Somewhere right now, a potential customer is typing a question like this into ChatGPT:
"Who's a reliable wedding photographer in the Cape Winelands?"
"Best guesthouse in Paarl for a weekend away?"
"I need a good plumber near Wellington — any recommendations?"
The AI answers confidently, names two or three businesses, and the customer phones one of them. No Google search, no scrolling, no comparing ten tabs. If your business was named, you just got a lead while you slept. If it wasn't, you'll never know the enquiry existed.
So how does the AI choose? It's not advertising (you can't pay ChatGPT to be recommended), and it's not random. Here's what's actually happening, in plain English.
Where the answer comes from
AI assistants build recommendations from two layers.
The first layer is everything the AI learned during training — an enormous snapshot of the public internet. If your business has been written about over the years (in news articles, travel guides, industry round-ups, directories, blogs), traces of that made it into the AI's general knowledge.
The second layer is live web search. Modern AI assistants — ChatGPT with search, Perplexity, Google's Gemini — go and look things up in real time, then summarise what they find. When someone asks for "the best guesthouse in Paarl", the AI quietly runs searches, reads a handful of pages (round-up articles, review sites, business websites, Google Business listings), and composes its answer from them.
That second layer is where the game is won or lost today, and it explains the single most important idea in this whole subject: the AI mostly doesn't recommend businesses directly — it recommends businesses that trusted sources already talk about.
The signals that get a business recommended
Across the major AI tools, the same handful of signals keep deciding who makes the answer:
1. Independent mentions. When several sources the AI trusts — local news, industry publications, "best of" articles, established directories — all mention your business, the AI reads that as consensus. One glowing page on your own website is your opinion; five independent mentions is evidence.
2. Consistent basic information. Your name, address, phone number, hours and services should match everywhere they appear: your website, your Google Business Profile, directories, social profiles. AI systems are built to be cautious. Contradictory information ("open Sundays" on one site, "closed weekends" on another) makes recommending you feel risky, so it doesn't.
3. Reviews with substance. Star ratings matter, but AI reads the words. A steady flow of recent, detailed reviews ("the owner personally sorted out our geyser on a Sunday") gives the AI concrete, quotable reasons to recommend you. Fifty reviews from 2019 and silence since reads as a business that may have declined.
4. A website that answers questions clearly. AI systems parse your site the way a hurried reader would. If a machine can't quickly establish what you do, where, for whom, at what price range, and why you're different — it moves on to a competitor whose site makes that effortless. Behind the scenes, structured data (a technical layer that labels your business information for machines) makes this even easier.
5. Actually being the answer to specific questions. AI queries are conversational and specific: "pet-friendly guesthouse in Paarl with a pool", "electrician who does solar installations near Wellington". Businesses whose websites address those specific questions, in those words, get matched to them.
Why great businesses stay invisible
Notice what's not on that list: how good you actually are. AI has never tasted your wine, slept in your rooms or seen your workmanship. It can only judge the paper trail — and plenty of excellent, decades-old businesses have almost none. No press mentions, an out-of-date Google profile, a website last touched in 2018, reviews nobody ever asked customers to leave.
To a human in your town, you're an institution. To an AI assembling an answer from sources, you barely exist. That gap — between how good you are and how visible you are — is exactly what Generative Engine Optimization (GEO) closes.
What you can do about it
The encouraging news: every signal above can be deliberately improved. Fix the consistency of your information. Bring your Google Business Profile fully up to date. Rewrite your key pages so they plainly answer the questions your customers ask. Build a steady review habit. Earn genuine mentions in the publications and directories AI reads.
You can do a surprising amount yourself — our 10-minute DIY visibility check shows you exactly where to start. And if you'd rather have it handled while you run your business, that's what we do.
Get your free AI visibility audit → — five minutes, and you'll know exactly where you stand in AI answers today.
Quick answers
Can I pay ChatGPT or Google to recommend my business?
No. There's currently no way to buy a recommendation inside an AI answer. Visibility is earned through the signals above — which is precisely why early movers gain a durable advantage.
Does this replace normal SEO?
No — it builds on it. Most of the foundations overlap. We've written a full comparison: GEO vs SEO.
How long does it take to start appearing in AI answers?
It varies with your starting point and competition, but because AI tools search the live web, well-executed changes can start surfacing within weeks rather than years. We guarantee measurable improvement in 90 days — or we work another month free.