How AI is changing SEO: what is new and what stays

AI is changing SEO on two fronts. Searchers more often get a summarised answer instead of a list of links, and the work itself goes faster, from research to putting changes live. The foundation stays: a findable, fast site with content that genuinely answers the question wins in Google and in AI answers alike.

That is why, as an SEO agency, we combine classic SEO and GEO in one programme. GEO stands for generative engine optimization: making sure AI systems such as ChatGPT, Gemini and Google AI Overviews name your brand.

Below you will read what is shifting, what is not, and what you can do with it today.

What changes for people who search?

Searchers ask longer questions and get an answer with sources attached. Google writes that people in AI Overviews and AI Mode ask longer and more specific questions, plus follow-up questions to dig deeper ^[1].

Behind that one answer there are often several searches. AI Overviews and AI Mode can use a technique Google calls "query fan-out": related searches across subtopics and sources ^[2]. So your page can become a source for a question it does not literally contain.

ChatGPT, Gemini and Perplexity work with their own search function and their own crawlers. How each of them chooses sources is covered in how do ChatGPT, Gemini and Perplexity choose their sources? The result is the same everywhere: the goal shifts from ranking high in the list to being named in the answer.

What stays the same?

For Google, there is no separate AI SEO. No extra requirements or special optimisations are needed to appear in AI Overviews or AI Mode, and the usual SEO best practices still apply ^[2]. You don't need new machine-readable files, AI text files or extra markup for it ^[2].

What counts is what always counted. Google mentions, among other things ^[2]:

• Access: crawling is allowed in your robots.txt and by your hosting or CDN.
• Internal links, so your content is easy to find.
• A good page experience for whoever lands on your site.
• Important content as text, not only in an image or a PDF.
• Structured data that matches the visible text on the page.

On top of that, Google asks for unique content that offers more than what is already everywhere ^[1]. If you lack this foundation, you won't win AI visibility either. So every programme starts with technology and content, not with an AI trick.

What does GEO require on top of SEO?

GEO requires pages that are easy to quote, and a brand that is also mentioned outside your own site. The researchers who introduced the term GEO measured that changes to a source text could increase visibility in generated answers by up to 40%, with large differences per domain ^[3].

In practice, that means:

• one question per page, with the answer in the first sentences;
• figures, prices and deadlines with a source and a date;
• the same company details on your site, your Google Business Profile and your other profiles;
• mentions in places where your customers already read.

That is why your website becomes more important, not less. It is the place where you decide which facts AI systems read about your company. The full step-by-step plan is in how to get found in ChatGPT.

How is AI changing SEO work itself?

AI speeds up every part of the work, but the choices and the checks stay with people. That applies to research, to content and above all to implementation.

Research and structure. Google itself writes that generative AI can be useful for researching a topic and for giving structure to original content ^[4]. Clustering keywords, reading competitors and drawing up a briefing per page go faster as a result.

Content. AI text is allowed, but generating many pages without added value for the reader can fall under Google's spam policy for "scaled content abuse" ^{[4], [5]}. Google also asks for accuracy in titles, meta descriptions, structured data and alt text ^[4]. So have every text reviewed by someone who knows the field.

Implementation. This is where the shift is biggest. MCP, the Model Context Protocol, is an open standard that lets AI applications such as Claude connect to external systems ^[6]. WordPress gained an Abilities API in version 6.9 ^[7], and the MCP Adapter turns those abilities into tools an AI assistant can call ^[8].

For our clients with a monthly SEO programme, where another party manages the website, the site always runs on a CMS. In about nine out of ten cases, that is WordPress.

On those sites we now put our recommendations live via MCP, much faster than by hand in WordPress. Which route fits which WordPress version, and how to set it up safely, is covered in SEO via MCP in WordPress.

Still, MCP in a CMS is an intermediate step. Every change has to fit within what the theme and the plugins allow, and for an SME site the CMS today mainly gets in the way. A site that AI builds in HTML and CSS without a CMS does not have that limit.

How do you measure SEO when AI gives the answer?

Measure on enquiries and revenue, not only on rankings and clicks. Google says clicks from results with an AI Overview are of higher quality: visitors more often stay longer on the site ^[2]. It recommends also looking at conversions, such as sales, sign-ups or lookups of your business details ^[1].

Three things to know:

• Search Console counts visits from AI Overviews and AI Mode in the regular search traffic, in the performance report under the search type "Web" ^[2].
• Traffic from AI assistants such as ChatGPT is tracked in Google Analytics. How to keep it neatly apart is explained by my colleague Cédric in UTM codes explained.
• Mentions in AI answers are measured with a fixed list of questions, every month again, per engine.

Want to know where your site stands today, in Google and in AI answers? The free SEO scan looks at technology, indexing, content, internal links and your visibility in AI answers. You get the findings and a conversation, whatever you decide afterwards.