Welcome to LEO.
It’s not exactly a secret that when you use any popular search engine to find something (Google, let’s say) the first response you see is an AI-generated result.
Initially, most marketers have regarded this as a technical curiosity, but not necessarily a problem (or an opportunity) that directly impacted their online marketing strategy.
However, ChatGPT and other Large Language Models (LLMs) like Claude, Grok, Gemini, LLaMA, and Mistral continue to be adopted by millions of people. It’s time to pay a lot more attention to the impact this might be having on search engine marketing.
Before we assume traditional search engines are collapsing, it’s worth remembering that not every GPT use is a search-type query. In fact, you can find some good statistics online which indicate that, when comparing apples with apples, ChatGPT-derived search vs traditional Google search remains tiny by comparison. GPT search is estimated to be about two per cent of the daily volume of Google searches. Search the article by Sparktoro to read a lot more on this; it’s good reading.
The concern is more that even the search engines are leading with AI-derived results, and it’s not clear how to rank in those results.
In recent months, we have seen the emergence of a new marketing term: LLM Search Engine Optimisation, or LEO for short (pronounced L. E. O.). While there might be a term for it, the science and suppositions behind how to do LEO well remain opaque.
The immediate areas of concern and interest for marketers are:
- How relevant are traditional search engines for driving web traffic compared to searches performed directly from AI context windows?
- How do you ‘rank’ for results when it comes to LEO?
Let’s address the first question: How relevant are traditional search engines for driving web traffic compared to searches performed directly from AI context windows?
Clearly, there is a shift happening here.
Even the search engines themselves understand that AI can offer a high-quality end customer experience compared to publishing a list of links and short text descriptions.
For marketers, that doesn’t negate the importance of Google (or other search engines) as a channel to market; it just means that the process to achieve search results is evolving, but it could evolve very quickly and suddenly.
More and more consumer and business technology companies are embedding AI-search into their operating systems or core applications, and people will always do what’s easiest. That means the intuitive interface of GPTs is likely to drive an even faster uptake of GPTs as a primary search tool.
The recommendation today is not to stop your successful paid and organic search programs. That would make little sense. However, it is important to track search trend data, ROI, and third-party reports very carefully to make sure the inflection point isn’t overlooked when it happens.
Also, keep in mind that companies like Google may drive this change even faster as they seek to improve search engine experience and deploy their own LLMs more widely. All the long money is on AI eventually dominating.
To the second question: How do you ‘rank’ for results when it comes to LEO?
We’ve all come to understand (as much as any of us can) how to rank in an organic or paid search environment, though it’s still largely a black box. There is no manual, just a lot of clues from Google and other platform owners, plus a lot of A/B testing by marketers, to figure out what works.
We know that two things matter most: authority of content and quality of content. Do the same things matter when it comes to achieving ranking in a Google search AI recommendation, or a ChatGPT search response?
In short, the answer is yes. In fact, it has become more important. AI is much more cognisant of context, meaning, relevancy, and authority at an individual level than the garden variety search engine is.
Our own research consistently shows that links to sites with high levels of authority count for a lot. This is interesting on a couple of levels:
- High authority sites continue to be directories, marketplaces, and media sites when searching for products and services. The nice thing about AI search is that it will provide a list of citations which can provide clues as to where you might need to invest additional marketing (such as paying for directory listings, etc.). It’s not that the directories are themselves valuable (sometimes they are), it’s that you need to show up on AI’s radar. It’s an SEO tactic, but it also matters a lot to AI right now.
- Public relations, and specifically media relations, is going to be increasingly important if you want AI to find you. Media coverage has always been important for building brand authority with critical audiences and good media coverage has been useful for SEO. However it is clear that AI weights media articles heavily too. This may be a function of LLM training data, the network effect of media coverage (e.g., syndication, sharing, cross platform discussion, etc.), and how easy it is for an AI to access and consume media stories.
The takeaway, in summary, is keep watching this space. Talk to our team if you want to start working on improving your AI search results and keep checking in on the research for yourself, too.
Ironically (or not), ChatGPT’s deep research tool can provide some excellent insights into market statistics, trends, and ranking strategies for LEO, though the suggestions are principles rather than instructions.
