Google has done what Google always does.
It has wandered into a room full of anxious search marketers, handed everyone a cup of herbal tea, told them the house is definitely not on fire, and explained that the smoke is probably just seasonal warmth from the algorithm.
The document, an AI optimisation guide, sits on Google’s Search developers site. It talks about AI search, generative engine optimisation, answer engine optimisation, structured data, chunking, helpful content, crawlability, mentions, and all the other little baubles currently being rattled around the SEO industry’s Christmas sack.
On the surface, it looks calm. Helpful. Reasonable. Very Google.
Underneath, it is something else entirely.
It is a strategic positioning piece from the most conflicted company in search.
It tells publishers, business owners and SEOs to keep creating the expensive stuff… expert content, fresh knowledge, useful material, authentic mentions, and more. While Google continues to build interfaces that reduce the need for anyone to visit the original source.
And then, with the straightest face in Silicon Valley, it calls the whole thing “still SEO”.
Lovely.
No, Google, GEO is not “just SEO”
Let’s start with the opening act.
Google says:
“What about ‘AEO’ and ‘GEO’? ‘AEO’ stands for ‘answer engine optimization’ and ‘GEO’ for ‘generative engine optimization’. These are both terms you may see used to describe work specifically focused on improving visibility in AI search experiences. From Google Search’s perspective, optimizing for generative AI search is optimizing for the search experience, and thus still SEO.”
This is the first sleight of hand.
“From Google Search’s perspective” is doing an awful lot of work there. From Google’s perspective, everything is Google. The sky is Google. The buyer is Google. The internet is Google. Search is Google. And therefore anything that happens in the broad world of AI discovery can be neatly folded back into the nice old acronym that keeps everybody calm: SEO.
But ChatGPT is not Google. Claude is not Google. Perplexity is not Google. These are not conventional search engines in the old sense. They do not simply return a ranked list of blue links and ask the user to go rummaging through them like a pensioner at a car boot sale.
They synthesise. They compare. They recommend. They summarise. They often satisfy the user before a click ever happens.
That doesn’t sound like SEO.
SEO is primarily about improving the visibility of a website or web page in search results. GEO is about increasing the likelihood that your brand is found, understood, mentioned and recommended inside generated answers when buyers ask intent-rich questions.
Those are not the same job.
They overlap, of course. Nobody serious should deny that. A crawlable, useful, authoritative website still matters. Ranking in Google still matters. If you are a vet in Manchester, you probably still want to rank for “vet in Manchester”. Fine.
But saying GEO is “just SEO” is like Blockbuster saying Netflix is “just DVD rental” because both involved watching films. Technically adjacent. Strategically stupid.
The buyer’s behaviour is changing. In B2B especially, people are already asking AI tools to compare vendors, explain categories, shortlist solutions, write buying criteria, summarise reviews and generate options. They are not always going to Google first. They are using Google and ChatGPT. Google and Claude. Google and Perplexity.
This is not an either/or world. It is a repertoire world.
I shop at Aldi. I shop at Tesco. I shop at M&S. I occasionally wander into Sainsbury’s and wonder what the hell happened to the prices. Consumers do not use one channel because a platform would prefer a monopoly. They use whatever gets the job done.
Google still has immense power. But it no longer owns every search moment. That is the point. And that is exactly why pretending GEO is “still SEO” is such a fragile position to take.
Google’s “opportunity” is a downgrade dressed in lipstick
Then comes this little beauty:
“As we upgrade Search to meet these changing expectations, this transformation offers new opportunities to reach people who may be more inclined to engage with your site, spend more time with your content, or even convert by becoming a subscriber or making a purchase.”
Read that again, slowly.
“May be more inclined.”
This is the language of a platform changing the deal while smiling politely through the knife work.
Google is telling publishers and businesses that AI search will create wonderful new opportunities to reach people who might engage more deeply, spend more time, subscribe or buy. And perhaps, sometimes, that will happen. A user who clicks after reading an AI-generated overview may indeed be more qualified. There will be cases where fewer clicks produce better clicks.
