Google recently published its first official guide on optimizing for AI Overviews and AI Mode. For anyone who has spent the past year drowning in new acronyms and “GEO hacks,” it’s a welcome reset.
The headline takeaway: AI search is still SEO. AEO and GEO aren’t separate disciplines; they’re new labels for the same fundamentals. And most of the tactics being pushed in marketing circles right now, things like llms.txt files, content chunking, AI-specific schema, and there’s a lot of mixed messaging.
Like most official search documentation, Google’s guide only tells part of the story. So we read it against the recent independent research on AI visibility to see where the advice holds up, and where you still need more nuance:
What The Guide Actually Says
In Google’s framing, AEO and GEO are simply SEO with a generative layer on top. The reasoning is straightforward: AI Overviews and AI Mode run on the same core ranking systems as standard Search. They use retrieval-augmented generation (RAG) and query fan-out to pull from the existing index. There’s no separate “AI index” to optimize for.
Google recommends focusing on three things:
- Creating unique, non-commodity content with original insight
- Maintaining a clear technical structure so your pages can be crawled and indexed
- Optimizing local and e-commerce details through Merchant Center and Google Business Profiles
The guide also points to the next phase of AI search: Agentic experiences, in which AI systems interact with a site on a user’s behalf, comparing products, checking availability, filling out forms, or completing purchases. For commerce, Google is already pointing to emerging standards such as the Universal Commerce Protocol, which is designed to enable AI agents to connect with business systems for product discovery, checkout, and post-purchase actions. For non-commerce sites, the advice is to make sure your site is easy for browser agents to use: Crawlable content, clear navigation, accessible forms, and fewer UX patterns that block automated interaction.
What Google Is Telling You To Ignore
The guide includes a mythbusting list that aims to highlight several tactics adopted across the industry over the past few months:
- llms.txt files – not used, not required.
- Chunking content for AI – not needed; Google handles long pages fine.
- Rewriting content specifically for AI – synonyms and meaning are already understood.
- Structured data for AI visibility – useful for rich results, not for AI citations.
- “Inauthentic” brand mentions – seeking them out won’t help.
The first four hold up well against the data.
SE Ranking analyzed more than 300,000 domains and found that adding llms.txt had no measurable impact on ChatGPT citation likelihood. The model actually got more accurate once it removed that as a variable. Otterly.AI ran a 90-day log study and saw only 0.1% of AI crawler requests ever touch /llms.txt. Most AI crawlers don’t fetch the file at all.
On schema, Ahrefs tracked 1,885 pages that added JSON-LD against a matched control group. The treated pages saw no citation lift across AI Overviews, AI Mode, or ChatGPT.
So on the technical “hacks,” Google and the independent data are largely saying the same thing. If you’ve been spending hours on these, you can stop.
Brand Mentions Are Still Important
The point getting the most pushback is the one on “inauthentic mentions.” SEO analyst Lily Ray called it “classic Googlespeak” in her breakdown on X: The line is vague enough to mean almost anything Google wants it to mean later, and the practical takeaway for SEOs is unclear.
Google is right to warn against manufactured mentions. Chasing random mentions across low-quality sites is not a strategy. But that doesn’t mean off-site visibility is irrelevant. The data on authentic brand mentions tells a different story:
- SE Ranking (216,000+ pages): Domains with millions of Reddit mentions averaged 7 ChatGPT citations versus 1.8 for low-mention domains – a 3.9x multiplier. Quora showed 4.1x.
- Semrush: In its AI citation study, Reddit and Wikipedia were still ChatGPT’s two most-cited domains, even after a sharp September drop in citation share.
- Ahrefs: Branded web mentions correlated at 0.664 with AI Overview visibility – one of the strongest signals in the dataset.
- Contently: Roughly 75% of the citation signal lives off-site.
Across these studies, brand mentions across the web come up as one of the biggest drivers of AI citations. Google’s guide doesn’t frame it that way. It warns against “inauthentic” mentions and points you back to your own helpful content. That’s not wrong, but it’s a long way from the full picture.
