The New Rules of AI Search

14“Rank” in AI Responses

Ranking is a core tenet of classic SEO, but in GEO, ranking is a little different. Let's break it down.

If a tracking tool tells you that you rank number three on a particular LLM response, what they are actually saying is that you are the third‐mentioned brand in that specific output. Many tools will aggregate your position placements to give you an overall rank across a bank of queries.

But here is the challenge with that ranking system:

• The reasons why LLMs place brands in a specific order depend heavily on the query itself. For instance, I tested a query asking, “Can you list some significant country pop albums by women?” The list that came back was strictly chronological. Shania Twain was listed first, and Beyoncé's Cowboy Carter was last. The response order was a reflection of chronology, not necessarily of quality or authority.
• Not every query triggers a mention (of anyone). In classic search, there is always something to rank. Search engines will always show a list of websites unless you somehow trigger a Googlewhack. In LLMs, this is not the case. You don't necessarily get a brand mention for every single prompt. If your prompt isn't the kind that triggers a brand mention, you won't be seen, nor will you be ranked—because no one is seen or ranked.

These limits also change depending on the platform. Tools like Perplexity, OpenAI, and Gemini, which have more commercial capabilities, are far more likely to mention brands than base conversational ...