But as a general claim, it conveniently ignores the obvious mechanism of AI search: it reduces the need to click.
If Google takes your expertise, synthesises your information, answers the query and keeps the user inside Google’s interface, the original publisher has not gained a beautiful new engagement opportunity. It has been disintermediated.
That is the commercial shift.
For years, businesses used SEO because they did not want to pay Google for every commercial interaction. Organic rankings reduced dependency on AdWords. That was the point. Rank for buyer-intent searches and you could get leads, sales, attention and trust without feeding the paid search machine every time someone breathed near your category.
SEO damaged Google’s perfect economic model. That is why Google has always tolerated the SEO industry more than loved it.
At one end, black-hat SEOs created signal noise so extreme Google had to keep building defences through updates. At the other end, legitimate SEOs helped businesses win visibility that reduced paid dependency. Either way, SEO was never Google’s favourite child. It was the weird cousin invited to Christmas because it had to be.
AI search changes the economic bargain again.
Google can now use publisher content to answer the query, keep the user, protect the interface and maybe send a click afterwards. Maybe. If the user needs more. If the source is visible. If curiosity survives the answer.
Google is telling publishers there may be more opportunity to engage users after Google has eaten the answer, digested the answer, served the answer and kept the user at its own table.
Perhaps the publisher can come in later and wash the plate.
The obsession with “content” tells you everything
The whole guide is obsessed with content.
That word appears everywhere. Content. Useful content. Non-commodity content. Content for your audience. Content Google can discover. Content Google can process. Content Google can use.
This is not accidental.
It also explains why the piece reads like Danny Sullivan in developer-documentation form. Sullivan’s public line has long carried the same broad simplicity: make helpful content. Create useful things for people. Don’t overthink it. Don’t chase tricks.
There is truth in that. But only enough truth to make the simplification dangerous.
SEO is not “just creating helpful content”. Anyone who has ever had to rank a commercial website in a competitive category knows that. SEO is technical, architectural, reputational, editorial, commercial, analytical and often brutally competitive. Crawlability matters. Internal linking matters. Page templates matter. Authority matters. Intent matters. Category structure matters. Merchant feeds matter. Product data matters. Local signals matter. Reviews matter. Links matter. Brand matters.
Saying SEO is just helpful content is the kind of sentence that sounds wise on a conference stage and collapses the minute it meets a real e-commerce site with 80,000 SKUs, broken faceted navigation and a competitor buying every link in sight.
But this reduction is useful for Google. If SEO becomes “just make helpful content”, then Google can quietly strip the discipline down to something any business owner supposedly ought to do themselves.
And when that does not work?
Well, there is always paid search.
The machine has eaten the listicle. Now it wants the expert.
The most revealing advice in the guide is this:
“Create valuable, non-commodity content for your audience Creating content that people find unique, compelling, and useful will likely influence your website’s presence in generative AI search in the long run more than any of the other suggestions in this guide.”
Google then gives this example:
“Commodity content (for example, something like ‘7 Tips for First-Time Homebuyers’) is often based on common knowledge, which could originate from anyone, and typically adds little unique insight for readers. In contrast, non-commodity content (such as ‘Why We Waived the Inspection & Saved Money: A Look Inside the Sewer Line’) provides unique expert or experienced takes that go beyond common knowledge and the ordinary.”
At this point, Google is either being very clever or taking the piss. Possibly both.
Let’s be clear. There is nothing wrong with non-commodity content. Quite the opposite. It is usually the only content worth creating.
A restaurant that shows its dishes and explains its sourcing is useful. An accountant explaining a tax problem for a specific kind of client is useful. A SaaS company showing how a real customer solved a painful workflow issue is useful. A consultant writing a piece like this to demonstrate how they think is useful.
Good content is advertising. Good content is evidence. Good content is sales enablement. Good content is memory creation. Good content helps buyers understand whether you know what you are talking about.
The problem is not that Google recommends useful content.
The problem is the economic context in which Google recommends it.
