Does your website help AI understand your business?

0:00

–:–

AI search is changing how people find businesses like yours online.

Instead of going straight to Google, doing a search, and clicking through a list of results, more people are asking tools like ChatGPT, Gemini, Perplexity, Copilot, and Google’s own AI features for direct recommendations, comparisons, and summaries.

People follow the path of least resistance. Increasingly, that “lowest effort” path isn’t ten blue links anymore. Which raises a new question for website owners:

Should your website include a page written specifically to help large language models understand your business?

You’ll see these described in a few ways: LLM info pages, AI information pages, source-of-truth pages, /llm-info/ pages, and sometimes a related file called llms.txt.

The idea’s pretty simple really. You create one clear, structured page (or file) that explains who you are, what you do, who you help, where you work, and what you don’t do.

I think it’s worth me saying this straight away: most websites don’t have an “AI problem”. They’ve got a clarity problem.

Most websites don’t have an “AI problem”. They’ve got a clarity problem.

If a real person can’t quickly work out what you or your business do and whether you’re a good fit, AI tools are going to struggle too. The difference is that AI will often fill in the gaps, even when it shouldn’t. It’ll do it confidently too, and sometimes it’s going to be wrong.

If your website’s already clear and your SEO foundations are solid, an LLM page can be a useful supporting layer. If your site’s unclear or technically messy, it’s not going to save you. In fact, piling another contradictory page on top of other messy pages may even make things worse.

What’s an LLM info page?

An LLM info page is just a regular public webpage which is designed to summarise your business in a way that’s easy for AI tools, search engines, chatbots, and other automated systems to interpret.

In practice, it’s a “single source of truth” that clearly states what your site probably already says across your homepage and all your other pages too. The difference is that it’s intentionally factual and unambiguous.

That matters because most websites aren’t written as one consistent, factual explanation. Each page has its own job. A homepage is broad. A service page persuades. A case study tells a story. That’s all fine, but it can leave automated systems with a skewed or confused impression of what the business actually does.

A good LLM page usually covers:

what the business does (in plain English)
who it’s for
where it operates
what it doesn’t do
links to your most important pages (services, case studies, testimonials, contact)

What’s llms.txt?

llms.txt is currently a proposed convention: a Markdown file placed at the root of a website, usually at /llms.txt

You can think of it as a cousin of the LLM info page we discussed earlier, but written more for crawlers than humans. Websites can be big, cluttered, and full of marketing copy and navigation noise. A short Markdown file can provide a cleaner starting point, plus direct links to the pages that matter. The ones that you hope AI will surface in the content it shows to users.

The key point is that it’s still a proposal. It’s not in the same category as robots.txt, XML sitemaps, or schema markup in terms of adoption or proven impact. So yes, you can add it. Just don’t assume it’ll be treated like an established standard, yet.

How does this actually work in Google’s AI features?

Google’s guidance is a useful reality check .

For AI Overviews and AI Mode, Google says the same fundamentals still apply. Pages need to be indexable, eligible to appear in Google Search, useful to visitors, technically accessible, and supported by strong content. Google also says there aren’t additional technical requirements for appearing in AI Overviews or AI Mode.

That’s helpful because it keeps expectations grounded. You don’t need to publish special AI-only files to be eligible for Google’s AI features.

That said, Google isn’t the only player here. Plenty of people now start with ChatGPT and never reach Google at all. So it’s worth thinking beyond “what does Google want?” and also thinking “can AI tools access and interpret what I’ve published?”

Do AI citations work like classic search rankings?

Not exactly. Ahrefs analysed 863,000 keyword SERPs and 4 million AI Overview URLs. Their March 2026 update found that around 38% of URLs cited in AI Overviews also appeared within the first 10 result blocks (Ahrefs link in the appendix).

That suggests AI systems can pull from a wider range of sources than “page one or nothing”.

38% of URLs cited in AI Overviews also appeared within the first 10 result blocks.

Google’s also said its AI features may use a “query fan-out” technique. Basically, instead of doing one search, it does a number of smaller related searches, and then builds a single response from what it finds.

So the win here isn’t an “AI trick”. It’s making sure your site says the important stuff plainly, in a couple of obvious places, so there’s less room for misreading or guesswork.

Can AI tools actually access your site?

OpenAI documents several user agents (including OAI-SearchBot and GPTBot) and how site owners can allow or block them via the robots.txt file. In other words, access controls still matter.

So if you do publish an LLM info page (or llms.txt), it’s worth checking it’s actually reachable. If your robots.txt, CDN/WAF rules, firewall, or security plugin blocks the relevant bots, you’re making it harder for AI tools to use what you’ve published.

