future models
User-agent: GPTBot
Disallow: /
User-agent: ClaudeBot
Disallow: /
User-agent: Google-Extended
Disallow: /
User-agent: CCBot
Disallow: /
Sitemap: https://yourdomain.com/sitemap.xmlWe'll be straight with you about this one, because most guides aren't.
llms.txt is a small Markdown file at the root of a site, meant to give an AI a short, curated map of your most useful pages. It is a genuinely good idea. It is also, as of mid-2026, a community proposal, not a formal standard. No major AI provider has committed to reliably reading it, and most of the big AI crawlers currently fetch and read your regular HTML pages directly rather than checking for this file first. No independent study so far has shown a measurable citation boost from having one.
None of that makes it a bad idea. It costs a couple of hours to build, some AI tools and coding agents already do use it, and the file itself does no harm sitting quietly at your root. Treat it as a small, deliberate bet on where things might go, not as a strategy on its own. If everything else on this page is already in place, adding an llms.txt is a reasonable next hour of work. It should not be the first hour.
# AIDRank
> AIDRank measures how well AI models like Claude and GPT
> already know a given domain, using training knowledge only,
> with web search switched off.
## Main Pages
- [Home](https://aidrank.com/): What AIDRank does and how to scan a domain
- [Why AIDRank](https://aidrank.com/why-aidrank): The reasoning behind the score
- [Glossary](https://aidrank.com/glossary): Every term used on the site, defined plainly
- [Pricing](https://aidrank.com/pricing): Free scanning, one-time paid domain slotsBefore an AI system repeats a claim, it looks for something to corroborate it against. An anonymous page gives it nothing to check.
A named author with a real bio, an Organization that clearly states who runs the site and from where, and a byline that is not "Admin" or "Content Team" on two hundred different pages, all give an AI system something to verify. This matters more, not less, as the web fills up with anonymous AI-written content. A page that is honest about who wrote it and why they are qualified has a real structural advantage over one that is not.
A dateModified field that says one thing while the visible page says another is worse than having no date at all. Keep it honest, and actually revisit your important pages every so often rather than setting a date once and forgetting it. A page that is quietly three years stale loses citation priority to a page that is current, even if the older page was better written to begin with.
If you only do a handful of things, do these.
Same policy as the rest of this site: our limits, stated plainly, before you invest time based on this page.
No, not directly. Your current score reflects a model that already finished training, and nothing published today reaches backward into that. What this guide affects is what the next model version learns, and how you show up in AI search tools that browse the live web right now. Both are real, just on a different timeline than a page edit today.
No. Start with clear headings and an answer-first opening paragraph on your most important pages. That alone helps human readers and AI systems at the same time. Add schema, sort out your robots.txt policy, and consider an llms.txt once the basics are solid.
That depends on what you want. Blocking the training crawlers does not remove you from AI search citations, as long as you leave the separate search and retrieval bots allowed. It is a genuine trade-off worth thinking through, not a setting to copy from a template without reading it.
No. It is optional, unproven so far, and most major AI crawlers do not consistently read it yet. It is cheap enough that many sites ship one anyway as a small, deliberate bet on where the field might go, not because it is currently required.
This guide is about what happens next. Your AIDRank score is where you stand right now, and it is free to check.
Scan My Domain