Your firm shows up on Google. So why doesn’t it show up in AI?
Your firm ranks on Google, but ChatGPT and other AI assistants keep naming its competitors. The problem may begin with what your website actually says.
A prospective client asks ChatGPT to recommend a firm like yours. It gives them three names. Yours isn’t one of them.
This is not the old problem of being invisible in conventional search. Your firm ranks well in Google, clients find you through it, and you have spent years making sure that happens.
What has changed is where some people begin. They ask ChatGPT, Claude, Perplexity, Gemini or Google’s AI features before visiting a single website. What comes back may be a name, a reason and a link. If your firm is not mentioned, it may never make the prospective client’s shortlist. And more often than you would like, the firm that does get mentioned is a competitor you do not rate.
Why does that happen?
Technical access, third-party mentions and existing search authority all play a part. But one of the most important factors is also the one you control most directly: what your pages actually say.
AI visibility is not only about ranking
Traditional search primarily gives people a set of ranked links. AI-assisted search often tries to assemble the answer itself, using information retrieved from websites and other sources.
Ranking still matters. But AI search visibility also depends on whether those sources contain clear evidence for the answer being produced.
Suppose an assistant is asked to recommend an accountancy firm for owner-managed businesses in East Anglia. Your website may rank for several related searches, but does it ever state plainly that this is who you help?
An AI system does not necessarily need an exact quotation. It does, however, need enough specific material to understand what you do and support what it says about you.
That is where many established firms become difficult to use as a source.
Why vague writing makes you harder to recommend
Read your service pages with that problem in mind.
“We’re passionate about helping your business grow.”
“Partner-led advice, tailored to you.”
“Trusted for over thirty years.”
A person skims past lines like these without much thought. An AI assistant finds little in them that identifies the firm’s actual relevance to a particular question.
Now compare them with this:
“We act for sole traders and owner-managed companies in the East of England on Self Assessment, Corporation Tax, VAT and Making Tax Digital.”
That sentence names the work, the client and the location. It gives a search system something concrete to retrieve and an assistant something it can use to support an answer.
Most professional firms have both kinds of writing on their websites. The problem is that the vague material often buries the specific material.
On a site that has grown for ten years, there may be service pages describing a firm that has since changed, four articles touching on the same subject and one strong technical page sitting somewhere nothing links to. The useful evidence exists, but it is scattered among pages that are outdated, repetitive or unclear.
The result is not necessarily invisibility. It is ambiguity. Faced with several possible firms, an AI assistant may find another one that’s easier to understand and tell the searcher about.
More content is not automatically the answer
The most common instinct to a content problem is often to publish more.
Sometimes that is necessary. A firm may have an important service, sector or location that its website simply does not cover. But established sites are rarely short of words. More often, nobody has decided what the existing posts and pages are for.
Old and overlapping posts can create conflicting signals. Important pages may be weakened by poor internal linking or surrounded by material that no longer reflects your business. Several articles may compete to answer the same question while none of them answers it particularly well.
Adding another article does not resolve that. It gives the website one more page to accommodate.
The more useful starting point is to decide what each page is doing: what still earns its place, what needs sharpening, what should be merged or removed, and what genuinely needs creating.
That is what a content audit is for. It was worthwhile before AI-assisted search began recommending firms. It matters even more when your content may be retrieved, compared and summarised before a prospective client reaches your website.
The judgement no visibility tool can supply
Software can test whether a firm appears for a defined set of questions. It can compare competitors, record citations and track whether the answers change.
The harder problem is deciding which questions matter.
“Insurance for dental practices.”
“R&D tax relief for engineering firms.”
“Employment solicitor for a board-level exit in Manchester.”
Questions like these determine whether the right prospective client finds you. Some may already be suggested by your website, but the site cannot reliably reveal which sectors the partners want to grow, which locations matter most or which instructions the firm is pursuing this year.
Those priorities come from within the business.
They come out in conversation before the audit begins. Without that conversation, you may get a thorough measurement of prompts that have little commercial importance. Anyone can hand you a visibility score. The valuable work is deciding what that score should measure and what to do about the pages behind it.
