What Is GEO and AEO (and Why SEO Is No Longer Enough)
SEO optimizes for Google. GEO and AEO optimize for being cited in AI answers — ChatGPT, Perplexity, Google AI Overviews. How they differ and what to actually do.
2024 was a turning point in how people search for information. ChatGPT surpassed 200 million weekly active users. Perplexity claims its search engine answers 100 million queries a day. Google rolled out AI Overviews — summaries above search results that show an answer before you click any link.
Customers are increasingly asking AI instead of scrolling through result pages. That changes the rules.
SEO: optimizing for rankings
SEO (Search Engine Optimization) has existed since the nineties. The fundamentals have not changed: write content, get backlinks, configure technical metadata — and Google ranks you higher in results. The goal is a click. The metrics are positions, impressions, CTR.
SEO works. But it works for the situation where a user opens a search engine, gets a list of links, and clicks.
That situation is ceasing to be the default.
GEO and AEO: optimizing for citation
GEO — Generative Engine Optimization — is a set of practices for getting generative AI systems (ChatGPT, Claude, Gemini, Perplexity) to include you in their answers and ideally cite you as a source. The term comes from academic research out of Princeton in 2023, which showed that different content styles affect how often generative models draw from a given source.
AEO — Answer Engine Optimization — is an older term, originally associated with voice assistants and Google featured snippets. Today it overlaps with GEO: both aim to make your content the source of direct answers, not just a link.
In practice they are not very different. Both rest on the same principles: content must clearly answer specific questions, be structured so a machine can understand it, and be technically accessible to AI crawlers.
Why this matters now
I have a concrete example. Someone searches for "how to connect Business Central to KSeF". They type it into Perplexity. They get an answer — with the names of companies that do it. They click one. The entire acquisition flow happened without Google.
If our website is not there, we do not exist for that query. It does not matter how well we rank in Google.
This is especially strong for technically oriented B2B queries. Customers looking for software, an integrator, or a consultant tend to ask precise questions. Those precise questions are exactly what generative AI systems are built for.
What to actually do
1. llms.txt
A llms.txt file in the root of your website tells AI crawlers what your site is about, which pages are most important, and how AI should use your content. Think of it as the analogue of robots.txt — but for AI, not for search engines.
The format is plain Markdown with a description of the site, links to key pages, and optionally instructions. For example: this site has an llms.txt with a description of what we do, links to product pages and articles, and an instruction for AI on how to work with the content.
2. Structured data (JSON-LD)
Schemas from Schema.org tell machines what each piece of content means. For GEO/AEO the most important are:
- Article — marks a blog post as an article with an author and date
- FAQPage — explicitly marks questions and answers; Google AI Overviews and Perplexity draw from this directly
- Organization — company name, what it does, contact
JSON-LD is inserted into the <head> of the page as <script type="application/ld+json">. Validation can be checked with Google Rich Results Test.
3. Content that answers questions
Generative AI indexes content but answers questions. If you want to be cited, your content must explicitly answer questions — ideally in the same phrasing your customers use.
That means writing paragraphs like: "What is X? X is..." or "How does Y work? Y works by..." — not just using a keyword in a heading and hoping.
FAQ sections at the end of articles are a direct implementation of this principle. Each Q&A answers one real query. If I wrote it clearly enough, AI may cite it verbatim.
4. robots.txt: allow AI bots
A large part of the web historically has various crawlers blocked in robots.txt. If you want to appear in AI answers, you must explicitly or implicitly allow:
GPTBot(OpenAI / ChatGPT)PerplexityBot(Perplexity)ClaudeBot(Anthropic)Google-Extended(Google AI training and Overviews)Googlebot(the foundation for AI Overviews)
A blocked bot = content AI cannot see = a citation that never happens.
5. Technical accessibility
AI crawlers are less tolerant of slow or broken pages than search engines. Content must be in HTML on the first request — JavaScript-only rendered pages, where the DOM arrives after JS hydration, may not be processed correctly by an AI crawler.
Prerendering or SSR is an advantage here. If the server returns full HTML with content at zero milliseconds, the crawler reads it reliably.
Where we stand
On this site, GEO/AEO infrastructure is deployed. Specifically:
llms.txtwith a description of the site and a map of key pages- Article and FAQPage JSON-LD on every article
- Trilingual hreflang (CS/EN/PL) — for queries across languages
- Prerendering — the server returns HTML with content without waiting for JavaScript
- 28+ articles focused on specific technical questions (KSeF, Peppol, ViDA, JSON-LD, e-invoicing)
- robots.txt that does not block any relevant AI crawler
Competitive analysis in our niche (KSeF, e-invoicing, .NET integration) shows that most competitors have no llms.txt, no FAQPage schema, and use JavaScript-first rendering. That is an advantage that will show up in citations — if we have the right content.
We handle GEO/AEO implementation and website audits for AI visibility as part of web presence projects for clients. If you want to discuss it, get in touch.
Summary
SEO = being found in Google. GEO/AEO = being cited in AI answers. These are different goals with partial overlap (good content helps both), but with different implementation.
The reason to address this now: customers searching for B2B software or expert services are moving to AI search faster than the consumer market. If your customers use Perplexity or ChatGPT when making decisions, GEO/AEO signals are a real acquisition channel — not a future experiment.
FAQ
What is the difference between GEO, AEO, and SEO?
SEO (Search Engine Optimization) optimizes pages for ranking in Google and Bing. GEO (Generative Engine Optimization) and AEO (Answer Engine Optimization) optimize content so that generative AI systems — ChatGPT, Perplexity, Google AI Overviews — cite it as a source. The goal differs: SEO wants a click, GEO/AEO wants a citation in an AI answer.
What exactly do I need to do to get cited by AI?
The basic steps: add llms.txt with instructions for AI crawlers, implement Article and FAQPage JSON-LD structured data, write content that directly answers specific questions (not just keywords), and verify that robots.txt has not blocked AI bots (GPTBot, PerplexityBot, ClaudeBot, Googlebot).
Is GEO/AEO necessary, or is it just hype?
It depends on your customer. If your customers ask technical or expert questions in ChatGPT or Perplexity and make decisions based on AI answers, then GEO/AEO is a real acquisition channel — not just hype. For consumer goods or local services, classic SEO may still dominate.