Google’s GenAI optimisation guidance is here – here’s what it means for you

For the past year, we’ve been drowning in buzzwords: GEO, AEO, LLM optimisation, AI-first content architecture, prompt engineering…the list goes on. Now, Google has FINALLY released its official guidance on optimising for GenAI in Search, giving businesses their clearest indication yet of what actually matters in Google’s AI-powered search experiences like AI Overviews and AI Mode. We are here to break it down.

Here’s what matters, what doesn’t and what you should do next.

This blog covers:

• What Google says about GEO
• Why llms.txt files can stay in the bin
• Chunking content..Yay or nay?
• How local visibility changes in AI-powered search
• Why “non-commodity content” might be the biggest ranking opportunity right now
• And what Google means by “agentic experiences” (because yes, that sounds mildly terrifying)

GEO is still fundamentally SEO

If you were expecting Google to reveal some revolutionary “AI ranking system” that changes everything…Sorry. Google makes it clear that optimising for GenAI search experiences still relies on established SEO principles:

• Technical SEO
• Crawlability
• Helpful content
• Strong internal linking
• Site authority
• Real expertise
• Content people actually want to consume

There’s no separate playbook being introduced here, these principles still form the foundation of visibility in both traditional search and AI-generated results.

So despite the new terminology floating around, the underlying reality hasn’t really changed. If your SEO isn’t strong, nothing else is going to sit on top of it and fix that.

via Giphy

No need for llms.txt files

One of the more noticeable myths Google has quietly shut down is the idea that websites need an llms.txt file – essentially a file designed to help large language models understand your content. This has been doing the rounds for a while, with people testing whether adding dedicated files for large language models has any impact on visibility or interpretation. Some people swore it would become essential, others thought it was complete nonsense. (We were leaning heavily towards the second one.)
Google’s position is simple: it’s not required.

Content chunking isn’t necessary, but clarity still is

Another myth Google quietly killed: you need to “chunk” your content into tiny blocks so AI can process it. Google says there’s no need to artificially restructure content specifically for AI systems.

That said, this is where nuance matters. While you don’t need to completely restructure content into rigid AI-friendly blocks, the idea of “answer-first” content still holds up. Content that is clear, direct and structured around real user intent tends to perform better across both traditional and AI-driven search experiences.

So it’s not about formatting for AI. It’s about removing friction from understanding.

You don’t need “special” AI schema markup

Another clear message: there’s no requirement to introduce experimental or AI-specific structured data.

Google’s advice is to sticktick with the structured data that already works and forms as part of a good SEO strategy to get you featured in Rich Results:

• Organisation schema
• LocalBusiness schema
• Product schema
• Article schema
• Review schema
• JobPosting schema

In other words, this isn’t a moment for reinventing structured data. It’s a reminder that the basics still work when they’re implemented properly. Which, reassuringly, is exactly where most good SEO strategies already sit.

📍Your Google Business Profile just got even more important

Generative AI responses can include product listings, product information, and information about local businesses, which means that your Google Business Profile needs to be in good shape. If AI systems are being used to surface recommendations, compare providers, or summarise options, then structured and reliable business information becomes critical. Make sure your contact details, name and address are consistent with what’s on your website, your categories and services meet search demand, your reviews are consistently trickling in and that your profile is active and fresh with posts and photos.

So if your Google Business Profile hasn’t been updated since someone uploaded a blurry office photo in 2017…that will be a problem. Local SEO was already important and AI search makes it even more valuable as a core visibility layer.

The bit we’re most excited about: “Non-commodity content”

This is arguably the most important part of Google’s guidance.

Google spends a huge chunk of its guidance talking about something called non-commodity content. This could change how brands approach content entirely. They draw a clear line between generic, widely available content and content that is grounded in real, unique experience.

Google’s example of commodity content: “7 tips for first-time homebuyers.”
Helpful? Sure. Unique? Not really. Anyone could write it, and there’s probably 30 other blogs just like it sitting in the Google SERP

Now compare that to their example of non-commodity content: “Why We Waived the Inspection & Saved Money: A Look Inside the Sewer Line”. Now that’s different.

That’s:

• Experience
• Expertise
• Emotion
• Real-world insight

Content which aligns with Google’s E-E-A-T is content AI systems actually want to reference.

And this is where a lot of brands will need to shift. AI systems can summarise generic information from their training data, without the need to head to your site. What they’re far less good at replacing is actual experience, nuance and real-world detail. This is where we need clients to lean in more. Not “set and forget” content, but stories that come from inside the business. Clients often expect to appoint a marketing partner and have content handled entirely for them, but the reality is it works best as a collaboration – they bring the insight and expertise from inside the business, and we turn that into polished content that performs.

We need stories. Real ones, like:

• If you’re an automotive company, try writing about a dramatic vehicle recovery story
• For accountants, we want to talk about how you saved client £Xm through one simple change
• A detailed case study

AI can mimic your expertise, but not your lived experiences! And increasingly, that’s the kind of content Google seems to be rewarding. AI doesn’t just want information anymore. It wants perspective.

“Agentic experiences”… 👀

And this one’s definitely worth watching.If you’re wondering what that means: browser agents and AI assistants may start interacting with websites on behalf of users.

Google breaks this down into three ways agents understand a site:

• Visual layer: interpreting layout like a user would
• DOM structure: reading the underlying HTML hierarchy
• Accessibility tree: using the semantic version of the page

Basically, AI might not just read your website, it might start using it. That could mean an AI agent visiting your website, understanding what’s on the page, comparing options, filling in forms, navigating product journeys or helping a user complete a task without the user manually clicking through every step.

This matters because traditional SEO has mostly focused on whether search engines can crawl and understand your content. Agentic experiences add a new question: can an AI assistant actually understand and navigate your website like a user would?

It’s early, and there’s still a lot of uncertainty around what this looks like in reality.

Google does suggest businesses should start paying attention to agent-friendly webdesign, even if the full implications aren’t fully defined yet.

Right now, it’s early days. We’re not pretending there are clear answers. But there are practical things worth reviewing now:

• Make sure key journeys are simple, logical and easy to navigate
• Where possible, use real HTML buttons and links rather than turning generic page elements into clickable actions
• Avoid moving key elements around from page to page – if your ‘Enquiry’ button changes position across categories, agents may struggle to follow the journey.
• Label forms clearly, ensuring fields like name, email, location or booking date are clearly labelled so agents understand what each input is for
• Avoid hiding important information inside awkward UX patterns
• Use structured data where relevant
• Test whether important pages make sense visually and semantically, not just from an SEO crawler’s perspective

But the fact that Google has mentioned it as one to keep an eye on, it’s definitely worth paying attention.

The takeaway

Now that Google has released its GenAI search guidance, the direction is clear: the fundamentals of search haven’t changed, but expectations have. SEO evolves all the time, and the GEO era is just another way for us to adapt to show up on these new platforms. GEO is fundamentally the same as SEO, but the bar for what counts as “good” content has gone up. Generic, surface-level content will find it harder to surface in both traditional and AI-driven results.

The focus is shifting towards real substance – experience-led content, specific examples, and insights that can’t be easily replicated or summarised.

In short: not more content, better content.

The businesses that adapt won’t be chasing every new AI acronym. They’ll be the ones already doing the basics properly, and backing it with something genuinely useful.