What fixing the content involves
On a small website, this is largely a reading job.
You assess the pages against the questions that matter, then decide what to keep, improve, merge, remove or create. The pages worth keeping may need to be rewritten so that they explain the firm’s work with enough precision for both clients and search systems to understand.
On a site with hundreds or thousands of pages, nobody can hold the whole thing in their head. Reading a sample leaves most of the site unchecked and makes it easy to miss patterns spread across different sections.
That is the gap Manticore helps me close.
Manticore is software I built for my own content audits. It examines every page in the agreed scope, applies consistent checks and records the page-level evidence behind its findings. I then weigh that evidence against the questions and commercial priorities established during discovery.
The software does not make the decisions.
It observes and records. I read the evidence, consider what the page is meant to achieve and make the recommendation. Your firm decides whether to act on it.
That distinction matters. A tool may find two pages covering similar ground. It cannot know that one supports a valuable client conversation, that the other is required by a regulator or that the partners intend to stop offering the service altogether.
Content is important, but it is not the whole answer
Some elements of AI visibility are not writing problems.
Your website must be accessible to the relevant search systems. Structured information should agree with the visible content. Your details should be consistent across registers, directories and review sites. Third-party coverage and citations may help confirm what your own website says. Someone should also track whether the firm’s visibility changes over time.
There is no special piece of markup that guarantees inclusion in an AI answer. Nor can stronger copy compensate for every technical or reputational problem.
That work matters, and there are specialists who do it well. Where an audit identifies those needs, I will point you towards them.
My part comes earlier: establishing whether your content explains the firm’s relevance accurately and specifically enough to be understood, trusted and repeated.
Your firm may perform well in conventional search because it has established authority and pages that satisfy the signals those systems use. AI-assisted results can expose a different weakness: your website may be discoverable without being clear enough to support a recommendation.
Fixing that begins with what your pages say, although it does not end there.
If your firm is missing from answers it should be winning, tell me what you want to be found for. I will tell you whether a content audit — carried out with Manticore where the scale calls for it — could help.
Practical questions
A few things people ask
A place for the questions that come up when a firm is missing from AI answers.
Why doesn’t AI recommend our firm?
Usually because of what your website says.
An assistant builds its answer from sources it can understand. If your pages describe the firm in general terms, there is not much for it to work with.
It is less a matter of being missing from the web than of being hard to read once you are found.
How does ChatGPT decide which firms to recommend?
Not the way Google ranks pages.
It works from the sources available to it, looks for clear and consistent information about who does what, and assembles a firm it can describe with some confidence.
Specific, accurate, well-organised content makes that easier. Outside signals – directory entries, reviews, other mentions – feed into it too.
Is AI search visibility the same as SEO?
No, though they overlap.
SEO is mostly about ranking your pages for particular searches. AI search visibility is about whether an assistant can use your website to describe and recommend you when someone asks.
Some of the groundwork is shared, like clear content and a sound structure. But a page can rank perfectly well and still be too vague to quote.
Do you handle the technical side, such as schema markup?
No, but I can tell you why it’s important.
Schema markup, keeping your details consistent across registers, directories and review sites, and tracking where you get cited over time all play a part. That work matters, and there are specialists who do it well, so I will point you to them where an audit shows it is needed.
My part comes first: making sure your content explains the firm accurately and specifically enough to be understood and repeated. Better markup cannot rescue vague copy, and strong copy cannot fix every technical problem.
Can you guarantee we’ll be cited by AI?
No. Anyone who promises it is overselling.
There is no setting that drops a firm into an AI answer. What I can do is find where your content is letting you down and make sure it says something accurate and specific enough to be understood and repeated.
That improves your chances. It does not buy a result.
Do we need more content, or an audit first?
Usually an audit first.
Most established firms do not lack words. They lack a decision about what the existing pages are for.
An audit shows what to keep, sharpen, merge or remove before you spend on anything new, so you are not adding pages to a site that is already hard to read.
Know someone wondering why AI recommends their competitors?
Send them this before they publish another article.
