If you’ve been wondering whether the SEO playbook needs a complete rewrite for the AI search era, I have good news. It doesn’t.
Google just published an official guide on optimizing for generative AI features like AI Overviews and AI Mode, and the message I took away after reading it twice is genuinely reassuring: the fundamentals haven’t changed. If anything, it’s the clearest validation I’ve seen that the work my team does every day still drives results, just in more places across the SERP.
Here’s how I’m thinking about it for our clients.
The big idea: AI search runs on the same engine
Google says it plainly in the guide. AI Overviews and AI Mode are rooted in the same core Search ranking and quality systems we’ve been optimizing for all along. They just layer two new techniques on top:
- Retrieval-augmented generation (RAG). The AI grounds its answers in real, indexed pages from Google’s index, then links back to the sources that informed the response.
- Query fan-out. The model runs related background searches to gather more context before it answers. Ask “how to fix a lawn that’s full of weeds” and behind the scenes it’s also pulling on “best herbicides for lawns” and “prevent weeds in lawn.”
The takeaway for me is simple. If your pages are ranking well and earning visibility in classic Search, you’re already in the running for AI features too. There is no second engine to optimize for.
Google even addressed the “AEO” and “GEO” labels directly, noting that optimizing for generative AI search is optimizing for the search experience, and thus still SEO. I appreciate the clarity.
What I’m telling clients to lean into
The guide lines up neatly with what we’ve been recommending. A few things stood out as worth doubling down on.
Create content only you could write. Google contrasts commodity content like “7 Tips for First-Time Homebuyers” with non-commodity content like “Why We Waived the Inspection and Saved Money: A Look Inside the Sewer Line”. First-hand, expert, specific. That’s the bar.
When I review content with our team, I keep coming back to the same gut check: could a language model produce this without your business in the room? If yes, we rework it. The brands winning in AI search are the ones bringing genuine perspective.
Make sure your pages are eligible to appear. Google was explicit that a page must be indexed and eligible to be shown in Google Search with a snippet before it can surface in AI features. That’s the floor, and it’s why our audits still start with the unglamorous stuff: crawlability, indexability, canonicals, and snippet eligibility. Get this right and everything else has a chance.
Keep technical SEO healthy. Page experience, Core Web Vitals, JavaScript rendering, and reduced duplicate content all still matter. These are the boring wins that compound over time.
Lean into local and ecommerce signals. AI responses pull in product listings and local business information, so Google Business Profiles and Merchant Center feeds are part of the AI surface area, not just the regular SERP. If you have a physical footprint or a product catalog, this is high-leverage real estate.
What you can let go of
This is the part I think every marketing leader will appreciate. Google specifically called out a list of tactics circulating online that you don’t need to do for Google’s AI features (more on other LLMs in a moment):
- Creating llms.txt files or other “AI-specific” markup (though llms.txt is a quick lift if it might help in other AI tools)
- Chunking content into tiny pieces “for AI”
- Rewriting content in a special way for AI systems
- Chasing inauthentic “mentions” across the web
- Adding new schema markup specifically for AI visibility (keep using structured data for rich results, just not as an AI play)
I bring this up not to throw shade at anyone, but because I’ve watched well-meaning teams put real time and budget into these tactics. Google giving us permission to set them aside is freeing. It lets us put that focus back into the work that genuinely compounds.
One nuance worth calling out: Google’s guide is specifically about Google’s AI features. The broader AI search landscape is bigger than that, and tools like ChatGPT, Claude, and Perplexity each crawl and cite the web a little differently. Some of these tactics, especially the lightweight ones like an llms.txt file, take maybe ten minutes to put together. If a quick implementation gives you a non-zero chance of showing up in another AI tool your customers are actually using, why not ship it. My rule of thumb with clients: if it’s low effort and the downside is zero, do it. Where I push back is when something like that gets sold as the centerpiece of an AI strategy, because it isn’t.
One forward-looking area worth tracking
The guide also flagged agentic experiences as an emerging area. AI agents (browser agents, shopping agents, research agents) are starting to interact with websites the way people do. They read the DOM, the accessibility tree, sometimes screenshots.
This isn’t a sprint-this-week problem. But clean semantic HTML, solid accessibility, and a predictable UX are quietly becoming competitive advantages again. The teams that have always cared about those details are well positioned for what’s next.
The work hasn’t changed. The stakes have.
Do real SEO well: content only your team could write, technical hygiene that holds up, and signals that prove you’re a real business worth ranking. By Google’s own definition, you’re already optimized for AI search.
If you’d like a second set of eyes on where your site sits today, that’s exactly what our team does. Always happy to take a look.
