AEO, GEO, and the future of search: how to prepare for the change

Why the rules of classic SEO are changing and what you need to do about it

The world you knew

For 25 years, the game has been the same. You had a website, worked on keywords, built backlinks, and optimized for Google. If you did it right, you appeared in the famous ten blue links on the first page. That was the prize.

It worked. Until now.

The call for change

AI is a revolution affecting everything: talent, productivity, usability, and user behavior dynamics. And as expected, it is also transforming how we search for information and interact with the internet.

Google has integrated AI Overviews into its search results, those AI-generated answers that appear at the top before the traditional links. Search Generative Experience (SGE) is expanding globally. And tools like ChatGPT Search, Perplexity, Claude, or Gemini are rapidly gaining ground as real alternatives to traditional search engines.

The data confirms the shift. Ahrefs reported in 2025 that the integration of AI Overviews reduced the CTR of the first Google result by 58% (previously, that decline was 34.5%). ChatGPT, for its part, processes more than 2.5 billion prompts daily, of which approximately 65% are searches that would have previously gone to Google.

Users no longer want ten links. They want a direct answer.

The current problem

What is clear for now is that current models (ChatGPT, Claude, Gemini, Perplexity) are not yet profitable. The investment required to maintain and train these systems far exceeds subscription revenue. OpenAI is still losing money, as is Anthropic, and while Google and Microsoft have the financial muscle to hold out, the pressure to monetize is real and growing.

Token-based model costs (what APIs charge for processing text) are causing many companies to reduce or cancel their enterprise plans because expenses skyrocket as they scale their usage.

This whole situation is reminiscent of Facebook's early days. They were slow to implement ads despite market pressure, and when they finally did, they transformed our relationship with digital advertising. We were used to an internet full of pop-ups and aggressive banners, but Facebook changed the rules by integrating ads as native posts within the feed. The format was so organic for the time that it was often hard to distinguish between real content and paid advertising.

Something similar is happening with conversational AI. For fear of losing user trust, platforms are still looking for the best way to serve ads without breaking the experience. But they will get there. Without a doubt.

The new landscape: AEO and GEO

In the midst of this shift, two concepts are taking hold that you absolutely need to understand.

What is AEO?

Answer Engine Optimization (AEO) is the practice of optimizing content so that answer engines cite it when responding to user queries. It covers everything from classic Google featured snippets to responses from voice assistants (Alexa, Siri) and conversational AI engines.

What is GEO?

Generative Engine Optimization (GEO) focuses specifically on generative engines (ChatGPT Search, Perplexity, Google AI Overviews, Gemini). The goal is for your brand to be cited as a source in the responses these AIs generate.

Are AEO and GEO the same thing?

Practically yes, but with nuances. AEO came first and covers a broader spectrum (including voice and traditional snippets). GEO is more recent and specific to the post-ChatGPT era. In day-to-day practice, the techniques overlap by 80-90%, and most SEO professionals are treating them almost as synonyms.

Don't obsess over the difference. If you optimize well for one, you are optimizing for the other.

What's next: AEM and GEM

These optimizations don't have much to do with paid campaigns yet, but I firmly believe they will be the foundation for a new system. Concepts like AEM (Answer Engine Marketing) and GEM (Generative Engine Marketing)are already being discussed, where brands pay to appear in AI-generated answers, just as they pay for other types of ads. This practice is already being tested on platforms like Google with its AI Mode and AI Overviews, and ChatGPT for free conversations. Due to the European Union's strict laws, this progress will be much slower here than in other parts of the world.

It is only a matter of time. When these platforms truly need to monetize, the ads will arrive. And those who are already organically positioned will have an advantage over those who enter late.

What are the essential tools?

To survive and stand out in this new landscape, you need to understand some technical tools.

What is robots.txt?

Robots.txt is a text file located at the root of your domain (yourdomain.com/robots.txt) that tells search engines which parts of your site they can crawl and which they cannot. It has existed since 1994 and remains fundamental.

What is sitemap.xml?

Sitemap.xml is another file at the root (yourdomain.com/sitemap.xml) that acts as a map of your website. It lists all the pages you want search engines to crawl, their hierarchical relationship, update frequency, and priority. It is the official guide you provide to Google regarding your content.

What is llms.txt?