Google does not need more “7 tips for first-time homebuyers” articles. The machine already ate that sludge years ago. It has consumed the listicles, the beginner guides, the definitions, the “what is” pages and the 1,200-word SEO porridge that content teams have been ladling onto the internet since 2012.
What it needs now is fresh expertise. Original experience. Case studies. Opinions. Data. Edge cases. Awkward stories. Commercial nuance. Things only real experts know because they have actually done the work.
In other words, Google is saying: please stop giving us the cheap stuff we can already summarise. Please give us the expensive stuff we need to keep the answer machine fresh.
And here is the rub. Non-commodity content is often harder to search for before it exists in the buyer’s mind. A keyword like “first-time homebuyer tips” has volume. You can see it. You can brief it. You can build a content calendar around it and pretend the traffic is a strategy.
But “why we waived the inspection and saved money after checking the sewer line” is not a neat keyword strategy. It is judgement. Experience. Editorial instinct. Sector knowledge. A story. A proof point.
Most SEOs cannot manufacture that with a keyword tool and a Surfer login between 9am and 5pm. That is not an insult. It is a category distinction.
That work belongs to subject-matter experts, copywriters, strategists, PR people, content marketers, founders, sales teams and customers. The SEO might help surface the opportunity, structure the page, make it discoverable and measure outcomes. But the expertise is not theirs.
So when Google tells the SEO industry to create non-commodity content, it is not handing them a playbook. It is exposing the limits of their current product.
The machine has eaten the listicle. Now it wants the expert. And they will also eat and synthesise that too. You won’t win by giving Google what they want.
That doesn’t mean don’t create content, I think you should. What it means is: don’t expect traffic to that content from Google… ever.
Crawlability is not strategy. It is table stakes.
Next, Google moves into technical structure:
“Build and maintain a clear technical structure The way Google Search finds and processes your pages remains the core of how our AI systems access your data. Technical clarity ensures your content is ready for discovery and indexing, and all existing technical SEO best practices continue to be worthwhile.”
This is true.
It is also basic.
Yes, make your site crawlable. Yes, avoid burying all your important information inside a JavaScript swamp created by an AI website builder with the structural integrity of wet cardboard. Yes, make sure Google can access, render and understand your pages. Yes, technical SEO still matters.
But telling a business to make its website crawlable is like telling a restaurant to have a front door.
Sensible, yes. Strategy, no.
The more interesting part is what this reveals about Google’s own dependency. Google’s AI systems need accessible data. They need pages they can find and process. They need the web to remain legible because their answer machine relies on the continued production and availability of human information.
This is why the whole document keeps returning to the same quiet request: please make things we can use.
Please create useful content. Please create non-commodity content. Please keep your site technically accessible. Please do not hide your material from us. Please let the machine eat.
And, again, Google’s worldview is narrower than the market’s. Plenty of businesses grow with very little dependence on their website. They grow through Facebook pages, TikTok, Instagram, YouTube, communities, referrals, partnerships, marketplaces, reviews, local reputation and paid social. There are businesses making serious money whose “website strategy” is little more than a digital business card and a booking link.
Google does not like that world because Google cannot index all of it neatly. But marketers have to live in it.
Crawlability matters. But it is not demand. It is not preference. It is not positioning. It is not proof. It is the price of entry.
A technically perfect page that nobody trusts, nobody cites, nobody talks about, and nobody chooses is just a well-structured corpse.
The GEO hacks are mostly bollocks too
Now, to be fair to Google, the guide is not wrong about everything.
In fact, some of its best advice is aimed at the emerging GEO grift market. And there is plenty of grift.
Google says:
“LLMS.txt files and other ‘special’ markup: You don’t need to create new machine readable files, AI text files, markup, or Markdown to appear in generative AI search. Note that Google may discover, crawl, and index many kinds of files in addition to HTML on a website: this doesn’t mean that the file is treated in a special way.”
Correct.
