AI is changing search. Here’s what Google says we should do about it.

Lynch’s view isn’t paranoia. It’s a response to a measurable shift in how people find information. AI Overviews at the top of Google, rows of shopping carousels, “People also ask” boxes — they all reduce the chance that a user ever clicks through to a website. It’s the rise of so-called zero-click search, and it’s changing the maths of organic traffic for every business that depends on it.

So what should you actually do about it? Fortunately, the answer isn’t to throw out everything you know about SEO. In fact, Google has just published its own guide to optimising for generative AI search — and the message is more reassuring, and more grounded, than much of the noise online suggests.

The headline: SEO is not dead

Google’s first big point is simple: the principles that earn visibility in AI Overviews are the same principles that earn visibility in classic search results. AI features on Google are powered by the same core ranking and quality systems. Get those right, and you’re already optimising for AI search.

There’s been a lot of talk in the industry about new disciplines like “AEO” (Answer Engine Optimisation) and “GEO” (Generative Engine Optimisation). Google’s position is firm: from their perspective, this is all still SEO. The acronyms have changed; the fundamentals haven’t.

How AI search actually picks your content

To understand how to be useful to AI search, it helps to know what’s happening under the bonnet. Google describes two key techniques:

• Retrieval-Augmented Generation (RAG): Sometimes called “grounding”. When an AI generates a response, it pulls in real, current information from Google’s search index — and the pages it pulls from are the same ones that rank well in normal search.
• Query fan-out: When you ask something complex, the model quietly generates a series of related questions in the background and pulls in results for each. A search for “how to fix a lawn full of weeds” might trigger sub-queries about herbicides, chemical-free removal, and prevention.

The takeaway: AI search isn’t picking content randomly or through some mysterious new mechanism. It’s reaching into Google’s index — the same index your existing SEO work is already targeting.

What to focus on

Google’s guidance distils into a few priorities worth doubling down on.

First, create non-commodity content. Generic “7 Tips for First-Time Buyers” articles compete with thousands of identical pieces and add little to an AI’s summary. Pages that share genuine first-hand experience, expert insight or a distinct point of view stand out — both to readers and to the AI systems summarising them. If a generative AI could have written it, there’s a good chance it already has.

Second, keep the technical foundations solid. A page can only appear in AI Overviews if it’s indexed and eligible to appear in normal search results. That means good crawlability, clean structure, sensible URLs, fast loading and a decent mobile experience. For agencies hosting client sites — like ours, where the bulk of our SEO clients run on Craft CMS and Webflow — this is well-trodden ground.

Third, organise content for humans. Clear headings, well-structured paragraphs, supporting images and video all help readers and AI systems alike. Semantic HTML matters, but Google’s guidance is pragmatic: aim for clarity over perfection.

Fourth, local and product details matter. For businesses with physical locations or e-commerce, keeping Google Business Profiles and Merchant Center data accurate gives AI search the structured information it needs to surface you in relevant answers.

What you can safely ignore

Just as useful is Google’s list of things you don’t need to worry about. If you’ve seen advice along these lines online, you can probably set it aside:

• LLMS.txt files: A proposed convention for telling AI systems how to read your site. Google says it doesn’t use them.
• “Chunking” content: Breaking pages into tiny, bite-sized blocks for AI consumption. Not necessary. Write at the length your topic deserves.
• Rewriting content “for AI”: AI systems understand synonyms and meaning. You don’t need to stuff in every keyword variation.
• Buying or engineering inauthentic “mentions” across the web in the hope that an AI will pick them up. Google’s spam systems already filter this out.
• Going overboard on structured data: Useful as part of normal SEO, but not a special lever for AI search.

In short: a lot of the “AI optimisation” tactics being sold are noise. The most reliable signal you can send to AI search is the same one that’s always mattered — high-quality, original content on a well-built site.

Looking ahead: agentic search

One area worth keeping an eye on is the rise of “agentic experiences” — AI agents that browse, compare and even transact on a user’s behalf. Google’s guide nods to emerging standards like the Universal Commerce Protocol, which suggest a future where your site needs to be readable not just by people and search crawlers, but by autonomous agents too. It’s early days, but for clients in retail, hospitality and any service that gets booked or bought online, this is a trend worth tracking.

Our view at CLD

Roger Lynch’s “plan for zero” framing was deliberately provocative, but the underlying message is sensible: don’t assume search traffic will look the same in three years as it does today. The right response isn’t to chase every shiny “AI optimisation” trend. It’s to focus relentlessly on the basics Google has now put in writing — useful, original content, on a technically sound website, with a clear brand voice that’s worth quoting.

For the clients we work with, that’s not a new playbook. It’s the playbook we’ve been following all along — now with renewed urgency.