Green Consulting, home

Digital Marketing · GEO

Generative engine optimization to get you cited in AI answers

A share of travel searches no longer ends on a page of results. It ends inside an answer already written, which cites two or three sources and closes the matter. GEO, generative engine optimisation, is the work of becoming one of those sources. Nobody can promise you a place there, because nobody controls those systems. What you can do is earn it.

Request a free quoteRead the story

When the answer replaces the list

Until recently a search produced a list of links, and the choice stayed with the person searching. Today it increasingly produces a text already written, with two or three sources cited alongside. Whoever gets named in there joins the conversation. Whoever is missing isn’t second or third. For that question they don’t exist.
Four systems matter, namely ChatGPT, Perplexity and, on the Google side, the generative answers above the results (AI Overviews) and Gemini. None of them publishes its rules, but the observable behaviour all points one way. What gets used is information that is verifiable, repeated consistently elsewhere, and written so it can be lifted out without interpretation.
In tourism it lands earlier than elsewhere, because travel questions are the ones an assistant answers most willingly. Where to stay in an area, what to do when it rains, how long it takes from the airport. Open questions, with no official answer, where the advice gets built from whatever is out there.
The work splits into three parts. Structuring content so it can be quoted, building the sources these systems read, and measuring whether your name shows up in the answers to the questions that matter to you, next to your competitors.

The systems visibility is measured on

We name them because they are the systems visibility gets measured on, not because any agreement, partnership or certification exists. They are third-party products, and none gives guarantees to anyone optimising for them.

ChatGPT

The assistant that entered everyday use first, and the one where travel questions already arrive as conversation. When it searches the web it shows its sources, so it is the clearest place to see which page you get named through, or who gets named in your place.

Google’s generative answers (AI Overviews)

The answer block Google builds above its results. Nobody opts into it, because it sits inside the search engine people already use, often without their noticing, and it decides whether a search reaches your site or stops short of it.

Perplexity

Built as an answer engine, it keeps its sources visible and numbered. Smaller audience, but the most readable system for seeing which pages count as reliable on a subject.

Gemini

Google’s assistant, sitting inside tools plenty of people already open daily. It draws on Google’s index, so it is affected by how clear and consistent the information about you is.

And the other assistants

Claude and Copilot round out the picture, working the same way. Whoever is quotable for the first four tends to be quotable here too, so no separate workstream is needed for each.

What can be guaranteed and what nobody can promise

On the right is work someone can take responsibility for. On the left, systems that decide on their own and change without notice.

AspectWhat nobody can guaranteeWhat can be guaranteed
On contentThat a system will pick those pages, because the selection criteria aren’t public.Pages that answer one question in full, with explicit, verifiable facts.
On sourcesThat an outside source will agree to correct what it wrote about you.Consistent data, names and numbers on every page that mentions you, on your site and off it.
On measurementThat two identical queries return the same text, because answers vary and the data reads as a trend.A reading repeated over time on a question list agreed with you, competitors included.
On timingA date by which you appear, because that depends on when sources get re-read, and no supplier sets it.A schedule of work and a check at fixed dates.
On the outcomeA citation percentage or guaranteed presence, because whoever promises it cannot deliver it.An account of what changed, what didn’t move and what comes next.

The five stages of a GEO project

They apply just as much to a project someone else is already running for you. The order isn’t accidental, because the second stage and the fifth are there to measure, and without a recorded starting point you can’t say whether anything moved.

  1. 01

    The list of questions that matter

    You don’t work on a keyword, you work on whole questions, the ones a guest would really put to an assistant before booking, in the languages of your markets. They have to be written with someone who knows the property, because no generic list contains the question about coach parking or the beach you can walk to. Usually between thirty and a hundred.

  2. 02

    The baseline reading

    We put those questions to the systems one at a time, recording who gets cited, through which page, and how often. The picture that comes out is uncomfortable and useful, because the name that shows up is almost never yours, it’s a portal’s or a competitor’s.

  3. 03

    Content that can be quoted

    We write, or rewrite, the pages that have to answer on a single rule, namely one question per page, the answer in the opening lines, facts with numbers, dates and full names. Form counts before quantity, and a long text with no liftable sentence gets used by nobody. Your site is probably full of them.

  4. 04

    The sources these systems read

    An assistant doesn’t read only your site. It assembles what it finds repeated on pages that aren’t yours. If the number of rooms, the address or the name of the town disagree from one source to the next, that fact goes unused and whoever holds a single version gets named instead. The work here is an inventory of where you are mentioned and the alignment of details that don’t match. Building authority and links is a different craft, and it sits on the SEO page.

  5. 05

    Recurring measurement and correction

    The reading has to be repeated at fixed intervals and compared with the baseline and with competitors. A serious report tells you which questions you appear on, through which source, and what changed since last time. It won’t promise you a ranking, because in generative answers there isn’t one. From there you decide where to act next, three or four questions at a time.

What makes content quotable

The checks we run page by page. Not a technical novelty but writing discipline, and it’s why one page gets used and the one beside it doesn’t.

  • The answer sits in the first three lines, ahead of any preamble.
  • Every statement stands alone when read away from the page, with no “here at our place” and no “as mentioned above”.
  • Numbers are explicit, with the unit attached, whether that’s distances, travel times, capacities or starting prices.
  • Dates are written out in full, not left as a vague “this year”.
  • Proper names are spelled out and always identical, from the property to the town to the province, with no shifting abbreviations.
  • The text that matters is in the body of the page, not inside an image or a PDF to download.
  • The Italian and the English say the same thing, numbers included, because two versions that contradict each other are worth less than one.
  • Who signs the page and when it was updated are both stated. A stale fact is worth less than a fresh one.

Frequently asked questions

GEO stands for generative engine optimisation, the work of getting used as a source by the systems that answer with a text instead of a list of links, meaning ChatGPT, Google’s generative answers, Perplexity and Gemini. It comes down to three activities, writing content that can be cited, keeping the sources that mention you consistent, and measuring what those systems say about your field.

No, and be wary of anyone who does. Those systems choose sources by criteria they don’t publish and change without warning, so nobody can decide what they will write. We commit to the work on content and sources, a periodic reading of where you stand against your competitors, and honesty about what moved.

It starts from a list of sector questions agreed with you. They are put to the assistants at regular intervals, and for each answer we record who is cited, through which source, and how many times. That gives a frequency, comparable with competitors and with the previous reading. There is no public ranking to look up. This is the only honest way to put a number on the channel.

No. We hold no agreements, special access or certifications with the companies that build these assistants, and for GEO none exist. We name them because they are the systems a brand’s visibility gets measured on, the way you would name a search engine. Anyone promising paid placement inside an assistant’s answers is describing something that doesn’t exist.

We don’t give a date, because it depends on when those systems re-read their sources, which isn’t up to whoever writes them. The baseline reading is there straight away, the first rewritten pages go live within a few weeks, then each measurement is compared with the last. If nothing has moved after several cycles, we change approach.

No, it works alongside it. They are two crafts with different measures, because one looks at positions and clicks and the other at citations inside a generated text. The sources assistants read are often the same pages that have to rank, so the two share one editorial plan. Search rankings themselves are covered on our hotel and tourism SEO page.

Let’s talk about your project

Request a quote to bring your project to life, win more enquiries and keep your customers coming back.

Request a free quote
+39 0744 431492 info@greenconsulting.it