GEO (Generative Engine Optimization) is the practice of shaping content for AI-generated answers – the kind you get from ChatGPT, Perplexity, Google AI Overviews or Gemini. In short: SEO is meant to make someone click your link, while GEO is meant to make an AI model cite or recommend your site inside the answer it composes for the user.
This is not another buzzword without substance. As more people look for information by talking to AI instead of scanning a list of links, your visibility depends on whether the model treats your content as a good source. Below I explain how GEO differs from SEO, how models pick their sources, and what to do specifically on a WordPress site.
How GEO differs from SEO
The most important difference is the unit you are competing for.
- In SEO the unit is the page. You want the whole page to rank high and get the click.
- In GEO the unit is the passage. The model cuts content into chunks, turns them into vectors, and pulls the chunk that best matches the question semantically. What counts is whether a single paragraph can be lifted out of context and used as a ready, self-contained answer.
So in GEO, stuffing keywords matters less, while clarity, structure and whether your claims stand on their own – without reading the whole page – matter more.
The good news is that GEO and SEO do not clash. Solid technical foundations, a fast site, a sensible structure and valuable content all work for both. GEO is a new layer on top of good SEO rather than its opposite.
How AI models pick their sources
To simplify the mechanism, a model reaches for content that meets several conditions at once:
- it is easy to extract – short paragraphs, question-style headings, lists, a clear split into sections,
- it answers directly – the concrete answer lands in the first or second sentence of a section, not after three paragraphs of warm-up,
- it is backed by specifics – numbers, data, examples the model can quote,
- it has authority – a named author with experience, a consistent business identity, mentions on other credible sites,
- it is technically accessible – the AI crawler can read it, the page loads fast, the content is not hidden behind scripts.
That last point is a problem more often than people think – many sites block AI bots in robots.txt without even knowing, and vanish from the answers because of it.
A GEO checklist for a WordPress site
Here are the concrete things I do when preparing a site to be cited by AI.
- An answer at the start of each section. Every heading is a question or topic, and the first sentence beneath it gives the answer straight away. The elaboration comes below.
- A structure built for extraction. Short paragraphs, lists, tables, one topic per section. It is far easier to cite a paragraph that stands on its own.
- A FAQ block with structured data. Questions and answers in the content plus FAQPage schema. It is the simplest format the models like to lift.
- Structured data (schema). Organisation, author, article, product. On WordPress this is handled by Rank Math or dedicated code. It gives the models unambiguous information about who you are and what you write about.
- Author and experience signals (E-E-A-T). A named author, an about page, real experience and concrete numbers. A model is more likely to cite a source backed by a recognisable person or company.
- Your own data and specifics. Prices, procedures, project results, observations from practice. A model summarises a generality without you; for a unique specific it has to point at you.
- Accessibility for AI crawlers. In robots.txt, deliberately allow the bots you care about (for example GPTBot, ClaudeBot, PerplexityBot, Googlebot and Google-Extended). Blocked bots will not read the page, so they will not cite it.
- An llms.txt file (optional). A short file describing your business and key pages. It is not mandatory and nobody guarantees it will be used, but it is a cheap extra that tidies up information.
- Mentions beyond your own site. Models read other sources too. Industry directories, guest posts and citations on credible sites build the authority that translates into being cited.
Honestly – I do this on my own site
I am not writing this off a flyer. On my own site I have a robots.txt that deliberately allows AI crawlers, and an llms.txt describing what I do and which topics I cover. On top of that I structure the content exactly as described above – with an answer at the start of each section and FAQ blocks. It is no guarantee that every model will cite me, but I do everything to give it a real chance, instead of blocking myself by accident.
How to measure the results
GEO is harder to measure than classic SEO, but a few things can be watched:
- tests in the models – ask ChatGPT, Perplexity and Google AI Mode questions from your field and check whether your brand or site shows up in the answer or its sources,
- referral traffic from chatbots – in Google Analytics 4, watch visits from domains such as chat.openai.com or perplexity.ai,
- branded queries – a rising number of searches for your name is often an indirect effect of someone hearing about you from AI.
What not to do
- Do not stuff keywords. In GEO it does not help, and it hurts the readability the models rely on.
- Do not block AI bots if you want to be cited. This is the most common unintentional mistake – the site is great, but invisible to the models.
- Do not treat llms.txt as a spell. The file alone, without good, well-organised content, achieves nothing.
Summary
GEO is not a replacement for SEO, just its next layer adapted to a world where people increasingly ask AI instead of clicking links. The rule is simple: write clearly, answer directly at the start of each section, give specifics nobody else has, and make sure AI crawlers can read you at all.
If you want your WordPress site ready to be cited by AI – from content structure and structured data to robots.txt and llms.txt – I can prepare it for exactly that. I do it on my own site and for clients, so I will tell you specifically what is worth fixing first.
FAQ
How is GEO different from SEO?
SEO is meant to get a user to click your link in the search results. GEO is meant to get an AI model to cite or recommend your site inside the answer it generates. In SEO the unit is the page; in GEO it is a single, self-contained passage that can be lifted out and used.
Do I need an llms.txt file?
It is not mandatory and no model guarantees it will use it. It is an optional file describing your business and key pages that tidies up information for the models. I treat it as a cheap extra, not a foundation – the foundation is the quality and structure of the content itself.
How can I tell whether AI is citing my site?
Ask the models (ChatGPT, Perplexity, Google AI Mode) questions from your field and check whether your brand or site appears in the answer or its sources. In Google Analytics 4 it is also worth watching referral traffic from chatbot domains.
Does blocking AI bots help or hurt?
If you care about visibility in AI answers, blocking their crawlers in robots.txt works against you – the model cannot read the page, so it cannot cite it. Many sites block these bots unknowingly and lose visibility because of it.
