Home / Insights / AI Search

Does ChatGPT Recommend Your Business? How Irish Firms Show Up in AI Search

By Paul O'Dwyer, Founder, Irish AI Consultancy · Published 26 July 2026 · 8 minute read

Try a small experiment. Ask ChatGPT or Claude the question one of your customers might ask: who to use for your service, in your county. Read the answer carefully. Is your business named? Is a competitor? Is the description of you accurate, or two years out of date?

For most Irish business owners, that experiment is uncomfortable. The assistant either has no idea they exist, or describes them vaguely, or confidently recommends someone else. And unlike a low Google ranking, which at least puts you somewhere on page two, being absent from an AI answer means you were never in the conversation at all. There is no page two.

Why does this matter now?

Because the behaviour has already shifted. People who would once have typed "solicitor Drogheda" into Google and compared ten blue links now ask an assistant "I am buying a house in Drogheda, who should I use?" and get a short written answer naming two or three firms. Google itself now answers many searches with an AI Overview above the traditional results, meaning even people who never open ChatGPT are reading AI-written answers about businesses like yours. The firms named in those answers collect the enquiry. Everyone else is invisible, regardless of how good they are.

The honest caveat: nobody outside these companies knows exactly what share of buying journeys start this way, and it varies by sector. But the direction is one way, the answers are already being generated about your market today, and the cost of checking where you stand is an afternoon.

How do AI assistants decide which businesses to recommend?

There is no submission form and no ad auction. Assistants build answers from what they can read and corroborate, and a few patterns hold consistently. They favour websites that state plainly what the business does, for whom, and where, over websites full of vague slogans. They read structured data, meaning the machine-readable markup that spells out your name, address, services and expertise. They lean on third-party corroboration: directories, review platforms, industry listings, press mentions and articles that confirm you are who your website says you are. And above all they pull from content that directly answers real questions, because that is what they are trying to produce: an answer.

This is why a small, specific business can outrank a bigger, vaguer one in AI answers. The assistant is not measuring your size. It is measuring how confidently it can describe and cite you.

How can I check what AI says about my business today?

That last list is your gap analysis, and it is worth doing before any money is spent. In our experience, owners are rarely upset by being absent; they are upset by discovering the assistant confidently recommending the competitor down the road.

Why is my business invisible to AI?

The usual culprits are mundane. The website talks in slogans rather than facts, so there is nothing concrete to cite. There is no structured data, so machines cannot reliably tell what the business is, where it operates, or what it sells. Key information lives inside PDFs, images or brochureware pages that machines read poorly. Business details are inconsistent across the web: one address on the website, another on old directories. There is no content answering the questions customers actually ask, so when an assistant looks for an answer about your specialty, someone else's page supplies it. And occasionally the website is actively blocking the AI crawlers in its technical settings without the owner ever knowing.

What actually improves AI visibility?

The work falls into four stages, and they build on each other. First, an audit: establish exactly how each assistant currently describes you and against whom, so improvement can be measured rather than guessed. Second, foundations: restructure the website so machines can read it, with plain statements of what you do and where, schema markup, consistent details, and crawler access. Third, content: publish pages that directly answer the questions your customers ask, in plain language with specific facts, because those pages are what assistants quote. Fourth, measurement: re-run the same question set monthly and track whether your presence and accuracy improve, alongside your normal Google rankings.

None of this is exotic, and that is the point. It is disciplined, structured work, and because so few Irish businesses have done it, early movers are competing in a near-empty field. This applies doubly outside Dublin, where AI answers about local services are often built from thin material and one well-structured local firm can dominate them.

Is this just SEO with a new name?

They are related but not the same, and the difference matters when buying the service. Traditional SEO tries to rank a page in a list of links; you win by being clicked. AI search optimisation, sometimes called generative engine optimisation or GEO, tries to make your business easy to read, verify and cite inside a written answer; you win by being named. Good GEO stands on good SEO foundations, and the two reinforce each other, but a site can rank reasonably in Google while being nearly unusable to an AI assistant. If someone is selling you one as automatically including the other, ask them to show you the difference in your own results.

What should I do first?

Run the 20-minute self-audit above before anything else; it costs nothing and it tells you whether you have a problem. If you find gaps, fix the foundations before chasing content volume, because a hundred blog posts on an unreadable site achieve little. And treat it as an ongoing practice rather than a one-off project, since the assistants, and your competitors, keep moving. For what the full programme looks like when we run it, see the AI search visibility section on our site, which is itself built on the same playbook it describes.