Your Next Customer May Ask ChatGPT About You Before They Ever Visit Your Website

A plumbing firm in Leeds, a payroll bureau in Bristol and a software start-up in Manchester have something in common this year. A growing share of the people who might buy from them never type a search into Google and scroll through ten blue links. They ask a question in ChatGPT, Perplexity or Google's own AI answers, read a short paragraph, and pick from the two or three names that paragraph mentions.

If your business is one of those names, you get the enquiry. If it isn't, you never find out you were in the running.

That shift is what people mean by AI SEO. The term sounds technical, but the idea is simple: making sure the AI tools your customers use know who you are, describe you accurately and recommend you when someone asks the questions you can answer.

Why this is no longer a problem for "later"

Google started showing AI Overviews to UK searchers in the summer of 2024, and they now sit above the normal results for a large share of everyday questions. ChatGPT and Perplexity added live web search, so their answers cite real pages rather than relying only on what the model learned in training.

The click data was already moving before any of that. SparkToro's 2024 zero-click study found that out of every 1,000 Google searches in the US, only 360 clicks went to the open web. In the EU the figure was 374. Most searches already ended on Google's own pages or with no click at all. AI answers push further in the same direction: the reader gets a summary and a shortlist, and the shortlist is where the competition now happens.

Gartner predicted in February 2024 that traditional search engine volume would fall by 25% by 2026 as people turn to chatbots and AI agents. Whether the real number lands at 15% or 30%, the direction matters more than the exact figure. Buyers are asking machines first.

What AI tools actually look at when they recommend a business

There is no single algorithm to "beat", and anyone promising a guaranteed top spot in ChatGPT is overselling. But after watching thousands of AI answers in different industries, a few patterns are consistent.

They lean heavily on third-party sources. When an AI tool answers "which accounting software is best for a small UK business?", it rarely quotes the software companies' own homepages. It quotes comparison articles, trade publications, review sites, forums and the occasional detailed blog post. If you are missing from the pages it trusts, you are usually missing from the answer.

They reward pages that answer the question directly. A page that opens with a clear, specific answer, followed by detail, gets quoted far more often than one that opens with a slogan. "Our standard boiler service costs £85 and takes about an hour" is something an AI tool can lift. "We deliver excellence in heating solutions" is not.

They need consistent facts. If your prices, opening hours, service areas or product names differ between your website, your Google Business Profile and a directory listing from 2021, the tools hedge or skip you. Consistency reads as reliability.

Each engine behaves a little differently. Perplexity shows its sources prominently and tends to favour recent pages. Google's AI answers lean on pages that already rank well in Google. ChatGPT's search mode often quotes well-known domains and distinctive, specific statements. A business can be cited by one engine and invisible on another for the same question.

A practical starting point for a small team

You do not need a new department to start on this. A few hours a month goes a long way if it is spent on the right things.

1. Ask the questions your customers ask

Write down ten questions a real prospect would type, in their words rather than yours. Not "best B2B payroll solution", but "who can run payroll for a 12-person company in Bristol?" or "is it cheaper to outsource payroll than use software?". Put each one into ChatGPT, Perplexity and Google, and note three things: whether you are mentioned, who is mentioned instead, and which web pages the answer cites.

That last column is the most useful. It is a ready-made list of the sources the AI already trusts in your market.

2. Fix the pages you already have before writing new ones

Most businesses already have a page that should be winning those questions: a service page, a pricing page, a FAQ. Often it simply doesn't say the thing plainly. Add the direct answer near the top. Put real numbers in where you can: prices or price ranges, turnaround times, areas covered, minimum order sizes. Use headings written as questions where that fits naturally, and keep the copy about the customer's problem rather than your company history.

Rewriting a good page from scratch is rarely necessary and can lose the Google rankings it has already earned. Small, specific additions usually do more.

3. Get onto the sources the AI is already citing

If every answer about your category cites the same two trade publications, one comparison site and a review platform, those four are your priority list. That might mean pitching a useful article to the trade title, making sure your profile on the review site is complete and current, or contacting the author of a comparison piece that left you out with something worth adding. This is ordinary PR and partnership work, aimed more carefully than before.

4. Make the technical basics boring and correct

Check your robots.txt file isn't blocking the crawlers that power AI search. OpenAI, for example, uses a separate crawler for ChatGPT search from the one it uses for model training, and some sites block both without realising. Make sure your key pages are indexed by Google and Bing, load quickly, and carry structured data for your organisation, products or services and FAQs. None of this wins on its own, but getting it wrong can quietly take you out of the running.

5. Measure it like any other channel

Rerun the same set of questions every month and keep a simple log: mentioned or not, position in the answer, which sources were cited. Within a quarter you will see whether your changes are moving anything. The answers do vary from one run to the next, so look at the trend across several questions rather than reacting to a single result.

When it is worth getting help

For a sole trader with a handful of local competitors, the manual approach above is enough. It gets harder when you sell in several markets, compete with dozens of firms, or need to show results to a board or investors. Checking hundreds of questions across four or five AI engines by hand every week is not a sensible use of anyone's time.

That is where specialist AI SEO services come in. Platforms such as RankCited track the questions buyers ask across ChatGPT, Perplexity, Google AI Mode and Gemini, show which sources each engine cites and which competitors are recommended instead, and turn the gaps into a plan: which existing pages to improve, what new content is worth writing, and which publishers the engines already trust. Whichever route you choose, ask any provider to show you the specific questions they will track and the evidence behind each recommendation, and be wary of anyone who guarantees rankings in a system nobody controls.

The businesses that move first will set the shortlist

AI answers are still young, and the sources they lean on in most industries are not fixed yet. A local firm that publishes clear answers, keeps its facts consistent and earns a mention in the two or three places its market reads can appear next to much larger competitors today. In two years, when those answers have settled and the same names keep being repeated, breaking in will cost a lot more.

The question worth asking this week is not whether AI search will matter to your business. It is what ChatGPT says when one of your customers asks about the thing you sell, and whether you would be happy with the answer.

‍