How to get your business cited by ChatGPT, Claude and AI Overviews

To get cited by ChatGPT, Claude or Google's AI Overviews, publish pages that answer one question completely in one self-contained passage, attribute every figure to a named source, and structure the page so a machine can lift the answer cleanly. Then make sure the rest of the web agrees on who you are: a consistent name, service list and location across the directories, reviews and registers that assistants read. That is the whole discipline in two sentences. It has a name, generative engine optimisation, or GEO, and for service firms it is fast becoming the second half of being found.

Generative engine optimisation is the practice of shaping your content and your wider web presence so that AI assistants name your business when they answer a relevant question. Where SEO earns you a position on a results page, GEO earns you a mention inside the answer itself.

The two are not rivals. Assistants still retrieve pages much the way a search engine does before they quote them, so a page that cannot rank cannot be cited either. The groundwork that makes programmatic local SEO work, genuinely useful pages with real information on them, is the same groundwork GEO builds on. What changes is the finish: the page must now read well to a machine assembling an answer, not just to a person scanning a list of links.

Google reported that AI Overviews reach over two billion users a month (Google, 2025). A growing share of buying research now starts and ends inside a generated answer: the prospect asks who they should use, reads the names the assistant offers, and contacts one of them. If your firm is not in that answer, you have not lost a click. You were never in the running.

For professional and local services this lands hard, because the questions assistants handle best, who does this near me, is it worth it, what does it cost, are exactly the questions your next client asks first.

Assistants prefer passages they can quote without having to repair them, from sources they can identify and trust. The clearest evidence is the Princeton-led study that named the field: across thousands of queries, the right content changes raised a source's visibility in generated answers by up to 40% (Aggarwal et al., 2024). The changes that worked were not tricks. Adding statistics, quotations and citations from credible sources, the things that make a passage verifiable, were among the strongest performers.

The study, GEO: Generative Engine Optimization, is published in full (Aggarwal et al., 2024). Its authors also found that the effect varied by domain, which is a reason to test these levers on your own pages rather than assume them.

On your own site, optimising for citation means writing so that any single passage can stand alone. Six habits cover most of it:

• Put the question in the heading and answer it fully in the first sentence or two underneath, the way this post's opening does.
• Make each passage self-contained, so it still makes sense lifted out of the page and dropped into an answer.
• Attribute every figure to a named source and year. An unsourced number gives an assistant a reason not to quote you.
• Use plain structure: descriptive headings, short paragraphs, and a real table when you are comparing options.
• Mark up question-and-answer sections with FAQ schema generated from the same text the page displays, so the markup never drifts from the words.
• Define your terms once, in plain language, the way a glossary does, so engines can connect the term to your business.

Assistants do not form their view of you from your website alone. They read the wider web: directories, review platforms, professional registers, local press, the places that confirm a firm is real and reputable. The practical work here is unglamorous consistency. The same business name, the same services, the same location everywhere you appear, with reviews that describe what you actually do. A firm described consistently in five independent places is a safer entity for a machine to name than one that is brilliant on its own site and invisible everywhere else.

There is no Search Console for assistants yet, so checking is manual but quick. Ask ChatGPT, Claude, Perplexity and Google the questions your clients actually ask: who they should use for your service in your town, whether the thing you sell is worth it, what it should cost. Note whether you are named, whether your pages are cited as sources, and who is named instead. Repeat it monthly and you have a working visibility log, and a list of the questions your site does not yet answer.

One honest caveat. Being named by an assistant produces an enquiry like any other, and enquiries are where service firms have always leaked. In one audit of 433 B2B companies, 55% took more than five working days to reply to an enquiry, or never replied at all (Drift, 2017). A firm that wins the citation and then leaves the resulting message in an unattended inbox has paid for visibility and thrown away the return.

This is why we treat GEO as one half of a system. The SEO and GEO work earns the mention; the AI employees behind the site answer the enquiry it produces the moment it lands, qualify it, and book it, at any hour. Visibility and follow-through, run as one thing, with pricing set against the value of the work it wins.

If you want to know what the assistants currently say about your firm, that is a question with a quick answer. A thirty-minute call covers where you are cited today, where your competitors are, and what closing the gap would be worth.