How to get ChatGPT, Gemini and Perplexity to recommend your business

AIs don't invent who they recommend: they learn it from what's published online about you. These are the concrete levers to get on their radar.

When you ask ChatGPT to “recommend a web design agency in Palma de Mallorca”, the model doesn't flip a coin. It gathers what it has read on the topic — websites, directories, reviews, articles, forums — and builds an answer with the names that appear consistently and reliably. So the practical question is: what is the AI reading about you?

The answer isn't in one single place. It's spread across your website, your Google listing, what your customers say and what others publish about you. That's why there's no magic setting, but rather four levers that reinforce one another. Let's take them one by one, with what we see every day in businesses on the island.

How an AI decides who to recommend

Models work in two ways, and often combine them. One is what they learnt during training: everything they read on the internet up to a certain date. The other is real-time search: faced with a local question, ChatGPT, Gemini or Perplexity look up the web at that moment, read several pages and summarise them. In both cases they're looking for the same thing: names that appear in several independent sources, with details that match and opinions that back them up. To an AI, a business that only its own website talks about is a claim with no witnesses.

This explains something that baffles many owners: the AI recommending a competitor who, in their view, does a worse job. Almost always, the reason is that there's more information about that competitor, more consistent and easier to read. It isn't a question of fairness but of available data — and that can be changed. If all this sounds new, start with what GEO is and why SEO is no longer enough.

Lever 1: a website that says clearly who you are

Your own website has to make crystal clear, in plain text and without ambiguity, who you are, what you do, where and for whom. No empty slogans. An AI needs sentences like “we're a web design and development studio in Palma de Mallorca, specialised in SMEs and local shops”. What's obvious to a human is raw material to the model.

Test it on your home page, your “about us” page and every service page: if someone read only that text, would they know what you do, which area you work in and what kind of customer you serve? The typical faults are that the town only appears in the footer, that services sit in a menu with creative names nobody searches for, or that the website exists only in Spanish when a good share of your customers ask in English or German. If someone asks the AI in German, it has an easier time citing you if you have content in that language; it's one of the reasons we champion multilingual websites for businesses on the island.

Lever 2: consistent presence beyond your website

Your Google profile, industry directories, mentions in local press, consistent profiles: same name, address and phone everywhere. The more your information repeats coherently across the web, the more confident the AI is and the likelier it is to name you.

Start with your Google Business Profile, which Google uses in its own answers and which is one of the most consulted sources on local businesses: correct category, real opening hours, complete services. We have a list of the Google listing mistakes that push you off the map, and nearly all of them confuse AIs as well. Then the directories that genuinely matter in your sector: TripAdvisor or TheFork if you run a restaurant, Booking if you offer accommodation, Doctoralia if you're a clinic, property portals, your professional association or the island's business associations. You don't need to be on fifty sites; you need to be on the relevant ones, with the same details.

Lever 3: reviews and brand mentions

This lever is increasingly decisive. In the AI world, being talked about matters as much as — or more than — links. A real review saying “they rebuilt our website and now bookings come to us directly, without intermediaries” is exactly the kind of evidence a model uses to recommend you with confidence. Ask for them, reply to them, look after them.

What helps most isn't the average rating but what the review actually says. A review that mentions the specific service, the place and the situation — “they fixed our air conditioning in the middle of August in Santa Ponça” — gives the AI the exact context to recommend you when someone asks precisely that. You can't dictate reviews, and you shouldn't try, but you can ask at the right moment and encourage customers to describe what you solved for them. You'll find practical ideas in how to get more Google reviews, and they work for almost any sector. Reply to all of them, in the language the customer wrote in.

Lever 4: make yourself easy to quote

Content that answers specific questions, with data, with your real experience, written the way you'd talk to a client. Models pull out snippets; if your pages are full of clear, verifiable statements, you become a convenient source to cite.

Think about the questions that reach you every week by phone or WhatsApp and turn them into content: one page or section per question, with the answer in the first two sentences. “Do you quote without a site visit?”, “Do you work in the Serra de Tramuntana?”, “What does maintenance include?”. When someone asks an AI that same question, your answer will already be written, with your name beside it. Avoid padding: one specific, checkable sentence is worth more than three adjectives.

Mistakes that keep you out of recommendations

The business name spelt differently on every site, with and without “S.L.”, or the trading name in one place and the registered name in another. Old phone numbers and addresses in directories nobody has checked for years. A website that won't let its content be crawled or hides it in images. Unanswered reviews and an unclaimed Google listing. And the most common of all: never having checked what the AI says about you. Very often the problem isn't that it doesn't name you, but that it describes you wrongly: a service you no longer offer, the wrong area or opening hours from several seasons ago.

How to tell whether you're already being recommended

Test it yourself with the questions a customer would ask, in Spanish, English and German, and on all three platforms. Vary the wording now and then, because answers change. Note whether you appear, where in the answer, what it says about you and which source it cites, and repeat it every few months to see how things evolve. If you'd rather have an organised diagnosis without doing it by hand, the AI visibility test does it for you.

None of these levers is a trick: it's building real digital reputation — only now machines read it too. If you want us to build that presence for your business — website, listings, reviews and content — at Teix we do it end to end as part of our AI search optimisation service.

Sources: OpenAI: overview of OpenAI crawlers (GPTBot and OAI-SearchBot) · Perplexity: Perplexity crawlers (PerplexityBot) · Google Business Profile Help: tips to improve your local ranking · Google: creating helpful, reliable, people-first content