You almost certainly know how your website performs in Google. You have rankings, impressions, maybe a Search Console tab open right now. What you very likely do not know is how your website reads to ChatGPT — or Claude, or Perplexity, or the system generating Google's AI Overviews.
That gap matters more every month. An LLM audit closes it. It is a short, specific piece of work that answers one question: when an AI system encounters your business, what does it actually understand?
What an LLM Audit Actually Checks
An LLM audit is not a traditional SEO audit with different branding. A conventional audit measures the things Google's crawler cares about — page speed, crawl errors, keyword placement, backlink profile. An LLM audit measures comprehension. Those are genuinely different questions, and a site can score well on one while failing the other.
Concretely, the audit works through five things.
Extraction. We feed your pages to the major language models the same way they encounter them in the wild, and read what comes back. Did the model correctly identify what your business does? Did it get your services right, or did it invent something adjacent? Does it know where you operate?
Entity clarity. AI systems work with entities — your business as a thing with properties, not just a string of keywords. Does your site make it unambiguous who you are, what you sell, and where? Sites that describe themselves obliquely tend to be summarised badly.
Structured data. Schema markup is how you hand a machine clean facts instead of asking it to infer them from prose. The audit reports what markup exists, what is missing, and what is present but wrong — mismatched schema is worse than none.
Factual consistency. If your pricing says one thing on your services page and another in a blog post, an AI system has to pick one. It may pick the wrong one, and it may repeat it. Contradictions across your own site are one of the most common findings.
Citability. Given a question you should plausibly be the answer to, does the model cite you? Cite a competitor? Cite nobody in your sector at all? This is the closest thing to a bottom-line result the audit produces.
What the AI Readiness Score Means
The audit returns a score out of 100, broken down by platform, because the platforms genuinely differ. A site can read well to ChatGPT and poorly to Perplexity, usually because one of them leans harder on live retrieval and the other on structure.
The number is a diagnostic, not a grade. Below 40 usually means something structural — thin content, no schema, a site built so heavily in JavaScript that most of it never gets read. Between 40 and 70 is where most UK business sites sit: the content is fine, the machine-readability is not. Above 70 generally means the fundamentals are in place and the remaining work is competitive rather than corrective.
What matters more than the headline number is the platform spread and the list of specific gaps underneath it.
If you would rather just see your own numbers than read about them, the audit is free and takes us 48 hours.
Call +44 20 4572 4940 →What You Get Back
A report, and a conversation about it. The report covers your score and platform breakdown, what each model correctly understood about your business, what it got wrong or missed entirely, the schema you are missing, any factual contradictions we found across your pages, and a prioritised list of fixes ordered by likely impact rather than by ease.
The prioritisation is the part that tends to be useful. Most sites have thirty things they could fix and four that would actually change anything.
Why This Is Worth Doing Now Rather Than Later
Three reasons, in order of how much they should worry you.
People are already searching this way. Not everyone, not for everything — but a meaningful and growing share of the research that used to happen on a results page now happens in a conversation. If you are absent from that conversation, you do not see it happen. There is no impression count for the answer that never mentioned you.
Ranking well does not protect you. This is the finding that surprises people most. A page can sit on page one of Google and still be close to unusable to a language model, usually because the useful facts are buried in marketing prose, or locked in an image, or contradicted three pages away. Google's crawler is far more forgiving of this than an AI system deciding whether to stake an answer on you.
The fixes compound. Schema, clear factual statements, consistent information, content that answers a question directly — these help AI systems and traditional search simultaneously. This is rare in SEO. Most of what you would do after an LLM audit improves your Google performance too.
What Happens After the Audit
Some businesses take the report and act on it in-house. That is a completely reasonable outcome and we write it to be usable that way — the fixes are specific, not vague.
Where the work is larger, it usually splits into two disciplines. Answer engine optimisation covers being cited in direct answers from ChatGPT, Perplexity and Copilot. Generative engine optimisation covers appearing inside Google's AI Overviews and similar generative results. Related, overlapping, but not identical work.
If you want the broader background before running anything, our LLM SEO guide covers the underlying mechanics, and how to track brand visibility in ChatGPT covers ongoing measurement once you have made changes.
The Practical Case
The audit costs nothing, needs no access to your site — we read what the models read, which is your public pages — and comes back inside 48 hours.
The realistic outcome is one of two things. Either it confirms you are in decent shape, in which case you have removed a genuine unknown from your marketing at no cost. Or it finds something, which is the more common result, and you now know what it is rather than continuing to guess.
Most UK businesses have simply never checked. That is the whole reason this is worth ten minutes of your time.
Frequently Asked Questions
What is an LLM audit?
An audit of how large language models such as ChatGPT, Claude and Perplexity read and understand your website. It scores your content on extraction, entity clarity, structured data, factual consistency and whether the models will cite you, then lists the specific gaps holding you back.
How is an LLM audit different from an SEO audit?
A traditional SEO audit measures what Google's crawler cares about: speed, crawl errors, keywords, backlinks. An LLM audit measures comprehension — whether a machine reading your page can extract clean facts and trust them enough to repeat them. Sites frequently pass one and fail the other.
Do you need access to my website to run it?
No. We read your public pages, which is exactly what the AI systems do. There is no plugin to install, no analytics access needed and no login to hand over.
How long does the audit take?
We return the report within 48 hours, along with a call to walk through the findings if you want one.
What is a good AI readiness score?
Above 70 generally means the fundamentals are in place. Most UK business sites land between 40 and 70, where the content is sound but the machine-readability is not. Below 40 usually indicates something structural rather than cosmetic.
Is the LLM audit really free?
Yes. No card, no contract, no obligation to buy anything afterwards. Plenty of businesses take the report and implement the fixes themselves.
See how AI actually reads your site
Send us your website and we will run it through ChatGPT, Claude and Perplexity, then come back within 48 hours with your AI readiness score and the specific gaps worth fixing.
Call +44 20 4572 4940 to talk it through →