GEO: How to Get Found by ChatGPT (and Why Google Isn't Enough Anymore)

I open Perplexity on a rainy Wednesday morning in London, in a small café near Marylebone High Street that serves a surprisingly honest espresso — almost worthy of a bar on Via Roma. Outside, the grey London sky wraps around passers-by darting between shop windows. Inside, the air smells of cinnamon and steam.

I type: "agenzia comunicazione e pubblicità Londra con esperienza mercato italiano".

I wait. The cursor blinks for a moment, as though it is thinking. Then I read the response.

Silence. Three awkward seconds — the kind you feel when you tell a joke that nobody at a gala dinner understands. Perplexity lists two agencies in Manchester, one in Liverpool with an "international team", and a boutique London agency with an "Italian division". But Insight ADV? My background, twenty years of work in communications between two cultures? Nothing. Vanished.

I try again with narrower variations: "agenzia pubblicità Londra comunicazione integrata", "marketing e comunicazione inglesi per aziende italiane nel Regno Unito". Nothing. For the artificial intelligence composing that response in real time, drawing on billions of parameters, we do not exist. Or rather: we are not "citable" enough to be included as an authoritative source at that precise moment.

And yet Insight ADV is very much there. A full-service London agency, founded by an Italian with twenty years of history in communications, working every day between the British market and Italian clients. Communication campaigns, brand identity, advertising, content strategy: daily work across two markets, two cultures, two languages. But in this new ecosystem we are, to use a brutally precise term, a ghost brand.

This is the exact moment when you realise the game is no longer simply about outranking the competition on Google. A new digital space is emerging — an "answer engine" — and many of us have simply not yet received an invitation to the party.

What does it mean, in concrete terms, to exist for an AI in 2026? Welcome to the era of GEO: Generative Engine Optimisation.

First of all: what is really happening (beyond the hype)

Let us be clear about this, without beating around the bush: SEO is not dead, but it has just undergone an irreversible genetic evolution. Where until two years ago our goal was to appear in Google's SERP (the famous list of blue links), today the entry point for search has multiplied and fragmented.

The paradigm shift: from index to conversation

The Google model (2000–2024): it worked like a vast library. You searched for a book, the librarian pointed you to the shelf. You went there, opened the website, and found the answer yourself.

The AI model (2025 onwards): it works like a consultant sitting across from you. You ask a question, they have already read every book in the library and synthesise the verdict for you. They do not say "go and read this" — they say "here is what you need to do". And only if you ask them will they tell you which source they drew from.

The figures are stark and come directly from Gartner: by 2026, the volume of searches on traditional search engines will fall by 25%, with search traffic progressively ceding market share to AI assistants and chatbots (source: Gartner, February 2024).

But there is more: according to analysis by Bain & Company, 60% of online searches already end today without the user clicking on any website. Semrush, in a study from September 2025, estimates that 93% of AI search sessions end without a visit to a website. We are not talking about future trends. We are talking about right now.

Someone searching for "marketing consultant Caserta" on Perplexity is not browsing through websites. They are collecting a verdict. If you are not in that verdict, you are simply not in the game.

Google has responded by transforming search with AI Overviews (previously in beta as SGE — Search Generative Experience), a feature now progressively available in European markets as well. Where before you searched and clicked, today you read and act. This is what we call "the answer era" (we have written about it at length in this in-depth piece: The Answer Era). We are no longer searching for information: we are searching for pre-processed solutions. And if you want to be the main ingredient in that solution, you need to radically change the way you communicate with machines.

GEO: the new name for an old problem

At Insight ADV we are fond of simple metaphors because they make strategy digestible. If the old SEO was like putting your name in the telephone directory in the right place — a matter of order, labels, and alphabetisation — GEO is something far deeper.

It is like becoming the example that an esteemed professor uses during a university lecture. If the professor cites you as a case of excellence in sustainable packaging in Campania, dozens of versions of that citation circulate among students, in notes, in online summaries. Nobody went looking for you actively on a directory. You have simply become part of shared knowledge. You have become a thematic authority.

The fundamental distinction nobody explains to you

The acronyms are often muddled, so let us set the record straight:

GEO for content: the technique for optimising an article so that an AI uses it as a textual source to generate a response. It is based on structural clarity, verifiable data, and direct answers (we discussed this here: The Answer Era).

