The short version: AI assistants like ChatGPT, Gemini and Claude are quietly becoming a recommendation engine — when someone asks “who’s the best [X] near me,” these tools name specific businesses, and increasingly specific people to talk to. For an Aussie small business, earning a spot in those answers is worth real money. But it isn’t a website trick you can buy as a “package.” It’s an integrated play: a credible, well-structured site plus the third-party trust signals — Google reviews above all — that every AI engine reads. Here’s what I found testing it on a real client, and what’s actually worth your time.
There’s a new question worth asking about your business, and most owners haven’t thought to ask it yet: when someone types “who’s the best [what you do] in Brisbane” into ChatGPT or Google’s AI, does your name come up?
Because more and more, that’s where the buying decision now starts. Not a page of blue links your customer scrolls through and judges for themselves — a single, confident answer from an assistant they’ve decided to trust. And whoever gets named in that answer walks into the conversation already vouched for.
I went down this rabbit hole properly on a client project recently, and what I found changed how I think about the whole thing. So this isn’t theory. It’s what happened.
What actually happens when you ask an AI to recommend a business?
It recommends specific businesses — and, increasingly, specific people — pulling heavily from your online reputation to decide who to trust.
I was doing some AI-visibility work for a client and ran what I thought was a simple test prompt: “I need to buy a deep fryer for my restaurant in Brisbane, where can I go?”
Both Gemini and Claude came back naming businesses — and here’s the part that stopped me — naming people at those businesses. Not “try Company X,” but “speak with John at Company X, he’s very attentive.” I hadn’t asked who to talk to. I hadn’t dug into any particular company. The assistant volunteered a named human being to go and speak with, unprompted.
That’s when it clicked. The AI was pulling from Google reviews and treating them as a human trust signal — and weighting them so heavily that a reviewer’s offhand praise of “John” was surfacing as a recommendation to go and ask for John by name. My client, at that point, wasn’t in the list at all.
Can you actually influence it — or is it a black box?
You can influence it, and the lever that moved the needle fastest was the least glamorous one: Google reviews.
Knowing what I’d seen, I set the client up with a simple review-capture system — nothing fancy. An automated follow-up after each purchase asking for a review, plus a short email template we gave their sales reps to send by hand whenever they finished a job. That’s it. No “GEO package,” no magic.
Then I waited, and re-ran the same prompt sequence — this time in a fresh, public session with no history, so there was no chance the AI was just remembering my earlier questions. The result: the client came back second on the list of companies, where before they’d been completely omitted. And exactly like the others, the assistant singled out the person named in their new reviews as the one to speak with.
The outcome showed up in the numbers too. In Google Analytics, sessions referred from AI tools — ChatGPT, Gemini, Claude — went from essentially zero to roughly 150 over about six weeks. A traffic channel that simply didn’t exist for this business before the reviews went live.
Now, I’ll be straight about what this is and isn’t. The reviews were the change I made, the timing lines up tightly, and the clean-session re-test rules out the obvious flukes — but this is real-world correlation plus a clear mechanism, not a lab experiment. I’m confident enough to have built it into how I work. I’m not going to pretend I ran a controlled trial.
Does this even matter yet for a small business?
Yes — and I say that as someone who’s usually the first to tell clients not to chase shiny things.
Here’s why it’s worth it. When an AI recommends your business, or Google drops you into its AI Overview, you get something an ad can’t buy: the assistant’s borrowed credibility. More and more people treat an AI’s answer as fact — for better or worse — so being the name it hands over is like earning the top spot a paid ad would charge you for, except with trust attached that no ad ever carries.
Even modest effort here is worth it, and if it doesn’t pay off immediately, it compounds. It’s the same shape as SEO: a long game where the goodwill you build now keeps paying later. Getting in early, while most of your competitors haven’t clocked it, is the whole advantage.
What the hype-merchants get wrong
There’s a growing crowd selling “AI visibility” or “GEO packages” and shouting that this is mission-critical, do-or-die, buy-now. Treat that noise with the suspicion it deserves.
It’s not the end of the world if you’re not in every AI answer tomorrow. And it is emphatically not a switch someone flips for a fee. Anyone promising guaranteed AI rankings for a monthly retainer is selling the SEO-snake-oil playbook with a fresh coat of paint. There’s no pay-to-appear mechanism in these tools — that’s rather the point of them.