Page Format Can Influence AI Retrieval
Another point worth questioning is the guide’s stance on rewriting content for AI. Google says you don’t need to write in a specific way because AI systems already understand synonyms and intent. Fair enough. You don’t need awkward AI-only pages or separate versions of every article.
But format still matters for retrieval. CXL’s analysis of 100 Google AI Overview citations found that 55% of cited snippets came from the first 30% of the source page. AI Overviews were more likely to cite answers that appeared early, were easy to extract, and directly addressed the query. Ahrefs’ research points the same way: In a study of 1.4 million ChatGPT prompts, cited pages aligned more closely with the fan-out queries ChatGPT generated in the background than with the original prompt alone.
The takeaway isn’t to create AI-only content. It’s to structure pages so the answer is easier to retrieve. Put the clearest answer near the top, use direct sections, keep titles aligned with the specific question being answered, and give each page a clear role. Google’s framing implies that on-page format doesn’t matter much. The data says it matters a lot, just not in the way the old SEO playbook assumed.
What To Keep In Mind
The guide is Google-specific. It covers AI Overviews and AI Mode, not Gemini standalone, ChatGPT, Perplexity, or Claude. It’s useful for understanding how Google wants you to think about AI search, but it shouldn’t be treated as the full AI visibility playbook.
That matters because the wider ecosystem relies on a broader mix of signals. Google is anchored in its own index and ranking systems. Other platforms lean more heavily on forums, review sites, third-party articles, and product databases. OpenAI, for example, partnered with Reddit to bring more Reddit content into responses. The advice overlaps, but it isn’t identical.
What To Do Now
If you’re wondering where to focus, start by cutting the GEO hacks that don’t move citations. Don’t add llms.txt files, chunk pages, or build a new schema just for AI search. Google’s guide and the independent studies agree here: These look like AI optimization, but there’s little evidence they change whether you get cited.
Then look off-site. Run your top buyer queries through ChatGPT, Perplexity, and AI Mode. Don’t just check whether your own pages show up; look at which third-party sources keep appearing. Reddit threads? G2 pages? Quora answers? Comparison posts? Review sites? Industry publishers? That gives you a simple map of where AI systems are already pulling evidence. If the same sources keep showing up, that’s where your brand needs to be visible – not through fake mentions or spammed comments, but through real participation: Better reviews, stronger partner pages, useful comparisons, and content buyers trust.
On-page, don’t rewrite everything for AI. Start with the parts that help retrieval: Titles, headings, intros, and the first few sections of your most important pages. Query fan-out happens before the page gets read, so if AI Mode breaks a broad query into smaller sub-queries, your page needs to make clear which specific questions it answers. That means clearer titles, faster answers, and sections that map to real buyer questions, with less vague category content.
Finally, don’t treat the guide as the full playbook. It’s Google’s guidance for Google’s AI search features. Useful, but limited. AI Overviews, AI Mode, ChatGPT, Perplexity, Claude, and Gemini don’t all work the same way. The overlap is still SEO: Crawlable pages, useful content, clear structure, and trusted signals. But AI visibility also depends on where your brand shows up outside your own site.
Stop chasing technical GEO hacks. Keep the SEO fundamentals. Make your best pages easier to retrieve. And build stronger visibility in the places AI systems trust.
Not Sure Where AI Search Is Already Pulling Answers About Your Business?
That’s exactly where we start. At First Rank, we map the queries your buyers are actually asking, find the sources AI systems trust to answer them, and build a plan to get your brand showing up on your own pages and across the places that influence citations. No GEO gimmicks, just the SEO fundamentals done well and pointed at where search is heading.
Book a free consultation with us, and let’s find out where you stand in AI search.
Jacob Kettner is the owner and CEO of First Rank Inc., a digital marketing agency based in Winnipeg, Manitoba. He currently sits on Manitoba Chamber of Commerce Small Business Advisor Council which assists people grow their small businesses in Manitoba.
