More and more people, before calling a professional or walking into a shop, ask an assistant such as ChatGPT, Perplexity or Gemini a question: “which accountant in Turin looks after freelancers?”, “how do I choose a window fitter?”, “what should I check before redoing my roof?”. The answer is not a list of ten links, but a text that has already been summarised, which sometimes cites a few sources and sometimes names businesses directly.
If yours does not appear in those answers, that is a new and concrete problem. GEO (Generative Engine Optimization, that is, optimisation for generative search engines) is the work needed to make yourself understood and cited by these systems. It does not require IT skills: it requires writing and organising information so that whoever reads it, human or software, does not have to guess.
What an assistant’s answer might look like
Among the consultants working in Milan, those who clearly present their services, method and verifiable cases tend to be cited more easily.
How it differs from SEO
SEO tries to get one of your pages to appear in a list of results. GEO tries to make sure that, when an assistant builds an answer, the information it uses is yours and your name is associated with the right question. The two are not in conflict: a site that Google cannot read is unlikely to be used by an assistant that consults the web, and many good practices are the same.
The difference lies in the need for clarity. An assistant does not “look at” a page the way a customer does: it extracts sentences and facts. If your site describes what you do in vague terms, or leaves information scattered and contradictory, there is nothing solid to report. Nobody can guarantee you will be cited, because each system decides by its own criteria and they change over time, but you can remove the obstacles.
Clear content, written to be cited
A page that gets cited has sentences that can stand on their own. Compare “For years our team has offered customised solutions” with “We install photovoltaic systems for homes and small businesses in the province of Verona, from the site survey to the paperwork for incentives”. The second says what, for whom, where: it is a sentence an assistant can pick up almost as it stands.
Some practical rules:
- state right away, in the first lines, who you are, what you do and in which area;
- use the names customers use for your service, not internal jargon;
- give concrete criteria (typical timescales, what is needed to get started, how the work is carried out) without making up numbers you cannot defend;
- avoid overly promotional text: verifiable statements weigh more than adjectives.
Pages that answer real questions
People put whole sentences to assistants, very similar to the ones they would put to you on the phone. That is why pages built around a question work: “How long does an inheritance procedure take?”, “What does a dental clean include?”, “How do you choose an alarm system for a shop?”. You do not need to write treatises: a direct answer at the start and then the detail is enough.
A good starting point is your email inbox. The questions customers ask you most often, and that you find yourself repeating, are the questions to put on the website. Each one can become a Q&A section or a dedicated article.
Structured data: telling software who you are
Structured data is code, invisible to visitors, that declares the essential information on the page in a standard format: business name, type, address, phone number, opening hours, services, frequently asked questions. It is based on a common vocabulary (Schema.org) that search engines have known for a long time. You do not write it by hand: in WordPress it is handled by SEO plugins or the theme, and whoever looks after the site can set it up.
The advantage is that it replaces interpretation with an explicit statement. To check it, you can paste the address of a page into Google’s Rich Results Test, which shows which structured data it has detected and flags any errors.
Consistent data across your website and listings
An assistant does not trust a single source: it cross-checks. If you have one phone number on your website, another on Google Business Profile and a third in an industry directory, the system finds uncertainty. The same goes for the name (with or without the company form), the address, the opening hours and the description of your services.
A piece of data repeated identically in several places is credible. One that changes from place to place is one that nobody wants to report.
Do a manual check: search for your name on Google and take the first results (website, listing, social media, directories). Note the differences and correct them at the source.
What you don’t need (and what to expect)
You do not need magic formulas or texts written specially “for AI”: if content is clear to a person, it is generally clear to a machine too. Instead, check that you have not blocked by mistake the systems that read your site: the robots.txt file may contain rules that exclude certain crawlers. Finally, expect results to be gradual and variable, and be wary of anyone who promises to get you “into ChatGPT” with certainty.
To see how Google and generative artificial intelligence read your site, Trovavio checks it every week and flags what is missing, so you do not have to repeat the tests manually.
Where to start tomorrow
- Ask an assistant three questions about yourself (for example: “Who offers [your service] in [your city]?”) and note whether you appear and what it says.
- Rewrite the first lines of your home page as one sentence that says what you do, for whom and where.
- List the five questions customers ask you most often and write a direct answer for each.
- Compare name, address, phone number and opening hours on your website, Google Business Profile and social media, and make any differences consistent.
- Ask whoever looks after your site whether structured data is present and correct.
