According to a recent study, the appearance of websites in AI searches is significantly correlated with their rankings in Google. Traditional SEO is therefore still important, but no longer sufficient on its own.
There’s currently a lot of discussion about whether traditional SEO will still be needed in the future, and whether “LLMO” or “GEO,” i.e., optimization for large-scale models and generative AI, will be more important in the future. Not to mention the question of whether all of this should still be called SEO, or whether the new terms will prevail.
At least for now, it seems that traditional SEO, i.e., optimization for organic rankings, still plays a role. A recent analysis by Ziptie of 25,000 search queries on various AI platforms revealed a correlation between page rankings and their likelihood of being mentioned in AI search results.
A website ranking first in Google has a 25 percent chance of appearing as a source in the associated AI Overviews. The higher a website ranks, the greater the likelihood of appearing in the search results of various AI platforms. This is shown by the results from ChatGPT, Perplexity, and the Google AI Overviews:
The reason for this likely lies in the way AI Overviews work. During a search, a few documents are first selected from the search index that best match the user’s query. Relevant information is then extracted from these documents. Finally, Google’s LLM (currently Gemini 2.5) uses this content to generate a comprehensive answer. Grounding, i.e., incorporating documents from the search, is intended to reduce errors in the K answers. Thus, there is a strong correlation between the AI answers and the search results.
However, there are also many exceptions. Often, sources mentioned in AI overviews are not found in the top rankings. This is because other factors come into play when generating AI search results, such as query fan-out. This is used, for example, in the new Google AI Mode. Here, a user’s search query is split into several queries. The answers to these queries are then used to generate the comprehensive answer for the user. Contextual factors and personalization also come into play here. Other AI searches, such as Perplexity, also use query fan-out.
Evaluation
Optimizing for traditional search results remains important, as AI searches incorporate these into their results generation. So, traditional SEO isn’t dead. However, it would be reckless to rely on tried-and-tested techniques and hope that they will continue to be sufficient to maintain visibility in the future. LLMs aren’t search engines. They generate results in other ways. Understanding these processes and aligning with them is the job of SEOs. Whether you continue to call this SEO or rather LLMO or GEO plays a subordinate role. What’s important is to stay up to date with developments and incorporate them into your strategies.