GEO for brand: the construction of the reputation of a brand or professional figure within the datasets and sources from which LLMs (Large Language Models) draw. It is not just about what you write, but about who you are within the digital ecosystem.

In this article we focus on the second point: how to ensure that ChatGPT mentions your name when someone asks it for a recommendation. It is not merely a technical matter. It is digital reputation for algorithms.

The anatomy of the verdict: how AI decides who to cite

Artificial intelligences are not magical, even if they sometimes seem to be. They are omnivorous, tireless readers — and, above all, statisticians. To decide who to cite, they operate across three levels of depth that we must occupy if we want our SMEs in London, Naples, or Caserta to emerge from invisibility.

1. Primary sources: the "training data"

LLMs were trained on an oceanic mass of texts gathered up to a certain date. If your brand has never appeared on Wikipedia, on historically significant regional newspapers — I am thinking of Il Mattino, Corriere del Mezzogiorno, Repubblica Napoli — or in authoritative professional directories, you are not part of the model's basic vocabulary. You are a neologism the AI has not yet learned. You start with a reputational handicap that has nothing to do with the quality of your work, and everything to do with your absence from the written conversations of the past.

2. Real-time sources: augmented search

Tools such as Perplexity and the "connected" versions of ChatGPT perform a rapid scan of the web before responding. Here, traditional SEO still counts, but the AI does not stop at the first result. It looks for cross-referenced confirmation. If I search for an "expert in food photography in Naples" and find your name on three different industry blogs, in a local news article, and in a verified directory, the AI applies a triangulation logic: "All right, this is a real reference, cited by different sources in similar contexts."

3. Co-occurrence: thematic authority by association

This is the least well-known and most powerful mechanism in GEO for brand. LLMs cite those who are associated through co-occurrence with a given topic.

If every time the internet discusses "bespoke marketing" (Bespoke Marketing) the name Insight ADV appears in the same paragraph — on LinkedIn, in an interview, in an industry thread, in a podcast comment — the AI learns statistically that the two concepts are twins. When asked about bespoke marketing, our name will emerge naturally as the logical completion of the sentence.

This is not about backlinks. It is about repeated contextual presence. It is the difference between having a phone number in the directory and being the name everyone uses as an example.

The Campania case: why we are behind (and why that is a huge advantage)

Let us play devil's advocate. I am Italian, I come from Campania, and I work every day with Italian clients from London. This gives me a privileged vantage point on a problem that concerns me directly: the Campanian market — and the southern Italian market in general — is almost entirely absent from AI responses on topics of marketing, communications, and high-level professional consultancy.

If I run the same test in London, searching for agencies in Camden or Shoreditch, the results are dense, precise, and up to date. British agencies understood the "extended digital footprint" game earlier. They invested in digital PR, attended to structured data, and established themselves in international directories.

In Campania we are still anchored to word of mouth, or at best a Facebook page managed in spare time that publishes photos of coffee and sunsets. Lovely for the soul, disastrous for GEO.

Why do I say this is an advantage? Because for an SME in Caserta or Naples, the cost of "colonising" this space today is still extremely low. Those who move now are not competing for a position already occupied by a thousand others: they are planting their flag on virgin territory. If the AI finds nobody genuinely authoritative in a specific niche in our area — say, "Vietri ceramic design consultancy" or "Campania food export UK" — and you begin sowing the right signals, you will become the default answer for anyone who asks that question. The AI has no geographical prejudices. It is simply hungry for reliable data.

The Naples–London factor: a digital bridge yet to be built

Between my frequent working stays in London and my daily dealings with Italian clients, I have noticed something very clearly. The British market is far more accustomed to conversational search. People there no longer search for "marketing agency" but for "marketing agency for sustainable fashion in North West England". That semantic specificity is precisely the level of granularity that AI systems love and reward.

This culture of precision is arriving in Italy. Neapolitan SMEs that have always prided themselves on bespoke quality have an extraordinary competitive advantage here — because craftsmanship is specificity. And AI loves specificity.

If you are a tailor from Arzano, do not try to compete on the keyword "men's suits". You will never win against the giants of fast fashion or the major global brands. But if you optimise your presence for "Neapolitan bespoke wedding suit with hand-stitched shirt shoulder", you become the global point of reference for that niche. The AI will be your best salesperson, because it will bring to you only those searching for exactly that standard of perfection.

