From SEO to GEO: Making Sense of Search in the Age of AI

In our latest podcast episode, I sat down with marketing veteran Pietro Mingotti to talk about a buzzword that’s been popping up everywhere lately: GEO – Generative Engine Optimization .

Everyone seems to be asking: is this the end of SEO as we know it? Or just another hype cycle?

This post unpacks the key ideas from our conversation, cuts through the noise, and highlights what businesses really need to know about search in the age of AI.

From SEO to GEO: What’s Really Changing? Link to this headline

Traditional SEO Funnel                 GEO Funnel
──────────────────────────             ──────────────────────────
User                                   User
  ↓                                       ↓
Google Search                          ChatGPT / Claude / Perplexity
  ↓                                       ↓
Ranked SERP                            Generated Answer
  ↓                                       ↓
Click                                  Citation (maybe)
  ↓                                       ↓
Website                                Website (if clicked)

For decades, SEO has been about one thing: ranking on Google. But user behavior is shifting fast.

Instead of typing a query into Google, people now open ChatGPT, Claude, or Perplexity to get answers directly. That changes the playing field: content isn’t just competing for clicks on a search results page anymore—it’s competing to be cited in a generative model’s answer.

That’s where GEO comes in. Not a brand-new industry, but a new layer on top of the old game.

Can You “Rank” on ChatGPT? (Spoiler: No) Link to this headline

Here’s the catch: you can’t actually rank on ChatGPT.

Large Language Models (LLMs ) don’t work with indexes like Google does. They generate text probabilistically. When search is involved, it comes through RAG (retrieval augmented generation), where the model fetches facts via an external search engine.

So the goal isn’t ranking—it’s increasing your probability of being cited.

Pietro explains this in detail in his blog post:
👉 Why you can’t rank on ChatGPT (and other LLMs) .

If you want to understand how LLMs actually work under the hood, that research is a must-read.

Who Should Care About GEO? Link to this headline

Not every website will feel the impact in the same way.

• E-commerce and B2B lead-gen sites: conversions still happen on your website. LLMs may influence discovery, but they won’t replace transactions.
• Media, news, and tools: informational queries are at risk. If users get the answer directly in an AI summary, they won’t click through to your page. That’s where ad-driven business models take a hit.

For brands, this means GEO isn’t optional. It’s about positioning yourself so AI systems (and the people using them) can find and trust you.

Making Your Content LLM-Friendly Link to this headline

Bad Example (Hard to Extract)           Good Example (Easy to Extract)
───────────────────────────             ─────────────────────────────
- Long unbroken paragraphs              - TL;DR summary at top
- Missing headings                      - Clear H1 > H2 > H3 hierarchy
- Salesy marketing fluff                - Bullet points for facts
- Info buried deep in text              - Schema markup / quick data

The good news? Much of what helps with GEO looks a lot like good SEO. A few key things to focus on:

• Technical hygiene: make sure crawlers can access your site. Tools like llmstxt.site let you explicitly manage AI crawler permissions.
• Visibility: target long-tail queries and provide clear answers to specific questions.
• Content structure: use TL;DR blocks, bullet points, and concise summaries that can be easily extracted.
• Formatting: clean HTML, semantic headings, schema markup—make your content quick to parse, for humans and machines.

Think “extractable and citable” rather than just “rankable.”

The Limits of GEO: Measurement & Uncertainty Link to this headline

Traditional SEO was measurable: you could test, optimize, and see ranking changes. GEO is murkier.

LLMs are probabilistic systems, not rule-based algorithms. You can influence your chances of being cited, but you can’t guarantee it—or easily measure success.

There are tools emerging to track brand mentions inside LLMs, but they’re expensive and not yet reliable. For now, businesses need to get comfortable with experimentation and uncertainty.

Beyond the Hype: Staying Human Link to this headline

Here’s where Pietro’s advice lands hardest: in a sea of AI-generated “slob,” original human-created content stands out more than ever.

• Don’t outsource your thinking.
• Keep content authentic, specific, and reflective of your brand.
• Remember: marketing is still about connecting with people, not machines.

Or in Pietro’s words: “Now is the moment to be human.”

Takeaways Link to this headline

• GEO isn’t replacing SEO—it builds on it.
• You can’t “rank” on ChatGPT, but you can make your content more likely to be cited.
• Technical hygiene and clear, structured content are non-negotiable.
• Conversions still happen on your website—don’t panic.
• The biggest opportunity is to stay original and human in a world flooded with AI output.

🎧 Listen to the full conversation with Pietro Mingotti on our podcast for the deep dive .

And if you’re ready to make your website future-ready—whether that’s with SEO, GEO, or beyond—we’d love to talk .