My Site Got Picked Up by a Language Model: A Deep Dive into GEO
01 My Site Got Picked Up by a Large Language Model
I discovered the concept of GEO quite a while ago—back then, I called it AISEO. Regardless of the name, the core idea is the same: do whatever it takes to make large language models cite your independent website in their responses.
Why did this catch my attention? Because one of my websites actually got cited by a language model.
When someone asked, “What is the official website of XXX?”, the model returned my site, claiming it was the official source. That was the first time I understood the potential behind GEO.
Recently, I started paying attention to GEO again because Google has started using large language models to respond directly to search queries. This inevitably means it will siphon off traffic from traditional websites.
That got me thinking: Is this a trend? What’s the endgame? Who will be affected? Who won’t? And more importantly—what can I do about it?
02 What is GEO?
The goal of GEO is simple: make large language models cite your site or product as a source.
When I read a piece by Song Xing, he described GEO as “polluting the training data of AIGC platforms.” Though a bit blunt, the idea is clear:
“AI doesn’t distinguish truth from falsehood. If enough people say something, or authoritative sources repeat it, the AI will believe it’s true.”
That reminds me of my earlier experience: my website repeatedly claimed to be the official site of XXX, and since it ranked high on Google, the language model took my claim at face value.
03 How Do You Do GEO?
Once the goal is clear, the path becomes simple: do whatever it takes to make the AI say what you want it to say.
Here’s a breakdown of the typical LLM query pipeline (especially those that rely on real-time web search):
- User inputs a question.
- Model analyzes intent, determines whether to perform a web search.
- If yes, what keywords to search?
- Where to search? (Google, Bing, own ecosystem?)
- After results come in, which ones should be considered authoritative?
- Content is synthesized—what gets shown first?
- Final answer is returned.
The key to GEO is planting your content in the right places, using the right keywords, and formatting it in a way that’s “digestible” to AI.
04 Will Platforms Punish GEO?
My conclusion: Eventually, yes. But not now.
Why? Because once platforms dominate the market, their next focus is maintaining quality. If everyone manipulates AI training data with low-quality content, users will lose trust.
But to be clear—this only applies to low-quality GEO. If your site truly provides value, platforms will welcome you.
Right now, platforms are still in the “land grab” phase. They're too busy optimizing infrastructure to worry about bad actors—and some may even welcome GEO efforts that improve their models.
05 Is SEO Dead?
Nope. Not even close. In fact, SEO and GEO complement each other.
GEO brings user engagement and behavior data. SEO boosts your search ranking. They feed into each other.
Even more so, SEO today is no longer just about algorithms. It’s about product thinking—creating things that genuinely solve problems.
06 My Takeaways
- GEO is far from its peak. We haven't even hit the real gold rush yet.
- It's worth experimenting with GEO, but don’t pour your entire budget into it—models are still evolving.
- GEO isn't going away. It's a reflection of the times.
- Think of GEO as part of SEO—especially off-site SEO.