Five operational pillars for existing in AI responses

Let us move to practice. I have no intention of giving you a generic checklist from a digital self-help handbook. These are five actions designed specifically for those operating between Italy and the British market — Italian businesses that want to be found in London, or agencies like ours that need to be visible on both fronts. Tested on ourselves before ever being tested on our clients.

1. Take ownership of the data infrastructure

I am not talking about the old Yellow Pages stuffed with intrusive adverts. I am talking about sources that AI reads as "gospel truth":

Google Business Profile: it must be impeccable. Original photos of the office or studio (never stock!), precise opening hours, weekly updates with regular posts. LLMs use Google Maps data as a primary source for the geolocation of services. An abandoned profile is worse than an absent one.

Schema Markup (structured data): this is the secret language of machines. If your website does not explicitly declare via code (JSON-LD) that you are an "Organisation" based in "London" (or "Naples", depending on your situation) and specialising in "communications and advertising for international markets", you are asking the AI to guess. When AIs have to guess, they prefer not to take the risk — and ignore you. This applies to us just as much as to our clients.

Category directories: essential. If you are an agency you need to be on Clutch and UpCity. If you produce wine you need to be on Wine-Searcher and the consortium portals. If you offer export consultancy you need to be listed in the directories of the Italian-British Chambers of Commerce. These are the knowledge bases that AI consults to validate your existence.

2. Become "citable": digital PR 2.0

In GEO, a textual mention on an authoritative website is worth as much as — and sometimes more than — a link. The old traditional link-building strategy gives way to the pursuit of quality mentions.

A practical example: an interview in Il Mattino telling the story of how your Caserta business reinvented mozzarella packaging using biodegradable materials is worth its weight in gold. The AI registers the association: [Your Brand] + [Sustainable Packaging] + [Caserta]. The next time someone asks about sustainability in Campanian food packaging, your name will surface from the background noise.

A mention on a British industry blog dedicated to Italian food & beverage is worth just as much for the UK market. The digital bridge between Naples and London is not built with a Google Translate rendering: it is built through a genuine presence in the written conversations that matter on both markets.

3. LinkedIn as your professional "source of truth"

Forget the motivational posts. ChatGPT and Perplexity read LinkedIn as though it were the world's official CV.

An optimised LinkedIn profile — with an "About" section that includes precise geographical and technical terms, a list of verified skills, and a series of native articles published on the platform — is the fastest way to help machines understand who you are and what you do.

Do not treat LinkedIn as a social media platform to simply "be on". Treat it as the verification infrastructure for your expertise. If the AI finds your name, the first thing it does is look for your LinkedIn profile to validate your claims. If the profile is empty, or worse still last updated to your previous job in 2019, you are actively damaging your algorithmic credibility.

4. Answer the "exact question" through your content

Here we return to the link with our article on the authentic use of AI (How Not to Become a Digital Clone). Saying "we are good at what we do" is not enough. You need to create content that answers the complex questions your potential clients are posing to AI at this very moment.

A concrete scenario: you are a producer of handcrafted footwear in Aversa and you want to sell in London. Do not write merely "handmade shoes". Write a guide on "how to identify genuine Campanian vegetable-tanned leather compared to industrial imitations". Why? Because when a British buyer asks ChatGPT "how to verify the quality of Italian leather shoes", the AI will use your parameters to compose the response. You will become the benchmark for quality, not just another producer among many.

5. Monitor your "digital shadow" with a monthly AI Audit

Become your own detectives. Run this test on the first Monday of every month — we call it the Monday Audit:

Open ChatGPT, Perplexity, and Google AI Overviews. Then ask these three questions — in Italian and in English, if you operate across both markets:

• "Who are the best [your sector] in [your city — Naples, Caserta, London]?"
• "What is the reputation of the company [your brand] according to online sources?"
• "Recommend a [your service] agency with Italian expertise in London." (or the equivalent Italian variant)

If you do not appear, or if the data is three years out of date, you have a GEO gap. If a competitor appears, look at the sources the AI cites — Perplexity is excellent at this — and ask yourself: which blog did they write for? Which directory are they listed in? Which article mentioned them? Then, simply, act.

Anatomy of a response: why the AI chooses your competitor

Let us conduct a post-mortem analysis of an AI response to understand what the machine sees that we do not.

