You spend the time, effort, and marketing budget on optimising your content for AI, but your new pages never show up in any AI-powered answers. Sound like a familiar scenario?
This is really common, and the reason for it is simple: the AI tool you used has no idea your new content exists.
[cue gasps][cue head scratching][cue murmuring of angry villagers]
Don’t get out your pitchforks yet.
Knowing a little bit more about where AI gets its information from will help you understand why you just can’t seem to find your new stuff anywhere. It will also help you understand why optimising your content for AI-powered searches, otherwise known as generative engine optimisation (GEO), is still the best way to future-proof your brand against the changing nature of internet search.
Let’s figure out what’s going on:
Our players are:
- LLMs
- Search Engines
- You
Large Language Model Basics
What is it? A large language model, or LLM, is a type of generative AI that processes, understands (as far as robots go), and creates natural-sounding language.
How does it work? LLMs are trained on large amounts of text data (books, websites, articles, and more) to learn language: its patterns, structures, nuances, and relationships. All this training helps LLMs eventually come to predict the next word (or ‘token’) in a sentence.
So, by the time you start inputting questions, it ‘understands’ the context and meaning behind your query and generates a response that’s not only relevant (minus the occasional absurd hallucination) but also pretty indistinguishable from human lingo.
Examples: OpenAI’s ChatGPT, Claude, Gemini, Llama.
Why LLMs Aren’t Using Your Content
LLMs aren’t continuously being trained on new data. That would be an insanely expensive endeavour.
Instead, AI systems have a training cutoff date (for example, ChatGPT 4o-mini’s was mid-2024) and have no knowledge of anything beyond that point. This means that if you’ve published content after the cutoff date, your favourite LLM won’t know about it.
Unless, of course, it can use an LLM WebUI or API to search the internet. Which many can.
What to Know About Web-Search Functionality
When your favourite LLM searches the internet, it retrieves information beyond its training data and expands its knowledge base.
Each LLM uses specific search engines to help answer your question:
- ChatGPT uses Bing (and possibly Google, according to industry gossip - srsltid parameters, anyone?!)
- Gemini uses Google
- Perplexity uses its PerplexityBot + Bing APIs
- Claude uses Brave Search
- Google AI Overviews uses Bing (jokes)
AI won’t automatically search the web unless it deems it necessary to do so. You’re far more likely to get a web-assisted answer if it requires real-time data or if it’s more on the complex side.
How Do LLMs Find Your Content on Search Engines?
Let’s say, you’ve recently gotten into chess. One day, you started playing against your computer, and now you’re obsessed with watching YouTube videos of grandmasters using the Sicilian Defense on their opponents.
Your curiosity has evolved into fascination and now you want to study the moves of the most recent champions. You might type something like the following into your favourite AI (in this example we used ChatGPT):
Who are the most recent chess world champions and grandmasters and what were their moves?
We won’t give you the whole answer (because it’s long), but here is what ChatGPT said and did:
First, it displayed this:
Then, it showed various images it pulled from the internet, and its very first words were, ‘As of July 2025…,’ which happens to be the year and month that we made the query.
Needless to say, it had obviously searched the internet. Here’s how it works:
- The user types in the query (i.e., our chess question).
- The AI search tool reads the question and determines whether or not it needs to call on external sources to provide the best and most accurate answer.
- If it needs to utilise the web, it uses a dedicated web search tool or API to retrieve the info.
- It then processes the information it retrieves (i.e., filtering and crawling)
- The retrieved information is combined with its own knowledge to provide a well-rounded, comprehensive, and accurate response.
Why GEO Is So Important
The training data used to build LLMs becomes obsolete the moment they’re released, so why bother optimising your new content for these AI search engines?
It’s in the web search part. Combined with search engine optimisation (which you all know), GEO boosts your visibility in AI answers where external information retrieval is involved.
Here are some ways to improve your chances of AI mentioning your brand and citing your content in a related query response:
- Keep doing SEO: technical optimisation, keyword research, and backlinks are all important for LLM visibility.
- Create well-written and easy-to-read content that comprehensively covers the topic.
- Use conversational language as much as possible and choose long-tail keyword variations and semantically related phrases to help the AI crawler understand your content and context better.
- Use logical headers to help AI understand the information hierarchy.
- Interlink to related pages on your site to give the AI bot more context.
- Consider your online presence outside your website. As well as boosting your authority, AI sources information from everywhere, not just your specific website. The more spread out you are, the more likely your content will be used to answer user queries.
Now You Try
Now that you know a little bit more about how your go-to AI tool answers your query, why not play around with it to see what kinds of questions trigger a web search? It may just help shape your understanding of the kind of content AI looks for when it goes a-searching.