In WordPress, this is where things often go wrong in not always obvious ways. I’ve seen CDN challenges that bots can’t pass, staging noindex settings left switched on, and sitemap issues after migrations or plugin changes. None of it’s glamorous, but it’s the difference between “this might help” and “this can’t possibly help”.

Is there evidence that this works?

This is where it gets a bit awkward. There’s useful data on llms.txt, but much less robust data specifically testing human-readable /llm-info/ pages.

A public /llm-info/ page is just a normal webpage with structured business information. A root llms.txt file is a proposed machine-readable Markdown convention. They’re related ideas, but they’re not quite the same thing.

The best available evidence is mostly about llms.txt (sources linked in the appendix). SE Ranking analysed nearly 300,000 domains and found that only 10.13% had an llms.txt file in place. More importantly, their analysis found no correlation between having llms.txt and being cited more often by AI systems.

Search Engine Journal summarised the same research by saying citation rates didn’t change based on whether a site used the file, and that site owners shouldn’t expect a visibility lift from adding it today.

SE Ranking analysed nearly 300,000 domains and found that only 10.13% had an llms.txt file

It’s also worth separating two different outcomes here: visibility and accuracy. Those studies are mainly looking at whether a site gets cited more often. That’s useful, but it doesn’t tell us whether a business is described more accurately when it is mentioned.

Even if citation rates don’t increase, it’s still possible that a clear “source of truth” page reduces ambiguity and helps AI tools summarise your services, location, and positioning more correctly (and avoids inventing things you don’t offer). That’s harder to measure, but it can still be a worthwhile goal.

Search Engine Land made a similar point about LLM-only pages and files: the format itself doesn’t get special treatment. The content may help, but only if it’s better, clearer, or more useful than what already exists on your normal pages.

So the takeaway isn’t “don’t do it”. It’s this: The file format isn’t the point. The clarity is.

So, should you add one?

It’s worth considering if your business is easy to misread, your offer’s a bit nuanced, or your site doesn’t have one clear “this is what we do” page. But if your fundamentals aren’t solid yet (indexing issues, weak service pages, slow performance, poor internal linking), sort that out first. An LLM page’s a finishing layer, not a fix.

There are three solid reasons to consider it.

Improve Clarity

A lot of business websites are written to persuade. That’s fine. But persuasion copy can get vague, especially when it leans on phrases that sound nice without saying much.

Improve Consistency

If your homepage, service pages, Google Business Profile, and directory listings all describe your business slightly differently, you create ambiguity. A single “source of truth” page helps keep the story straight.

Reduce Risk

AI tools sometimes summarise businesses incorrectly. They might infer services you don’t offer, locations you don’t serve, or claims you haven’t made.

What should an LLM info page include?

It should be factual, structured, and easy to scan. You’re trying to remove ambiguity, not write marketing copy.
A “minimum viable” version looks like this:

Business: ICW Digital (Paul Edwards)
What it is: WordPress website continuity + SEO and technical optimisation support
Who it’s for: small and overstretched marketing teams (UK)
Where: Worthing, West Sussex (remote UK support available)
Core services: WordPress maintenance, technical troubleshooting, technical SEO audits, performance fixes
Not offered: general IT support, guaranteed rankings
Key pages: /about/, /services/, /content-seo-technical-optimisation/, /contact/
Contact: ******@icw.digital
Last reviewed: May 2026

That format helps machines, but it also helps people. It gives a quick “are we talking about the same thing?” reference, without digging through five pages.

Should you also add llms.txt?

Probably. It’s low effort, but expectations should be modest. There’s no strong evidence it increases AI citations today. The better reason to add it is that it gives AI systems (and humans) a clean, maintained list of key pages and an accurate description of what the business does.

If you do add it, keep it simple and keep it maintained. A stale “source of truth” is worse than none.

A stale “source of truth” is worse than none.

How would you measure whether it helps?

You probably won’t get neat attribution like “LLM info page = X leads”. Instead, you’re looking for a mix of signals over time:

crawl and access (including server logs, if available)
indexing and impressions in Google Search Console
whether AI tools describe the business more accurately
whether citations start showing up for relevant prompts
whether enquiries begin mentioning AI tools or AI summaries

You can also manually test prompts in AI tools, especially around “what does this business do?” and “is this business a good fit for X?”. It’s not perfect measurement, but it’s still useful signal.

Let’s make your website easier for AI tools to understand

My SEO and technical optimisation services combine content review, technical SEO, performance improvements, and practical recommendations to make your website clearer, faster, and more visible.

If you’d like to discuss your website, call me on 01903 527927 or get in touch via my contact page.