Insights
Do 'Normal' SEO
The major players in Search keep issuing similar advice…
At the recent Google Search Live event, Gary Illyes (from Google) said that you don't need to do GEO, LLMO or anything else to show up in Google AI Overviews.
You just need to do "normal SEO."
(via Kenichi Suzuki)
Meanwhile…
Fabrice Canel (Bing's Search Product Manager) said the way you optimize for AI Search is by following SEO guidelines.
I'd add that, while traditional SEO is still important, data shows that LLM and AI Search visibility requires a slightly different approach.
Conversational search is shaping our industry
Back in May, Google's promotional announcement for AI Mode stated:
People are coming to Google to ask more of their questions, including more complex, longer and multimodal questions.
An example query in their announcement is a massive 27 words long and reinforces the conversational nature of AI Mode.
To put that into perspective, Semrush published a study showing queries in AI Mode are typically longer, averaging ~7.2 words vs. ~4 words in classic Google search.
This is how Google wants you to use AI Search
AI Mode is designed to eliminate the need for numerous searches, so you can get the answer you need with a single query.
The "query fan-out" technique also performs multiple related searches across subtopics to provide you with additional information.
While Google tells us to focus on creating "people-first content," they offer close to zero direction on appearing in AI Search.
The point is, listening to Google's old catchphrase doesn't help us.
We can no longer just do '"normal SEO," because we can't ignore the machines that decide what to surface and show our audiences.
I'm not suggesting you stop creating human-focused content.
But while Google figures out how best to display search results and continues to launch new experiments (like Web Guide), we must keep up with how AI systems interpret and present information.
The good news is, SEOs and marketers are always testing stuff, tracking what's happening in AI search, and sharing it on social media, in forums, and in blogs.
Have you run any tests or discovered something you'd like to share in this newsletter?
Send me the details and I'll feature some replies next time.
Quoteworthy
Anthropic studied what gives an AI system its ‘personality’ — and what makes it ‘evil’
"It’s sort of getting peer-pressured by the data to adopt these problematic personalities, but we’re handing those personalities to it for free, so it doesn’t have to learn them itself. Then we yank them away at deployment time. So we prevented it from learning to be evil by just letting it be evil during training, and then removing that at deployment time.”
—Jack Lindsey, an Anthropic researcher
I want to know what happens if/when they forget to "yank them away at deployment time?"
via The Verge
Rank Tracking in AI Search Doesn't Exist
Rank tracking was slightly flawed even before AI because Google’s search results varied by user and query.
AI just makes this worse.
Despite what some tools (and 'AI opportunists') claim to offer, you can't track rank in AI search systems.
Every query generates a unique response based on context, user history, and how the question is phrased.
There's no static ranking to track because there's no consistent results page anymore.
Even if a tool shows your site in a snapshot today, that might change tomorrow with a slightly different query or user context.
What should SEOs do instead?
Audit AI visibility: Prompt AI engines and LLMs to see if your client’s brand or content appears in AI-generated answers across common queries.
Monitor factual accuracy: Check for brand mentions, outdated or incorrect facts, and areas where competitors are being cited more often or more credibly.
Optimize for inclusion: Structure content clearly, cite trustworthy sources, and create pages that directly answer the kinds of questions AI models summarize.
Focus on authority signals: Backlinks, brand mentions, and high-quality content still matter, because they’re inputs for the models and influence what gets surfaced. Being indexed in Google is fundamental to inclusion in AI answers.
Key Takeaways:
- AI-generated answers are dynamic, personalized, and not position‑based, making traditional rank tracking obsolete.
- Instead, monitor your brand’s factual mentions, correct misinformation, and prioritize visibility within AI-generated responses—not in rankings.
- Stay open‑minded: Our SEO strategies are shifting from "How high do we rank?" to "Are we the trusted source AI systems cite?"
Brand Mentions By LLMs
Are we clear on the difference between brand mentions and brand citations?
You might find yourself having to explain this at some point – I know I've had to.
The key difference between an AI mention and an AI citation lies in what they refer to within AI-generated responses.
AI mentions are general references to brands in AI-generated responses, e.g., ChatGPT or Google's AI Overviews.
A mention may be linked or unlinked – and sometimes mentions and citations both occur in the same answer.
AI citations are references to specific sources or pages that were used to generate the AI response. If you're cited, you got a link.
Semrush analysis of non-branded queries
I found this Semrush analysis of non-branded queries interesting.
They used 1 million non-branded queries across five LLMs and found that AI models include brand mentions in 26% to 39% of responses.
A breakdown of AI mentions’ prevalence by LLM
AI Overviews are built on Google Search. They synthesize information from sources that are already indexed and considered high quality.
We also have evidence that ChatGPT falls back on Google when Bing doesn’t have the answer or index.
Abhishek Iyer (a former search infrastructure engineer at Google) and Aleyda Solis both recently published articles on how ChatGPT uses Google Search.*
Abhishek decisively states, "the paid version of ChatGPT is using Google Search."
The bottom line:
The above underscores why traditional SEO and Google indexing remain vital even for AI-driven search visibility.
Brand building and SEO are two sides of the same coin.
*Aleyda Solis' article: Confirmed: ChatGPT uses Google SERP Snippets for its Answers
*Abhishek Iyer's article: ChatGPT web search uses Google Search and NOT Bing Search
Do you follow me on LinkedIn? I share regular tips and stories I don't have room for here. Come and join me.
