AI search: what changes when ChatGPT and Google AI answer for your customer

When someone searches for “best café in Santa Cruz” or “web designer in Tenerife”, more and more often they no longer see a list of blue links. They see an answer. Written by ChatGPT, Perplexity, or Google itself through AI Overviews.

The shift is quiet but significant. Your customer no longer chooses which website to visit: they read the answer, decide whether they need to verify it, and only then click on one of the cited sources. Or they don’t click at all.

For a local business, that rewrites part of the rulebook. It’s not enough for your website to show up on Google: it has to be the kind of source AI chooses to cite.

How we got here

AI search went from novelty to infrastructure in less than two years:

• Google AI Overviews was announced at Google I/O in May 2024 and rolled out across more than a hundred countries through 2025. It’s the generated answer box that appears above the classic results.
• ChatGPT Search launched on October 31, 2024 and opened to all users a few months later. It searches the web, summarises, and cites sources.
• Perplexity has been running since 2022 with the “answer + sources” format as its core. By 2024 it had become one of the most-used AI search apps in the world.
• Google AI Mode, a conversational tab inside Search itself, has been rolling out to key markets through 2025 and 2026.

What they have in common: AI reads several websites, writes an answer and cites a handful of sources. The user often reads the answer and, in many cases, doesn’t click on anything.

What changes for a local business

Three visible effects:

Fewer clicks, but better-qualified ones. If AI has already summarised your information, whoever clicks is already convinced and wants to book, call, or see photos. Less raw traffic, more decisive traffic.

Classic SEO still matters, with nuances. AI models cite websites that are already well ranked: they don’t make up sources. But between two sites in similar positions, AI prefers the one with more structured content, verifiable data, and literal answers to the question.

What your website says about your business becomes what AI says. If your website states you open at 9, AI will say you open at 9. If your Google profile and your website say different things, AI tends not to cite you because it can’t decide which source to trust.

Why AI cites some websites and not others

We call GEO (Generative Engine Optimization) the set of techniques to show up in generated answers. But many aren’t new: they’re solid SEO with a focus on how an AI reads.

Data served on a tray. AI prefers websites that hand them the key information (business, address, hours, prices, frequent questions) on a layer designed for machines, not only in the text the user sees. It’s the difference between leaving them a pre-digested summary card and asking them to deduce it from reading the whole site front to back.

Content that answers a literal question. If your page raises the user’s exact question and answers it directly underneath, the chances of AI picking up that fragment go up. It’s the same logic as Google’s classic featured answers, only now it carries more weight.

Consistency across sources. Your Google Business Profile, your website and the directories you appear in must all say the same thing: same address, same phone, same business name. AI cross-references sources, and if it finds discrepancies, it tends to trust another option.

Clear dates and live content. A website with a visible publication date on every article and recent updates weighs more than one without dates. That’s why a modest but up-to-date blog adds so much.

Speed and technical solidity. If your site is slow or buggy, the bots feeding these models crawl it worse or skip it. Speed is still SEO, but now it’s GEO too.

What to do if your business is local

You don’t need to change strategy or rebuild the website. You need to align what you already have so an AI model can read you without effort. In practice, that means looking at five fronts:

Your Google profile, up to date. Without this, nothing else matters. Correct category, current hours, recent photos, a steady flow of reviews. Any AI answer about a local business cross-checks against Google Business Profile, so if that profile is empty or stale, everything else counts for less.

A structured information layer on your website. AI reads a website twice: the visible part you see, and an invisible technical layer where it’s told, in machine format, who you are, where you are, your hours and what you offer. That second layer isn’t added from the WordPress editor or from Wix: it’s technical work that needs to be done properly once and kept up to date.

Question-and-answer content. If you don’t know what to publish, look at what customers ask you. Each frequent question can become a heading with a short answer below it. It’s the structure models digest best.

Consistent data everywhere. Same address, same phone, same name. On your site, in Google Business, in directories, on social. Inconsistencies hurt you twice: in classic SEO and in GEO.

One page per service or main product. Don’t dump every service into one “Services” page. One page per service, with the service name plus location in the title, gives AI a clear target to point at.

If some of this sounds opaque, that’s normal: most are adjustments that live under the hood of the website, and they’re not solved by writing more text. Whoever built your site should know where to touch them; if not, that’s a sign the foundation needs a review.

Classic SEO vs GEO: where to invest?

False dichotomy. Classic SEO is still the foundation AI builds its answers on. There’s no way to skip that layer.

What changes is the “how”: well-placed keywords are no longer enough. AI rewards structured content, consistent data, real authority and concrete answers. The old SEO of “stuffing keywords” doesn’t work anymore; the SEO of “clearly explaining who you are and what you do” works double.

If your website is well built (fast, with a solid technical foundation, with honest and well-structured content), most of the GEO work is already done. If not, fix that foundation before thinking about any advanced technique.

What not to do

Some advice circulating around doesn’t work, and sometimes hurts:

• Filling your site with “according to AI” or citing ChatGPT. It doesn’t give you authority: it takes it away. AI cites sources that look like authority, not sources that name it.
• Generating your website content with AI without review. Models detect patterns of generated text and distrust it. Content needs a human voice, even if AI helps with drafts.
• Creating a page for every variant of a question. That’s the trap of “ten near-identical pages with different keywords”. Google and AI models penalise that. Better fewer pages, more thorough.
• Publishing data and not maintaining it. If you state you open at 9 and then change to 10 without updating the website or your Google profile, you’re feeding AI false data. Outdated information weighs worse than no information.

The summary, without the jargon

AI search is not a fad and isn’t going away. It’s not magic either: behind it are still the same websites, better or worse structured. Your business shows up in those answers if your information is clear, consistent, verifiable and quick to read (for a machine too).

The good news: if you already have a website with solid local SEO, an up-to-date Google profile, honest content and a sound technical foundation, you’re already doing GEO without realising. The only thing that changes is that work now has a new audience: the models writing the answers.

---

Want to know how your website stands against AI search and classic local SEO? Tell us about your case and we’ll give you a no-obligation review.