Llms.txt is the new file on the block. Proposed by Jeremy Howard (co-founder of Answer.AI) in September 2024, it is designed specifically for large language models (LLMs). Its goal is to provide AIs with a structured and curated summary of your website's most relevant content.

But keep the real context in mind: llms.txt is not an officially adopted universal standard. Its adoption varies by platform:

• Anthropic (Claude): officially respects it in its retrieval flows
• Perplexity: uses it to prioritize page selection
• OpenAI (ChatGPT): no official confirmation, but there are observable patterns
• Google and Gemini: no confirmation; they continue to prioritize robots.txt, sitemaps, and structured data

There is also debate within the industry. Some experts argue that Howard never proposed llms.txt as a GEO/SEO tool, but rather for technical use by code agents (Cursor, Claude Code). Its use as a ranking lever is an interpretation by the SEO community, not the original intent.

That said, implementing it is low-cost (1-4 hours) and has no known downsides. If platforms end up adopting it en masse, you will already be ready.

The basic structure proposed for an llms.txt file is:

# Nombre del sitio
> Descripción breve del sitio

## Sección principal
- [Título de página](url): Descripción breve del contenido

## Documentación
- [Otra página](url): Descripción

‍

It is plain markdown, easy to generate and maintain. The trick to getting it optimized without any hassle is to ask your favorite AI to create it for you.

What is Schema Markup?

Schema Markup (structured data) is code you add to your HTML so that search engines and AIs understand exactly what type of content is on each page. If you have an article, you tell them, "this is an article, with this author, this date, and this title." If you have a product, "this is a product, with this price and this availability."

Schema.org is the standard, and Google uses it actively. The better your data is structured, the better AIs understand you and the higher the probability of being cited.

The most useful types depending on context: Article (for blogs), Product (ecommerce), FAQ (frequently asked questions), HowTo (tutorials), LocalBusiness (local businesses).

What stays the same (and what changes)

Otherwise, much of the game remains the same.

Authority remains key. AIs prioritize reliable sources, just like Google. Quality backlinks, mentions in relevant media, and a consistent presence on recognized platforms—all of that still matters.

Keywords are evolving. You are no longer fighting for single words; you are fighting for complete phrases, natural questions, and entire paragraphs that answer specific intents. Conversational search implies longer natural language.

Backlinks and mentions still count. AIs analyze what others say about you, not just what you say about yourself. Your digital reputation is built the same way it always has been.

Success in other channels adds up. The more visible you are on social media, podcasts, industry directories, and media outlets, the more likely you are to be referenced by AIs when they train or query information.

How to stay ahead

It won't be an easy task, but here is the opportunity: many companies with years of SEO history will fall behind because they won't update. That opens up space for you if you act fast.

What is recommended today:

Take your Schema Markup seriously. Don't just generate them automatically with a plugin and forget about them. Check that they faithfully reflect the actual content of each page. Use Article, Product, FAQ, HowTo, or LocalBusiness, as appropriate. Structured data is the technical language that AIs understand best.

Implement llms.txt, even if adoption is partial. The cost of implementing it is low, and it positions you for when it becomes more widely adopted. Anthropic and Perplexity already respect it, so there is already a real, albeit limited, benefit.

Restructure your content to answer questions. AIs look for clear answers to specific questions. If your content is organized as direct answers (with question-style headings, concise answers at the beginning of the paragraph, and concrete examples), you are more likely to be cited.

Stay alert to changes. This sector is moving incredibly fast. What is a proposal today could be the standard in six months. Follow the main platforms, industry professionals, and official documentation.

Don't abandon classic SEO. Google is still the dominant engine, and AI Overviews coexist with traditional results. Working on AEO/GEO doesn't mean abandoning SEO; it means expanding it.

The new normal

We are in a moment of transition that we haven't seen since Google replaced the old search engines. The rules are being rewritten in real time, and those who understand the shift early will have an advantage for years to come.

The blue links aren't dead, but they are no longer the only goal. Now you are also fighting to be the source that the AI cites when someone asks a question. To be the reference that Perplexity mentions, the article that ChatGPT considers an authority, or the site that Google AI Overviews highlights in its generated response.

The game has changed. The good news is that the manual is still being written. And that means anyone who starts now with a clear strategy can compete.

‍