Skip to main content
Aelo Brand Operating System logo
Log in
Try it for free
Knowledge

What is GEO?

GEO, or Generative Engine Optimisation, involves ensuring that AI chatbots such as ChatGPT, Gemini and Perplexity mention and recommend your brand when someone asks about your category without specifying the company name. The result of GEO is brand visibility in AI, measured references, page citations and recommendations, rather than its position in the search results, as in SEO. In a study conducted by Princeton and IIT Delhi Including quotations and figures in the text increased its presence in the responses generated by generative models by around 30–40%. GEO offers no guarantees: no one controls the model’s responses; one can only increase the likelihood and measure the change.
Author: Karolina Wcisło · Last updated: 24 October 2026 · reading time: approximately 6 minutes
Definition

Where did the term ‘GEO’ come from?

The concept of GEO was introduced in 2023 by researchers from Princeton University and IIT Delhi in their paper „GEO: Generative Engine Optimisation”, at a time when an increasing number of queries began to be answered with a ready-made response from the model rather than a list of links. For twenty years, the question was: how to rank highly in search results. The answer was SEO. For the past few years, a second scenario has been on the rise: a person asks the model a question and receives a ready-made answer containing the names of several companies. They don’t click on any links, because they don’t need to.
GEO describes the work of getting a brand into exactly these answers. The name can be misleading: it is not about optimising the model itself, but about what the model finds about the brand. The same work is also called GEO optimisation, GEO positioning or GEO marketing.
In practice, it encompasses three things at once: content on your own domain, brand signals outside it, and the technical ability to link to your website. The third element is sometimes described separately as AEO.
GEO in a nutshell

We’re working to ensure that the AI model has the information it needs to reinforce your brand when someone asks about your category.

DiagnosisThe model knows that your company exists and what it does
RecommendationThe model mentions your brand when someone asks about the category
Citation rateThe model can extract an answer from your page

Full name: Generative Engine Optimisation. The term came into common use around 2023.

Mechanism

How does the AI chatbot match brands to responses?

He chooses those brands that he can back up with knowledge gained from his training and from sources found online. There is no list of recommended companies. He constructs his answer from whatever he finds, so a brand with no online presence has no way of being found online.
User's question

Who does it? [service] in [city] and who can you trust?

Model response
  1. 1Company Ahas reviews
  2. 2Company Bin the ranking
  3. 3Company C

The model specifies selection criteria. These, too, come from somewhere.

01Asks about the category No brand name, as requested by the customer.
02Looking for the basics In training and in live feeds: websites, rankings, reviews.
03Selects names Those which he can back up with evidence.
04Adds criteria It offers advice on what to look out for when making a choice.
The second step is the most important. The model needs confirmation, and the easiest way to find it is outside your domain: in reviews, rankings and industry publications. Your own website says you exist. Other sources say you’re worth mentioning.
Three areas

What influences brand visibility in AI?

Three things: the content on your website, brand signals from outside it, and the technical ability to link to the site. Only the first of these is entirely up to you.
Content on your own domainYou do it yourselfSubpages under topics people ask about, with the answer in the first paragraph and specific details rather than buzzwords. This builds brand recognition: the model knows that your company exists and what it does.
Signals from outside the domainIt depends on othersReviews, rankings, industry publications, catalogues, presentations. These are what build a recommendation – that is, a situation where the model mentions your brand, rather than simply being aware of it. This is the hardest and longest part.
Permission to quoteTechnologyWebsite structure, company details, access for bots. Without this, even good content can be unreadable to the model.
The most common mistake is to focus solely on the first area. A company writes articles for half a year, its brand awareness grows, yet it still receives no recommendations. We discuss this in the article known, but not recommended.
Practical experience

Where should I start with GEO?

