Does ChatGPT actually know my business exists?
Only if the web says so plainly, and in more than one place. ChatGPT answers a local question by running a live search, reading the pages that come back and naming the businesses it can corroborate: a website that states the trade, the area and the services in ordinary sentences, a Google Business Profile that agrees with it, a directory listing or two, and some reviews. A business that lives mainly on Facebook, or whose site is a logo and a phone number, is invisible to that process. So is one whose name, address or services differ from one listing to the next, because the model cannot be sure it is looking at one business.
The quick test is to ask it about you by name. If it gets your services or your town wrong, the raw material is wrong or missing.
Is my website telling it enough?
Probably less than you think. Assistants lift specifics: the services you offer, who they are for, where you work, what you are accredited to do, which jobs you have done and what they involved. A single services page with six bullet points supplies none of that. RB Electrical in Halifax gives every service its own page, EV charging included, with enough on each to be understood on its own, and that structure is what a search engine and an assistant both read first.
One page for each thing you want to be recommended for, written in plain words, is the biggest single change most Halifax sites need.
Why does it recommend a competitor with a worse website?
Because the answer is built from more than the website. Say a rival has forty Google reviews, a listing on the trade body’s site, a write-up in the Halifax Courier and a named project on a supplier’s page. The assistant then has four independent sources telling the same story, and a smarter site with none of that loses to it. Third-party mentions are the part of GEO that most resembles PR, and they are the part businesses most often skip, because nobody is asking for them by name.
Do reviews and case studies really change the answer?
Yes, because they are the evidence the assistant repeats. Ask for a recommendation and the reply usually gives a reason: highly rated, long established, specialises in commercial work, covers the whole of Calderdale. Those phrases come from reviews, from About pages and from jobs that have been written up. Bond Brickwork in Brighouse has years of project pages and evidence from live sites behind it and is recommended by AI search for commercial brickwork as a result. Dolphin Heating in Huddersfield states plainly that Fay is a Gas Safe registered engineer with more than 25 years in the trade, and the AI answer names her for female gas engineer Huddersfield, which is the question her customers actually ask.
A review that mentions the job, the town and the trade is worth several that say nice bloke, would recommend.
What would Halizona fix first?
The facts, then the pages, then the proof. We start by asking the assistants the questions your customers ask and recording what they say about you and about the businesses they name in your place. Next we line up your name, address, services and area everywhere they appear, rebuild the service pages so each one answers a real question in its first sentence, and gather the reviews, photographs and case studies that give a model something to cite. Mentions on other sites come last, because they only count once your own site can back them up.
None of it is a trick, and all of it improves your ordinary Google rankings as well, which is where most of the answers are drawn from in the first place.
How long before ChatGPT starts mentioning us?
Weeks for corrections, months for recommendations. Assistants that search the live web see a changed page the next time they fetch it, so a wrong town or a missing service can be put right quickly. Being named for a competitive question in Halifax depends on rankings, reviews and mentions building up, and that is measured in months, the same as SEO. We run the same set of questions every month and report the count, so you can see the direction before you see the enquiries.
