Does AI SEO Even Exist? What Danny Sullivan Actually Said
AI SEO, LLM SEO, generative engine optimization. A new term shows up every week. Google's Search Liaison, Danny Sullivan, says most of it is not as new as it is being sold.
6 things you can act on after watching
- What Google's Search Liaison actually said about AI SEO
- Which parts of AI search are genuinely new and which are rebranded SEO
- How Google, ChatGPT, Perplexity and Claude decide what to cite
- Why pages cited in AI answers are usually pages that already rank
- What to ask a vendor selling generative engine optimization
- Where to spend the effort instead
Is AI SEO a real discipline or a rebrand
Every few weeks there is a new name for it. AI SEO. LLM SEO. Generative engine optimization. AI search optimization. Each one arrives with the implication that everything you knew is obsolete and there is a new discipline to buy.
Google's Search Liaison, Danny Sullivan, has been fairly direct that much of this is not as new as it is being presented. This episode works through what he actually said and separates the parts of AI search that genuinely changed from the parts that are an old practice with a new invoice.
What is actually different
Three things, and they are worth being precise about.
Answers get assembled, not just listed. For a growing set of queries the result is a synthesised answer with citations rather than ten blue links. That changes the shape of the page and the odds of a click.
Queries got longer and more conversational. People ask models full questions, and that behaviour is bleeding into search. Pages that answer a real question properly do better than pages built around a two word phrase.
The citation set is narrower. An answer cites a handful of sources, not a page of them. Being tenth is worth less than it used to be.
What did not change
This is the part that gets skipped in the pitch. The systems choosing what to cite are leaning heavily on the same signals that already determined ranking. Relevance, clarity, structure, credibility, and whether the page actually answers the thing being asked.
Analysis of what shows up in AI answers keeps landing in the same place: the cited pages are mostly pages that already ranked well. There is no side door. Visibility in AI results is downstream of conventional search performance, which means the work that earns one earns the other.
How to read an AI SEO pitch
A useful test is to ask what the service does that good SEO does not. The honest answers are quite short.
- Writing in a way that is easy to quote, with the answer immediately under the heading that asks the question. This is good SEO.
- Clean structured data so machines can parse entities and relationships. This is good SEO.
- Clear topical depth so the site is a credible source on the subject. This is good SEO.
- Making sure content is rendered server side so crawlers can read it. This is good technical SEO.
If the pitch contains something genuinely outside that list, it is worth hearing. If it does not, you are paying a premium for a renaming.
Where the real work is
The uncomfortable implication of all this is that there is no new lever. The pages that do well in AI answers are well structured, genuinely expert, and say something a model cannot generate on its own, which usually means firsthand experience and original data.
That was the advice before the models arrived. The models just made the gap between a page that has something to say and a page that does not considerably more expensive.
Key takeaways
- Nearly every tactic sold as AI SEO is a tactic that was already good SEO. The label is new, the work mostly is not.
- The pages getting cited in AI answers are overwhelmingly pages that already rank conventionally. There is no separate entrance.
- What genuinely changed is the click, not the ranking. Fewer informational queries produce a visit.
- Structure, clarity and direct answers help a model quote you, and they were already the right call.
- If a vendor cannot explain what their AI SEO does differently from SEO, that is the answer.
Questions people ask about this
Does AI SEO actually exist as a separate discipline?
Mostly not. Google's Search Liaison has pointed out that the tactics being sold under the AI SEO label are largely the same practices that already constituted good SEO. What changed is the shape of the results page and the likelihood of a click, not the underlying signals that decide which pages get surfaced.
What is generative engine optimization?
It is a term for optimising to be cited inside AI generated answers. In practice the recommendations under that label are familiar: clear structure, direct answers, accurate structured data, topical depth and credibility. The work is conventional SEO applied with attention to being easy to quote.
How do AI tools decide which sites to cite?
They lean heavily on the same relevance and quality signals that drive ranking, which is why the pages cited in AI answers are usually pages already performing well in conventional search. Clear structure and a direct answer placed where it is easy to extract make a page easier to quote.
Should I pay for an AI SEO service?
Ask what the service does that good SEO does not. If the answer is clearer structure, better schema, stronger topical depth and server side rendering, those are standard SEO deliverables and should be priced accordingly. If the pitch contains something genuinely outside that, it is worth hearing out.
Has AI reduced search traffic?
It has reduced clicks on informational queries that can be answered in a sentence. Commercial, comparison and local queries still produce clicks, because people making a decision want to see the source. The traffic loss is concentrated, not universal.
Want this handled for you?
Our in-house team works directly with US-based marketers. Tell us what you are trying to grow and we will tell you what we would do first.