The idea that an llms.txt file is going to make a serious generative system recommend your brand is wishful technical theatre. It might be useful in narrow contexts for certain tools. It might help some systems understand where to look. Fine. But if brand recommendation in AI search could be won by uploading a polite little text file, every half-competent spammer on earth would have colonised the channel by Thursday.
Generative engines do not recommend brands because they found a Markdown note in the lobby.
They recommend brands because enough accessible, credible, corroborated evidence points in that direction.
The same goes for structured data. Google says:
“Overfocusing on structured data: Structured data isn’t required for generative AI search, and there’s no special schema.org markup you need to add. However, it’s a good idea to continue using it as part of your overall SEO strategy, as it helps with being eligible for rich results on Google Search.”
Again, correct.
Structured data is useful for housekeeping. In e-commerce, local search, recipes, events, reviews and rich results, it can help. Use it where it makes sense. Keep your product data clean. Mark up reviews properly. Make entities easier to understand.
But structured data is not going to save an unremarkable brand.
A five-star review snippet may help a human click a search result. But the underlying reviews are the real signal. The customer experience is the real signal. The third-party corroboration is the real signal.
Schema is the label on the tin. The food is inside.
This matters because structured data is one of the SEO industry’s favourite deliverables. It feels technical. It can be audited. It can be implemented. It can be reported. It gives the client something tangible.
But GEO will not be won by adding schema and sending an invoice.
The machine can read. That is rather the point.
Chunking is just clear writing with a worse name
Google also takes aim at chunking:
“‘Chunking’ content: There’s no requirement to break your content into tiny pieces for AI to better understand it. Google systems are able to understand the nuance of multiple topics on a page and show the relevant piece to users. However, sometimes shorter (or longer!) pages can work well depending on your audience and subject matter. There’s no ideal page length, and in the end, make pages for your audience, not just for generative AI search.”
Again, broadly right.
The SEO influencer class discovered vector embeddings, cosine similarity, retrieval and chunking, then immediately tried to sell paragraphing back to clients as a new discipline.
Chunking is not the future of AI search. It is what competent writers used to call paragraphs before the industry swallowed a machine-learning glossary.
But there is a useful idea hiding under the nonsense.
A good commercial page is made of information packages. A headline. A proposition. A proof point. A benefit. A specification. A comparison. A testimonial. An objection handled. A call to action. Each package should carry meaning. Each should help the buyer move forward. Each should earn its place on the page.
That is not revolutionary. It is basic copywriting.
But it is useful to audit a page this way. Look at each package of information and ask: does this help? Does it reduce uncertainty? Does it add proof? Does it improve understanding? Does it make the brand more credible? Does it answer a real buyer question? Does it increase information gain relative to the effort required to consume it?
That is a much more useful conversation than “make your page 2,000 words because the competitor has 1,700”.
The old SEO obsession with word count was always crude. More words did not automatically mean more authority. It often meant more sludge. A product page should not read like War and Peace. A service page should be as long as it needs to be to do its commercial job and no longer.
So Google is right that you do not need to dice your content into tiny pieces for AI.
But marketers are also right to think carefully about the information packages on their pages. Not because AI needs baby food. Because humans scan, interpret and decide in pieces too.
Mentions are the new links, and Google accidentally admitted it
The most important part of Google’s guide is buried in a section about inauthentic mentions:
“Seeking inauthentic ‘mentions’: Just like the rest of Google Search, our generative AI features can show what’s being said about products and services across the web, including in blogs, videos, and forum discussions. However, seeking inauthentic ‘mentions’ across the web isn’t as helpful as it might seem. Our core ranking systems focus on high-quality content while other systems block spam; our generative AI features depend on both.”
This is the tell.
If Google is warning against inauthentic mentions, mentions matter.
Nobody warns you not to counterfeit Monopoly money in a casino. They warn you about fake money because real money has value.
This is where GEO becomes clearly different from old SEO. It is not only about what you publish on your own website. It is about what the network says about you.
