Contents7
GEO, for Generative Engine Optimization, is the work that gets a website cited in AI answers: ChatGPT, Perplexity, Gemini, Google’s AI Overviews. It is sometimes presented as a separate trade. In practice, it shares most of its method with SEO. Google says so itself about its AI Overviews: there are no additional requirements and no special optimisations needed to appear in them, beyond SEO best practices. A marketing team therefore has every reason to run both in the same routine. What an agency should deliver in this area is covered in AI SEO agency: what it should really deliver in 2026.
This guide proposes that routine, with three examples. The citation mechanics are detailed in How to get cited by ChatGPT and AI engines.
In short
Start from real queries, read in Search Console, not from an imagined keyword list.
Pages already close to the goal come before pages to create.
A page that answers at the top, with sourced facts, serves Google and AI engines at the same time.
Measure the two channels separately: rankings and clicks on one side, citations on the other.
What is shared, what differs
What serves both
A direct answer at the top of the page
One page per question, with a title that names it
Dated, quantified and sourced facts
A fast site, readable without scripts
What is specific to GEO
Being mentioned elsewhere than on your own site (articles, comparisons, directories)
Questions asked to AI engines the way you would ask a colleague, longer than a Google query
A different measurement: you replay benchmark questions, there is no official citations console
Example 1: a page that is seen but rarely clicked
In Search Console, a pricing page appears in the results for the query “wordpress maintenance price”, on average in seventh position, with many impressions and very few clicks. The page is not badly ranked: its title gives no reason to click.
The action: a title that repeats the query and promises the answer, and a first paragraph that gives the price range and what makes it vary. That same paragraph is the one an AI engine can quote as is. Note the date of the change and compare four weeks before and after, on the same query. The method to prove a change worked is in AI visibility in Search Console: proving a change worked.
Example 2: a topic where the company appears nowhere
You ask three AI engines the question a prospect would ask: “which provider can automate invoicing for a small business in France?”. The answers cite comparisons and guides, never the site. Two levers:
- a page that deals with exactly this question, with an honest comparison table, prices and limits;
- mentions elsewhere: a case study published by a partner, an up-to-date listing in the sector’s directories.
You replay the same question every month, under the same conditions, to see whether the site enters the answers. The protocol is described in Measuring your visibility in AI engines: an honest protocol.
Example 3: an old article that is slipping
A guide published two years ago loses positions month after month. The figures it quotes are out of date, and competitors have published more recent ones. Rewriting it from scratch would throw away what it has earned; you refresh it instead: same address and structure, figures updated with their source and date, a section added on what has changed.
The weekly routine
- Read.
The queries that are rising, those that are slipping, the pages seen but rarely clicked.
- Choose.
Three actions at most, starting with the pages already close to the first page.
- Write.
The title, the answer at the top, the missing facts, with their source.
- Record.
The date and nature of each change, so you can measure afterwards.
- Measure.
Once a month, rankings and clicks on one side, AI citations on the other.
From the field
Our own SEO engine runs this way
On brainplusai.com, Search Console has been read every night since 25 August 2026 by a script that spots slipping pages, titles to fix and queries within reach. Agents then prepare the draft adjustments. Nothing is published without human review: a client learned this the hard way, when a model published content that matched nothing real and it had to be taken down.
Frequently asked questions
Do you need an llms.txt file to be cited by AI engines?
It is not a must. It can help some tools read a site, but it replaces neither clear pages nor sourced facts. We take stock in llms.txt: what it is really for. The time spent on it is worth less than the time spent on your key pages.
Can you know how many visits come from AI engines?
Partly. Visits that arrive from ChatGPT or Perplexity show up as referring sites in your analytics tool. Answers that cite your site without a click leave no trace at all: only benchmark questions replayed every month measure them.
How long before you see the effect of a change?
On Google, a page that is already indexed often changes behaviour within a few weeks, hence the comparison over four weeks before and after. In AI answers, the delay varies by engine and is not published: you measure, you do not promise.
What we can take on
- Ongoing work on both channels: SEO and GEO.
- Measurement, on Google and in AI answers, by location and language: rank tracking on Google and in AI answers.
- An assessment before you start: SEO audit.
Sources
- Google Search Central, AI features and your website (new tab)developers.google.com
Jim · founder of Brain Plus AI
SEO and AI visibility
Want the same thing running at your place?
We run a multilingual content engine in production across six languages, and we apply to the site you are reading the very fixes we sell, in the order we recommend them.
Also in seo and ai visibility
Practical guide7 min
How to get cited by ChatGPT and AI search engines
What actually gets a site cited by ChatGPT, Claude or Perplexity: let the bots read it, answer straight away, source every fact, mark up only what is shown. And how to measure it without fooling yourself.
Practical guide8 min
Measuring your visibility in AI: an honest protocol
No tool gives you a Search Console for assistants. Three readings repeated every month show whether a site is read, cited and visited by ChatGPT, Claude, Perplexity or AI Overviews, and what each measure does not tell you.
Understand6 min
llms.txt: what it is really for
The llms.txt file promises to help AI read your site. What the proposal says, what can be verified about real usage, what Google, OpenAI and Anthropic say about it, and why we publish one anyway.