The Counter-Narrative Is Making the Rounds
A post is circulating on LinkedIn that's rattling many GEO practitioners: Generative Engine Optimization doesn't work. AI systems simply use the classic search index via RAG. Llms.txt provides zero advantage. If you don't rank in regular search, you won't appear in AI answers either.
Is there something to that? Yes – and no. Here's a nuanced take.
What's True: Solid SEO Remains the Foundation
The core argument – that high-quality content and technically sound websites remain the baseline – is correct. Language models do draw on indexed web content, either via Retrieval-Augmented Generation (RAG) or through training on crawl data. Brands that are completely invisible in search will also struggle in AI responses.
And yes: Llms.txt is not a magic bullet. It's a convention, not a guarantee.
> The foundation remains: great content, technical soundness, genuine expertise. That hasn't changed.
What's Missing: AI Visibility Is More Than SEO Ranking
But this is where the argument becomes too simple. GEO – properly understood – is not the same as "SEO for AI". It's about a fundamentally different question:
> SEO asks: Do I rank on page one? GEO asks: Am I mentioned in the AI answer, recommended, and how am I described?
These are fundamentally different questions – and they can have different answers. Companies that rank strongly for certain keywords are still not mentioned in AI responses. And conversely: some brands appear prominently in AI systems despite not holding top positions in classic search.
Why? Because AI answers are not distilled 1:1 from search results. Training data, citation patterns, how brands are framed across the web, presence in specialist publications, community mentions – all of this feeds in.
The Concept That Explains the Difference: Query Fanouts
To understand why GEO is not the same as SEO, it helps to understand how modern AI search systems work internally. The key concept: Query Fanouts.
When you ask ChatGPT or Perplexity a question – say: "Which marketing automation platform suits a mid-sized e-commerce company?" – the model doesn't answer it with a single search. It internally breaks it down into many sub-queries:
- "marketing automation software SME"
- "email marketing tools comparison 2024"
- "best CRM integration Shopify"
- "marketing automation reviews mid-market"
- "[Vendor X] vs [Vendor Y] comparison"
These sub-queries run in parallel. The results are synthesized into one coherent answer. That's a Query Fanout.
> What this means: Your brand doesn't need to rank for a single head term – it needs to be visible across an entire network of related, specific queries. And every AI system defines that network differently.
A brand that ranks on page one for "Marketing Automation" but barely appears in comparison articles, trade forums, or user reviews will likely be absent from the synthesized AI answer – even with strong SEO.
That's exactly why looking at your own ranking isn't enough. You need to know which AI answers actually include your brand – and for which implicit sub-queries your brand is considered relevant.
Three Reasons Monitoring Remains Essential
1. AI Systems Are Not Consistent
ChatGPT, Claude, Gemini, and Perplexity give different answers to the same question – with different brand mentions, different sentiment, different positioning. Relying solely on "ranking well" means you have no visibility into what's happening in the systems your customers actually use.
2. The Competition for Mentions Is Real
When a user asks ChatGPT "Which CRM would you recommend for a mid-sized company?" – which three vendors does the model name? That has nothing to do with your Google ranking. It has to do with your content, your brand perception in the training corpus, and your visibility in relevant sources.
3. You Can't Spot Trends in What You Don't Measure
That's the core of PromptScan: we make measurable what AI systems say about brands. Not once – but continuously. Not based on intuition – but on data. Without monitoring, GEO optimization is guesswork.
Conclusion: Hype Debate vs. Real Questions
Yes, there's GEO hype. Yes, there are providers selling companies expensive promises they can't keep. That deserves scrutiny.
But the right response isn't "ignore AI visibility." The right response is: ask the right questions. And measure what's actually happening.
PromptScan helps you do exactly that – no magic, no hype. Just data on how AI systems describe your brand today. Get started free at promptscan.de