Google has set out how it wants you to optimise for AI Overviews and AI Mode, and most of the advice is sound. But it is advice about Google's surface, written by the company with the most to lose from the idea that search has changed. Your buyers are increasingly asking elsewhere. This article explains what Google's guidance gets right, what it leaves out, and why marketing teams need an expert read across every platform.

By Helenor Rogers, CEO of KnownEntity.ai

Google has published official guidance on how to optimise a website for its generative AI features, AI Overviews and AI Mode. Its central message is that optimising for AI search is still SEO, because Google's AI features draw on the same ranking and quality systems that have always governed the index. The advice is sound, and it confirms that the content fundamentals still matter. It also describes one surface among several that buyers now use to decide who makes their shortlist, and it is written by the company with the most to lose from the idea that search has fundamentally changed.

If you read Google's guidance as a verdict on AI search as a whole, rather than guidance about Google's own surface, you will optimise confidently for the one platform you understand whilst potentially being invisible on the other platforms that your buyers are actually using.

This matters because the buyer journey is fragmenting across platforms. McKinsey's AI Discovery Survey (October 2025) found that a brand's own website accounts for only 5 to 10% of the sources AI search references, that half of consumers already use AI search, and that 44% describe it as their primary way of finding information. Most of the platforms shaping those answers are not Google's index. Understanding how to be found by Google is useful. It is not the same as understanding where your buyers are looking, or what they are being told when they get there.

What Google's guidance gets right

Google's core instruction is one most up-to-date practitioners already understand: create content with a genuine point of view, write for people rather than for a ranking system, and avoid commodity material that restates what is available everywhere else. Google draws the distinction sharply, contrasting a generic "seven tips" article built on common knowledge against content carrying real expertise and first-hand experience. The latter earns its place; the former adds little.

This is consistent with how AI Authority works. Appearing in an answer for the wrong reasons, or for things that do not matter to your commercial strategy, carries little value. The work that counts is ensuring AI systems understand what your business genuinely stands for and reference it for the things that matter. Google's emphasis on a unique, expert point of view sits comfortably alongside that principle.

The technical advice holds up too. Make content crawlable, use clear structure and descriptive headings, and provide a good page experience. These are foundations, and they remain foundations whether a human or an AI agent is doing the reading. When Google says the fundamentals still apply, the honest response is agreement.

Why this is guidance for Google, and why that distinction matters

Google's document is guidance for Google, and it says so on almost every page. That scope is the single most important thing to understand about it.

Google's AI features are grounded in its own index through retrieval-augmented generation and query fan-out. The model pulls from pages Google has already crawled and ranked, then synthesises an answer. Within that architecture, Google is correct that you do not need special files, content broken into artificial fragments, or markup invented purely for machines. Its systems are built to understand ordinary, well-made web pages, so ordinary, well-made web pages are what they reward.

ChatGPT, Perplexity, Claude and Copilot do not all work the way Google's index works. They draw on different training corpora, different live retrieval methods, and in several cases no connection to Google's ranking systems at all. A model reasoning about your business from what it absorbed during training is solving a different problem from ranking a freshly crawled page. A platform retrieving live results uses its own sources and its own logic for deciding what to trust. Advice optimised for one retrieval architecture does not automatically transfer to the others, and treating it as universal is where marketing teams will go wrong.

Read the incentive, not just the instruction

There is a commercial reality behind this guidance that deserves stating plainly. Google's business has been built on the click. The search result that sends a visitor to your website, and the advertising that surrounds it, is the engine. A future in which a meaningful share of buyer research happens inside AI answers, on platforms Google does not own, with no click and no ad impression, is a difficult future for that model.

It is reasonable to expect Google to frame the shift as continuity rather than rupture, because continuity protects the existing business. That does not make the guidance dishonest. It makes it partial. It is the most credible possible case for the proposition that nothing has fundamentally changed, made by the party that benefits most if the market believes it. The independent evidence points elsewhere: the McKinsey research above shows owned websites now contributing a small fraction of AI search sources, with 90 to 95% coming from third-party references the business does not control.

What this means for a marketing team

The practical trap is to read Google's document, feel reassured, and conclude the existing SEO programme has the job covered. That conclusion is safe for the share of buyers who research inside Google. It leaves everyone else unaccounted for.

A marketing team assessing its real position now faces a harder question than "how do we rank on Google." The question is what every major AI platform currently says about the business, what it cites when it says it, and whether any of that aligns with what the business actually wants to be known for. Answering it means testing the platforms directly, with the questions buyers genuinely ask, and reading the results across all of them rather than one. That sits outside the day-to-day remit of most in-house teams and most traditional SEO agencies.

This is where an external perspective earns its keep. Seeing your own entity clearly across platforms you do not control is difficult from the inside, and the platforms behave differently enough that a single-surface view will mislead you. At KnownEntity.ai, the specialist AI Authority Infrastructure consultancy I co-founded, we use a structured methodology called the AI Authority Ladder™ to take businesses from poor or misappropriated visibility to appropriately and strategically cited. It begins with a diagnostic that maps exactly how AI agents currently understand a business: what they say about it, what they get wrong, where they cite competitors instead, and where the gaps sit between commercial strategy and AI presence.

A sensible first step

Google has done marketing teams a real favour by publishing this guidance, as long as it is read in full context. It confirms that the content fundamentals matter, and it reveals, by what it leaves out, exactly where the new work lies: across the platforms beyond Google, in an environment that is becoming not only answerable but, with emerging protocols like the Universal Commerce Protocol, increasingly transactable.

The honest answer to "where do we actually stand" is that almost no one knows until they look properly across every platform. That is why we built the AI Authority Snapshot: a free, fast read of how the major AI platforms currently understand and reference your business, designed to show you the real picture before you decide what to do about it.

Google has told you how to be found by Google. It is worth knowing where your buyers are actually looking, and what they are being told when they get there.

If you would like to understand where your business currently stands in AI search, and what it would take to close the gap, request your free AI Authority Snapshot at hello@knownentity.ai or visit knownentity.ai.

Last updated: 28 May 2026

Sources referenced in this article:

Find out where you stand

An AI Visibility Audit gives you a clear, actionable picture of your current AI Authority and how you compare to your competitors.

Request Your Audit