You can claim to be the best CRM for dentists, the leading accountancy firm for contractors, the most trusted roofing company in Leeds or the smartest B2B content agency in Britain. Wonderful. Your website says you are brilliant. What a surprise.
But when reviews, customers, comparison pages, forum discussions, trade publications, YouTube videos, analysts, partners, podcasts and category experts start saying similar things, that becomes evidence.
Generative engines need evidence.
They need corroboration. They need patterns. They need co-occurrence. They need to understand that your brand is repeatedly associated with a category, a problem, a buyer situation and a set of performance attributes.
This is why fake mentions are already being sold. Of course they are. The spam economy moves faster than the conference circuit. There are already services offering injections of synthetic mentions across Reddit, forums, comments, social platforms and God knows where else.
Do not misunderstand me. I am not recommending that.
It is grubby, risky and likely to become increasingly detectable. It is also the natural successor to the link-buying hypocrisy that has defined SEO for years. The same people standing on stages talking about E-E-A-T, helpful content and white-hat purity often had link vendors quietly doing the dirty work in the background.
The white-hat sermon in public. The guest-post farm in private.
Now the link economy is mutating into the mention economy. Fake Reddit threads. Synthetic comments. Paid inclusions. Manufactured social proof. “Community activation” that looks suspiciously like spam with a nicer deck.
But the dark art proves the importance of the signal.
Authentic mentions are central to GEO. The goal is not to spray fake brand noise across the internet. The goal is to earn, build and amplify credible evidence that your brand deserves to be recommended.
That is PR. That is reputation. That is customer experience. That is brand. That is search marketing. That is not “just write helpful content”.
Google’s guide quietly undermines the SEO industry sharing it
Here is the funny bit.
Many SEO consultants have been sharing Google’s guide as proof that GEO is just SEO. But read the guide closely, and it does not defend many of the SEO industry’s favourite deliverables.
No special markup. No magic schema. No llms.txt. No chunking requirement. No inauthentic mentions. No AI-specific technical charm.
Just create non-commodity content and make sure Google can access it.
If I were an SEO agency spouting “it’s just SEO”, I would be a little more nervous about celebrating that.
Because the obvious client question becomes: what exactly are you doing differently?
There are different types of SEO professionals, and we should not lump them all together.
Some SEOs do technical hygiene. Audits, crawl analysis, indexation, templates, performance, migrations, and architecture. Useful work. A bit like servicing a car. Necessary, often valuable, and still relevant.
Some SEOs do e-commerce and product search. Merchant Centre, feeds, category architecture, product data, marketplace visibility, and commercial search surfaces. Still useful.
Some SEOs have built their businesses around traffic generation: keyword universes, topical maps, content calendars, topical authority and large-scale publishing. More vulnerable.
And then there is the SEO influencer class, which sells strategic ambiguity as a service. These are the people saying “SEO is GEO” while never quite explaining what they will actually do between 9am and 5pm to make a brand more likely to be recommended in ChatGPT, Claude, Perplexity or Google AI Mode.
That is positioning, not analysis.
And it is defensive positioning.
If your main message is “nothing has changed”, you are not describing the market. You are protecting your invoice.
GEO starts with positioning, not prompts
So what should businesses actually do?
Start where good marketing always starts: positioning.
Not schema. Not an llms.txt file. Not a 38-page AI content calendar full of reheated blog posts. Positioning.
What do you want to be known for? Which buyers do you serve? What problem do you solve? What alternatives are you compared against? Why should you be chosen? What proof supports that claim? Where does that proof appear? Can humans and machines both find it? Is the market saying the same thing you are saying about yourself?
Generative engine optimisation starts with these questions because AI recommendation is not just retrieval. It is selection.
You are trying to become the plausible answer.
That means the work is not simply optimising for the interface. You cannot just optimise for Google’s SERP, ChatGPT’s answer, Claude’s response or Perplexity’s citations as separate little boxes.
You have to optimise for the internet.
That means building visibility, credibility and association across the network that these systems use to understand the world.
