LLMs are helpful, but a little lazy
Many people assume AI models gather information, compare sources, and reach conclusions. They do not. They recognise familiar patterns and generate the wording most statistically likely to come next.
A recent copyright case against Cohere illustrated this well. An LLM reproduced most of a New Yorker article almost verbatim because it looked like the most relevant match. Not malicious. Just how the system works.
This is why GEO needs clarity. If a model tends to reproduce the simplest, most repeated phrasing it can find, our job is to create content that is easy to recall and reliably accurate, so the model has a clear, consistent pattern to follow.
LLMs.txt is not a standard today
You may see optimisation tools or CMS platforms promoting LLMs.txt as the next big requirement. It is an interesting proposal, but no major AI platform uses it yet. It may mature, or it may not. For now, it is optional and experimental, not a new pillar of optimisation.
Reddit manipulation is not a strategy
Some brands hear that LLMs draw from Reddit and immediately jump to tactics that resemble old SEO shortcuts. LLMs don’t reward one-off extreme signals; they reward patterns that appear reliable, repeatable, and grounded in high-quality data.
Authentic expertise and consistent patterns perform better than anything artificially seeded. It may work in the short term, but long term it’s likely to be detrimental. GEO rewards clarity and credibility, not clever tricks.
Publishing more content is not a GEO shortcut
Volume is not harmful in itself, but it is not the lever people think it is. LLMs do not rank pages. They generate responses based on their understanding of language and context, drawing on patterns of meaning rather than real-world page authority or popularity.
If your content is consistent, well structured, and semantically clear, it helps. If you publish a lot of overlapping or contradictory material, it does the opposite. Quality signals outweigh quantity.
Models prefer factual stability, clear phrasing, conceptual alignment and content that directly answers intent. Keyword repetition does not influence them. It simply introduces noise.
GEO is discovery through language, not crawling
SEO is built on crawling, indexing, ranking signals, and links. GEO is built on what the model remembers and how it abstracts meaning. That means the role of the content strategist becomes even more important, not less. We are shaping the patterns the model learns from.
GEO and SEO are related, but not interchangeable
Confusion between the two leads to most of the unhelpful advice circulating right now. When people treat GEO as “SEO but with AI”, they end up chasing false signals, producing AI-only articles, or misunderstanding what retrieval systems actually do.
GEO has its own logic, its own workflows, and its own optimisation targets. Understanding the difference is what helps teams invest in the right things.
Looking ahead
None of this should feel discouraging. In fact, it is good news. GEO is not a mysterious dark art or a race to out-publish competitors. It rewards clarity, authority, and purposeful content design.
If we approach it with curiosity rather than shortcuts, 2026 can be the year we build frameworks that genuinely help both users and generative systems understand our brands.
Author: Lorraine van den Biggelaar, Head of Organic Search