Imagine a user asks: "Can you recommend a photographer for corporate events in Naples who knows the historic venues well?"

The AI responds: "For events in historic venues in Naples, one of the most frequently cited names is [Competitor X], known for their work at Villa Pignatelli and the Museo di Capodimonte..."

Why did it choose them and not you?

Semantic specificity: Competitor X's website has a dedicated page entitled "Corporate Photography in Historic Naples Venues". It does not say generically "event photographer". It uses the exact words the user used.

External mentions: Competitor X was cited in an event planner's blog that listed the "10 best venues for events in Naples". The AI connected the photographer's name to the category "historic venues" via a third-party source.

Structured data: the website has the "ProfessionalService" Schema correctly filled in, stating Naples as the operational area and "event photography" as the service.

If you are a highly talented photographer but your website simply says "Mario Esposito — Photographer", the AI does not have enough statistical anchors to "risk" recommending you. AIs hate giving incorrect information. They prefer to cite those who have multiplied their confirmation signals, not those with the most beautiful portfolio.

The authenticity paradox: do not write for robots

There is a dangerous temptation in this new world: to start producing flat, keyword-saturated texts that look as though they came out of an automated generator from the early 2000s. It is the gravest mistake you can make, for a very precise reason.

Today's AIs are trained to recognise and reward EEAT (Experience, Expertise, Authoritativeness, Trustworthiness). In plain terms: they look for the human voice, the detail that only someone who has lived an experience can know. An article with invented data, generic examples, and an aseptic tone is perceived as "low-reliability content" — and relegated to the bottom of the hierarchy of citable sources.

If you write "That time I lost 40% of my traffic due to a mistake in the robots.txt and how I fixed it in 72 hours in Manchester", you are giving the AI something unique: a real experience, specific data, a recognisable voice. This is the paradox of GEO: to make yourself understood by machines, you must write in a profoundly human way. We discussed this in detail here: How Not to Become a Digital Clone.

What NOT to do: the three cardinal sins of the new visibility

At Insight ADV we are already seeing questionable advice doing the rounds. Before embarking on costly strategies, avoid these three mistakes:

The robots.txt suicide. Some "experts" recommend blocking AI bots (such as OpenAI's GPTBot) to protect copyright over content. Unless you have paywalled content like the New York Times, this is a self-defeating strategy. If the AI cannot read your website, for all intents and purposes you do not exist. It is like bricking up the shop window and then wondering why nobody comes in to buy anything. (For the technical deep-dive on robots.txt and LLMs.txt, see: The Answer Era.)

Synthetic over-optimisation. Filling your website with AI-generated texts without human oversight, purely to "bulk up" content and cover every keyword. AIs recognise their own kind better than we might think. A site packed with automated content lacking a human voice is classified as a low-reliability source. Three excellent, personal articles are worth far more than a hundred pages written by a soulless bot with no experience to draw from.

Ignoring the UK market if you export. Many Campanian SMEs think it is enough to translate their website. This is a serious mistake. The AI responding to a buyer in London or Manchester draws predominantly from English-language sources and British directories. If you want to sell in the UK, you need to exist in the digital conversation in the UK. GEO is, by definition, localised.

Let us reclaim our space

Back to that café in Marylebone. A few months after that first failed test, I ran the same search again. This time, having worked on our digital shadow, having attended to the signals we were sending out into the web, the result was different.

We are not yet the first result in every possible variant — and anyone who claims they can guarantee you that is lying through their teeth. But this time Insight ADV was there. There was a reference to our London office, a citation of one of our articles, a connection to our international work.

GEO is not a magic wand. It is a labour of patient sowing, consistency, and above all, truth. Artificial intelligence is not trying to understand how good you are: it is trying to understand how reliable you are as a source of information for its users.

In Campania we have a millennia-long history of commerce, beauty, and innovation that we often struggle to tell well beyond our own borders. It would be an unforgivable shame to remain absent from the maps of the future simply because we lacked the inclination to speak the language of the machines that are redrawing those maps right now.

The future does not belong to those who shout loudest on Google. It belongs to those who manage to become the most credible answer in a silent conversation between a human being and a machine.

Is your brand ready to step out of the shadows?

If you want to understand where you stand today in AI responses and build your Generative Engine Optimisation strategy, we are here to offer a consultancy that combines the soul of Campania with an international vision.