Your website matters. But so do third-party articles, reviews, YouTube videos, podcast appearances, customer stories, comparison pages, Reddit discussions, niche forums, directories, partner pages, social content, analyst mentions, case studies and expert commentary.
The job is to increase the co-occurrence between your brand and the things you want to be known for.
Brand plus category. Brand plus problem. Brand plus buyer situation. Brand plus competitor comparison. Brand plus proof. Brand plus performance attribute. Brand plus location, if relevant. Brand plus use case. Brand plus trust signal.
That is GEO.
Not because co-occurrence is a shiny new trick. Because this is how reputation works. Humans understand brands through repeated associations. Machines increasingly do the same.
If you are an agency, your positioning matters too. If you tell the market “GEO is just SEO”, that is your positioning. Fine. Own it. But understand the risk. You are telling clients that you will handle a new discovery environment with the old deliverables, even as Google’s own guide tells them not to overfocus on many of those deliverables.
My position is different.
I help brands become more visible, credible and recommendable across the internet for the buying situations that matter.
That is a boardroom sentence. That is a commercial sentence. That connects to revenue, not just rankings.
Good search marketers will win. Lazy SEOs will not.
Weak SEO labour is in trouble.
AI can already generate content briefs, keyword clusters, technical checks, schema suggestions, competitor summaries, redirect maps, internal linking ideas, and draft audits faster than many teams can. Not perfectly. Not without supervision. But fast enough to change the economics of the work.
A lot of low-level SEO output was always just organised labour dressed up as strategy. AI will devour that.
But good search marketers will become more valuable.
The people who can diagnose commercial problems, understand buyer behaviour, interpret data, separate traffic from revenue, prioritise technical work, extract expertise, build digital PR systems, improve product pages, understand category entry points, measure brand visibility, and connect organic discovery to business outcomes will do very well.
Because that is not just SEO.
That is search marketing. Or brand visibility. Or demand discovery. The label matters less than the job.
And the job is becoming clearer: help brands get found, understood, trusted and chosen wherever buyers ask questions.
Stop taking strategic advice from conflicted platforms
The most dangerous thing a marketer can do is take strategic advice from a platform without first asking what the platform wants.
Google is not stupid. Google knows more about search than anyone. Its guide contains sensible technical advice. It is right about plenty of GEO hacks being nonsense. It is right that useful, original, expert content matters. It is right that crawlability matters. It is right that spammy mentions are a bad idea.
But Google is not neutral.
It wants a web it can crawl. It wants fresh expert content it can process. It wants fewer spam signals. It wants publishers to keep feeding the machine. It wants businesses to believe organic visibility is still worth chasing while its own answer interfaces absorb more of the user journey. And, when uncertainty rises and organic clicks become less predictable, it will not be heartbroken if more businesses decide paid search looks reassuringly measurable again.
That is not a conspiracy. That is the reality we face.
The SEO industry has its own incentives too. Many consultants want GEO to be “just SEO” because it protects their positioning, retainers, and existing deliverables. That does not make them evil. It makes them human. But it also means their advice should be interrogated.
The market does not need more soothing ambiguity.
It needs clearer thinking.
Generative engine optimisation is not a magic text file. It is not schema. It is not chunking. It is not content sludge with an AI headline. And it is not simply SEO with a new badge slapped on the lanyard.
It is the discipline of making a brand more likely to be discovered, understood, trusted and recommended in a world where buyers increasingly ask machines for advice.
That starts with positioning. It grows through evidence. It compounds through visibility. It depends on what the market says about you, not just what you say about yourself.
So by all means, read Google’s guide. Some of it is useful. Some of it is obvious. Some of it is self-serving. And some of it accidentally tells you where the future is going.
But do not mistake Google’s incentives for your strategy.
Do not optimise for Google. Do not optimise for ChatGPT. Do not optimise for whatever acronym the SEO industry is trying to sell this week.
Optimise to be recommended by AI. And that’s optimising your presence on the internet, not the Google interface.
That is where the money is.
Andrew Holland
Originally published on LinkedIn on 19 May 2026.