Building a GEO strategy based on the leading LLM – ChatGPT

The ways in which people discover your brand are changing. Traditional search engine optimisation (SEO) strategies are simply no longer enough on their own.  

The search for, say, a SaaS product for a small tech company use to begin with, “best CRM SaaS solution”. Today, thanks to Large Language Model (LLM) tools like Gemini, Claude, or ChatGPT, it can be, “recommend a CRM SaaS solution for a small technology enterprise that leverages AI to optimise workflows and has powerful features that enable remote collaboration“ – and chances are, you’ll get all that information (and more) back in one concise paragraph – or better still, within a comparison table! 

That means planning a GEO (generative engine optimisation) strategy to ensure your brand, product or service is the one that surfaces among the search results. And if there’s one LLM you should prioritise your GEO strategy around, it’s ChatGPT.  

The growth trajectory of the leading LLM has been nothing short of extraordinary, solidifying its position as the most widely adopted tool of its type to date. As of May 2025, ChatGPT boasts approximately 800 million weekly active users and around 122.58 million daily users, processing over 1 billion queries daily. This rapid rate of adoption underscores its deep integration into users’ daily routines and its prominence in the AI landscape. 

In comparison, other LLMs are far behind in terms of their share of AI search. For instance, Google’s Gemini has garnered about 450 million monthly users as of July 2025. Notably less than ChatGPT’s user base. Then there’s an even greater drop off for popular tools like Claude and Perplexity (approximately 18.9 million and 22 million monthly active users respectively). This disparity highlights ChatGPT’s dominant position. 

While the results for LLMs are impressive, however, it’s not time to abandon your SEO strategy just yet. 

What’s comforting to know is that since Gemini’s (and the AI Overviews’) own algorithms are likely to be more closely aligned to Google’s search algorithm, if you prioritise ChatGPT for your GEO strategy, and Google for your SEO, you should have a really strong reach.  

So, let’s look at the main considerations for a ChatGPT-based GEO strategy…   

1. ChatGPT uses Bing to generate answers.

This is a fundamental – and perhaps surprising – point. When ChatGPT generates answers using its browsing tool, it retrieves content through Bing Search – not Google. The model looks at Bing’s website and content ranking, summarises the information, and cites or references brands based on what it finds.  

Why is this so important? Because if you don’t rank on Bing, you’re invisible to ChatGPT’s real-time answers. In essence, just as SEO was once all about Google, GEO for ChatGPT is now largely about Bing. B2B marketers who optimise for both engines can increase their brand’s visibility across the most important discovery pathways available today – and tomorrow.

2. If people talk about your brand, AI is more likely to give you a shout out. 

ChatGPT and other LLMs are trained on vast text corpus, scraped from the public web. During training, these models ‘learn’ word associations and frequencies. If your brand is consistently mentioned near key industry terms, your name becomes statistically entangled with those topics. For example, if the phrase “best enterprise CRM” frequently appears with the word “Salesforce” in the training data, the model learns to associate them. This matters because LLMs don’t rank links – they predict language. If your brand has been part of the public dialogue around a subject, the model is more likely to include you in its generated answer.

3. Retrieval-augmented generation (RAG) models look beyond the top link.

Most leading LLMs – including ChatGPT, Gemini, and Perplexity – integrate live search to fetch current content. They rely on retrieval to complement their training. In these cases, LLMs extract answers from top-ranked, recent, and contextually relevant web content. Studies show that when they cite content, the majority of these sources come from the top 20 search results. This means your organic ranking still matters and, crucially, you don’t need to be one of the first few links to be scanned and cited by LLMs. Overall, pages that rank well are more likely to be seen and selected by AI. 

4. ChatGPT surfaces content that demonstrates expertise. 

Beyond simply grabbing the top-ranked search results, LLMs like ChatGPT evaluate each potential source through a wider lens of relevance and authority. They favour content that is comprehensive and organised with clear headings or FAQ-style sections. This makes it easy for them to extract key information. If this content is published on domains already recognised for expertise and trustworthiness, its chances of being chosen increase further. Essentially, pages that meet these criteria – depth, clear structure, and proven authority – stand a far better chance of being surfaced in ChatGPT’s answers.

5. Choose helpfulness over hype – use utility-first, non-promotional content.

Promotional fluff is largely ignored. LLMs give priority to material that puts utility ahead of salesmanship. Copy laden with marketing hyperbole is typically filtered out, whereas educational resources that answer a user’s question are prioritised. Well-researched thought-leadership pieces, plain-language explainers, industry glossaries, and data-driven articles provide the kind of concrete, non-promotional value an LLM is designed to surface. As a result, content created with a ‘help-first’ mindset is far more likely to be summarised or cited in an AI-generated response. 

6. Originality and data points are the key to earning mentions.

LLMs actively seek out content that offers something new. This could be proprietary frameworks, fresh analysis, or hard-to-find data points. When you publish original statistics, you give the model a concrete fact it can cite verbatim and attribute back to your brand. Imagine your white paper reports that: 34% of manufacturers plan to reshore production by 2026; if that figure is novel and starts circulating, LLMs are likely to repeat it and credit you whenever the topic arises. In short, the more distinctive your insights, the greater your chances of becoming the authoritative source an AI surfaces and cites. 

And finally, keep it fresh…

What the above tells us is that ChatGPT and other AI models are constantly seeking credible sources with original data to cite. But as AI-generated content becomes more prevalent, a new challenge emerges: will originality become harder to find? 

Essentially, as AI-generated content floods the web, brands risk being lost in a sea of sameness. If you don’t at least give the model some net-new information to work with, the waters will soon be muddied with everyone citing the same few original sources.  

The point is: if you’re using AI to create content, do so cautiously. Human-generated content will be key for GEO. If you need to use AI as a tool to optimise content written by humans, great. But if you’re using it to churn out content without human input and insights, it’s unlikely to be valuable to humans, and it’s unlikely to help you on your GEO mission either.