Starting with the baseline measurement: ask the AI chatbots questions about their category without mentioning any brand names, and make a note of which ones they mention. All four steps can be carried out independently, without tools and without an agency.
Step 1Measure the initial condition Ask the models questions from your own category, without specifying the brand name. Make a note of the names they mention.
Step 2Find the barriers Find out what the models have no way of verifying: reviews, citations, inclusion in the lists.
Step 3Do two things, not ten Start with the ones that cover the largest proportion of search queries in your category.
Step 4Measure again After a month, on the same set of questions. Without a second measurement, there’s no way of knowing whether anything worked.
We explain how to ask models questions so that the results make sense in this article how to find out what AI says about a company. The way it does this Aelo in the module Visibility in AI, is described in methodology.
To be honest

What won’t GEO do?

GEO cannot guarantee recommendations or immediate results, as no one controls the model’s output. A great many promises have grown up around the concept, so we’re cutting four of them out.
It is not possible to „position” a brand within the modelThere is no single panel or parameter that boosts the brand’s rating in the responses. The only thing that can be done is to increase the number and quality of the signals on which the model is based.
No one checks the answersThe same query text may produce a slightly different result, and The models change with every update. That is why it is the trend that counts, not a single measurement.
A diagnosis is not yet a recommendationA model may be aware of a brand and still choose not to recommend it. These are two different jobs and two different terms.
You won’t see the results within a weekSignals from outside one’s own domain take months to build up, and models need time to pick them up.

What sources is this text based on?

FAQ

Questions about GEO

The questions we hear most often when people first come across this concept.
What is GEO?

GEO, which stands for Generative Engine Optimisation, is the optimisation of a brand’s presence in responses generated by AI models. It is about whether the model mentions the brand when someone asks about a category.

How does GEO differ from SEO?

SEO relates to a website’s position in search results and is measured from the moment a link is clicked. GEO refers to a situation where a person finds the answer they’re looking for and doesn’t click anywhere. We explain this in detail in the article SEO, GEO and AEO.

Does GEO replace SEO?

No. These are two parallel systems that measure different things. Search engine rankings are still important, and a website that ranks highly in search results usually has an easier start in these models.

Is it possible to check your GEO yourself?

Yes, on a small scale. You ask models questions from your own category without mentioning any brand names, and make a note of the brands they mention. We explain how to do this effectively in the article how to find out what AI says about a company.

How long does it take for the work on GEO to bear fruit?

From a few weeks to a few months. Models notice changes on their own website more quickly, whilst signals from outside it take months to build up. We discuss this in the article Why aren’t the changes visible straight away?.

Does GEO apply only to large companies?

No. The mechanism is the same regardless of size, and in narrow local terms It can be easier, as models have fewer brands to choose from.

What should I do if the model doesn’t know my company at all?

Start with the indicators that help to confirm this: reviews, listings in directories and mentions in publications outside your own domain. We describe this in the article when AI doesn’t know your business.

What is an AEO, and is it the same thing?

No. AEO is a narrower concept: preparing content in such a way that the model can extract a ready-made answer from it. It forms part of GEO as its editorial and technical component. Read more in the article What is an AEO?.

What does the GEO effect look like in figures?

Taking a lime manufacturer as an example: in June 2026, the brand did not appear in any of the 18 search queries; after five weeks of work on the website, it appeared in 17 out of 18. The first recommendation came after 11 weeks. We describe this in case study of a lime manufacturer.

Do you need to commission an agency to carry out a GEO?

No. You’ll carry out the measurements, make any corrections to the page and ensure a consistent company description yourself. We explain what GEO is like as an agency service in this post GEO services at an SEO agency.

Check whether the models mention your brand

Free Account No card required: you get the following to start with 250 AeloCoins and full visibility report Your brand on ChatGPT, Claude, Gemini and Perplexity. You’ll see who the AI chatbots mention in your category.
Try it for free
Related: What is an AEO?, What is SEO?, differences, dictionary, methodology.
Aelo
Aelo (aeloapp.io) measures brand visibility on ChatGPT, Claude, Gemini and Perplexity.

Product

Who is it for?

Knowledge

Company

© 2026 Aelo Prosta Spółka Akcyjna. All rights reserved.
Sąd Rejonowy w Bielsku-Białej, VIII Wydział Gospodarczy KRS, KRS 0001269539, NIP 9372784735

Check whether the AI recommends you or your competitors

A free report within 24 hours on working days