~ Joseph Pack

How AI Search marketing is different from SEO

It seems that AI search differs from SEO in several crucial ways.

When we use Google, we tend to write short, direct searches — “best football boots” or “cheapest radiators.” But when we use an LLM, we often write long, detailed paragraphs that give far more context to the problem we’re facing. As a result, the output — assuming the prompt is good — is tailored to the individual, in contrast to organic search results, which tend to be the same for everyone (unless the search is location-specific).

If the first few sets of results on Google for “best football boots” don’t answer the specific query posed to an LLM, the LLM will look elsewhere. Some research shows that 52 percent of citations come from outside the top 50 traditional results — meaning the LLM pulls what it believes is the best (more on that later) answer regardless of where a webpage sits in the SERP.

Several factors seem to determine which information an LLM will select from a website. Chief among them are clarity, simplicity, and quality of writing. This doesn’t mean long, detailed pages — the kind that often rank well on Google — will be chosen. In fact, shorter, more concise text written with LLMs in mind seems to perform better: clear, direct sentences that introduce a topic followed by bullets, much like the way LLMs present information back to us.

Links also matter far less in AI search. LLMs don’t care about the ranking of a page — we’ve already seen that more than half of citations come from outside the top 50 results. So a site optimised to the hilt with links and dense content might rank number one on Google yet fail to appear in an LLM response. (It might appear, but not because of those factors.)

So what helps an LLM decide what to cite?

Authority appears to be high on the list. Information from an expert’s site — or a site linked to by several expert sources — carries weight. This isn’t about the domain authority of the link but the authority of the site providing it. If you write a blog about diet, for example, and reputable doctors link to it, LLMs are more likely to cite your work.

Here’s a point that I think shifts the game from traditional SEO to AI search: success depends more on doing genuinely good work — strong marketing, creating information that truly serves customers, building a reputation, earning links from reputable people — and becoming a trustworthy source.

Traditional SEO has always had an element of gamesmanship: how to stay a step ahead of the algorithm so Google ranks you highly. I think creative marketers with some analytical skill will do well with AI search. Overly technical marketers may struggle (which is ironic, given the technical nature of AI).

LLMs don’t judge a website by how fast it loads, how stable its layout is, or how smooth it feels to use. Traditional search engines sometimes use those factors to rank pages. LLMs, however, focus mainly on the text — what it says, how clear it is, and how well it answers a question. That said, users still care about performance, so a site still needs to be optimised for them. But this shift shows that text quality matters more than ever, and the technical SEO expert who excels at gaming the algorithm has a smaller role in optimising for LLMs. With LLMs, meaning and clarity matter more than technical details.

You need to:

These practices already help with modern search engines, but they’re even more important with LLMs.

Key performance indicators also change. Traditional rankings are not relevant in LLM environments. Instead, success should be measured by brand citations in sources or by brand mentions and links within LLM answers.