The honest version is less sexy: it’s earned, it’s gradual, and a big chunk of it is stuff you should be doing anyway.
The uncomfortable truth: your website isn’t enough on its own
If there’s one thing to take from my deep-fryer rabbit hole, it’s this — you cannot win this from your website alone. It’s an integrated game.
The single biggest lever in my client’s case lived entirely off their site: Google reviews. The AI was reading third-party trust signals — what other people say about you — and weighting them heavily. No amount of clever copy on your own pages replaces that, because of course it doesn’t; you saying you’re the best means little, a wall of genuine reviews saying it means everything.
That said, your site still does real work, especially for Gemini, which leans on Google’s understanding of your business — and Google crawls your site directly. My honest take is that strong, credible traditional SEO and a well-built site gets you most of the groundwork done, because Google’s own AI leans on it and the other engines lean on the same public signals. But “most of the groundwork” isn’t “all of it,” and no single engine is the whole picture. Remember my test lit up across Gemini, Claude and ChatGPT — three different systems drawing on different sources. Win Google and you’ve got a serious head start. You haven’t automatically won everywhere.
From the builder’s chair: what your website actually needs
So what do you do on the site itself? Nothing exotic. You make it easy for a machine to understand exactly who you are, what you do, and where.
That means a clean, simple, correctly structured site — the same discipline I bang on about with heading tags and page structure. A crawler (and the AI reading through it) should be able to move through your site and come away with a clear, unambiguous picture of the business. I still use schema markup where it’s useful, even though Google has downgraded it as a direct ranking signal, because it helps crawlers and other engines understand the site — and understanding is the whole game here.
And the thing to avoid: rushing out a pile of thin, AI-generated slop to “feed the machines.” That backfires. The engines are actively getting better at spotting and demoting low-value content. A small, clear, genuinely useful site beats a big vague one every time. Structure simply, write for humans, mark it up so machines can follow along.
Most of that? It’s just good SEO and good UX — the work that makes your site better for actual customers too. Which is rather the point: the businesses that show up in AI answers tend to be the ones that quietly did the fundamentals well.
Where to start
If you want to get ahead of this while your competitors are still ignoring it, you don’t need a big budget or a buzzword-laden package. You need the fundamentals, done properly and consistently: a steady stream of genuine Google reviews, a clean and credible website a machine can actually read, and the patience to treat it as the long game it is.
That’s the work we do at Kartia — building sites that are structured to be understood, by customers and search engines alike, and helping Brisbane businesses get the foundations right rather than chasing the trend of the week. If you’re wondering whether AI is recommending you or quietly sending your customers elsewhere, let’s find out.
Frequently asked questions
How do AI tools like ChatGPT and Gemini decide which businesses to recommend?
They synthesise a recommendation from the information they can find and trust about you, and they lean heavily on third-party signals — Google reviews chief among them. In my own testing, the AI weighted reviews so heavily it would name a specific person mentioned in them as the one to speak with. Your own website matters too, especially for Gemini which draws on Google’s view of your business, but reputation signals from elsewhere on the web carry serious weight.
Can I pay to have my business appear in AI recommendations?
No. Unlike Google Ads, there’s no paid-placement mechanism in tools like ChatGPT, Gemini or Claude. Visibility is earned through genuine credibility signals — reviews, a clear and well-structured website, and being consistently and accurately represented online. Anyone selling a guaranteed “AI ranking” package should be treated with caution.
Do Google reviews really affect AI recommendations?
In my experience, significantly. After setting a client up with a simple review-capture system, they went from being omitted entirely to appearing second in a clean-session AI test, and AI-referred website traffic went from essentially zero to around 150 sessions in roughly six weeks. Reviews are a trust signal these tools read directly, so a steady stream of genuine ones is one of the highest-leverage things a small business can do.
Is optimising for AI search different from normal SEO?
There’s a lot of overlap. Much of what helps an AI understand and recommend your business — a clean site structure, clear plain-language content about who you are and what you do, schema markup, strong reviews — is simply good SEO and good UX done well. The main addition is paying deliberate attention to off-site trust signals like reviews, and making sure your site is genuinely easy for a machine to read rather than padded with thin, AI-generated content.
