It's occurred to me over the last year that we've talked endlessly about AI - new models and rollouts, updates, features, growth numbers, even the personalities - but we've barely touched on the one question that keeps coming up behind the scenes:
"How do you rank higher on LLMs?"
ChatGPT? AI Mode? Perplexity? Claude? Should each be optimised differently?
These are fair questions.
But here's the thing: they completely misunderstand how these models actually work.
And there are plenty of companies selling the AI dream - putting out misinformation and promising secret formulas to "rank number one in ChatGPT."
And I don't want it to cost any of you.
Last week, I started an AI course run by Britney Muller - and noticed her sharing thoughts on this exact topic on LinkedIn, which sparked hundreds of comments. We also dug into it during lesson one.
In essence, large language models like ChatGPT and Claude aren't search engines - they're probability engines. Continuation machines. Based on everything they've read across the internet, they predict what comes next - aggregating the most statistically likely response.
Unfortunately, you can't optimise your way into that statistical likelihood. The model has already been trained. You need to have been mentioned - often and positively - across the web months or years ago.
From that perspective, the more that's out there about you, the better - you could think of it as a kind of popularity contest.
If you come from an SEO background, brand mentions are like the new backlinks - but it's even closer to real-world recommendations. Every time your brand, product, or people are mentioned positively, it acts as a small vote of confidence in the model's training data. The more people who recommend something, the more likely others - and the model - are to trust it.
So, alongside digital PR, the other opportunity is figuring out how to get customers talking about you - and for you.
It's also worth remembering that most LLMs take about a year to train. And because those training sets take so long to build, much of what they know is already out of date. Refreshing your content or research regularly isn't going to propel you to the top.
Unless… some hybrid models can pull in real-time data - and when they do, they go to the same trusted sources you and I do: search engines, via APIs that tap into Google or Bing. If that's the case, we need to focus on SEO.
So, despite all the new tech and terminology, here's what actually matters:
Stop doing:
• Paying agencies for platform-specific "AI optimisation"
• Creating separate strategies for each AI platform
• Constantly refreshing content hoping it enters the next training cycle
Start doing:
• Build genuine brand authority through PR and thought leadership
• Earn mentions across reputable sites (digital PR)
• Focus on traditional SEO activities, such as publishing the best content on the internet (it powers the real-time layer in hybrid models)
• Get customers talking about you (reviews, advocacy programmes, social proof)
Whatever you call it - these are all the things we've already been doing in SEO. The fundamentals haven't changed. Which is exactly what SEOs like Britney, myself, and others keep saying.
