Your future clients are asking ChatGPT who to hire. Is it saying your name?

AI assistants recommend businesses that appear consistently across sources the model already trusts, rather than whichever business ranks first in traditional search. The engine breaks a question like "best emergency plumber in Manchester" into several sub-questions, retrieves candidate passages from its index, and assembles an answer from the ones that survive reranking for relevance and trust. What decides inclusion is therefore the breadth and consistency of your presence: a Google Business Profile with genuine reviews, listings in the directories relevant to your trade, a website that answers specific questions directly, and identical facts about your business everywhere it appears. Contradictions across sources are one of the most reliable reasons a model hedges or names a competitor instead of you.

No. Google's May 2026 guidance on optimising for generative AI features states directly that you do not need llms.txt, AI-specific markup, or Markdown mirrors of your pages, and that optimising for AI Search is optimising for Search. The eligibility floor for appearing in any AI answer is ordinary technical SEO: your page must be crawlable, indexed and eligible for snippet display. A page carrying a nosnippet directive or blocked in robots.txt cannot appear in AI Overviews no matter what else you do. Most of what is currently sold as "AEO", "GEO" or "LLMO" is conventional SEO with a new label and a higher invoice. Publishing an llms.txt file costs nothing and does no harm, but treating it as an AI visibility strategy is a mistake.

Self-contained passages that answer one specific question completely, in one place, without depending on the surrounding page. The practical shape is a heading phrased as the question a real person would type, followed immediately by a direct answer in the first sentence, then the supporting detail, the caveat and the evidence. Passages in roughly the 130 to 170 word range perform best as citation units: long enough to carry a claim plus its qualification, short enough to be quoted whole rather than paraphrased. Paraphrase matters because it is far less likely to carry a link back to you. Including a number, a named entity and a visible verification date measurably improves selection odds, because the model has no other cheap way to judge whether your figure is current.

In rough order of leverage for a local service business: your Google Business Profile, because it feeds both Google's AI features and general entity understanding; Bing Places, which feeds ChatGPT directly and is routinely neglected; Reddit, which is the single most-cited source across major AI engines and is where models look for candid opinion; trade-specific directories and review platforms such as Checkatrade, Yelp or Clutch depending on your sector; and LinkedIn for company and founder identity. The consistency of facts across those venues matters as much as being present on them. Same legal name, same address format, same founding year, same one-line description everywhere. Wire that consistency into Organization schema on your own site with sameAs links pointing at every profile you control.

Write down the ten questions a genuine customer would ask before hiring someone in your trade, then run each one through ChatGPT, Google's AI Mode, Perplexity and Copilot and record three things: whether you appear at all, whether you are cited with a working link, and which competitor is named instead of you. Do this monthly and keep the results, because answers are non-deterministic and personalised, so a single test tells you very little while a trend over three months tells you a lot. If you appear nowhere for questions you have written a dedicated, well-evidenced page about, the problem is almost always eligibility rather than content quality, and you should check indexing and crawler access before rewriting anything.