GEO Is Not a New Game. It's the Same Game With Worse Referees

Every week now, someone tells me that search is dead and a new discipline has replaced it. This week the phrase was GEO, generative engine optimization, sold to me as the thing I have to master before ChatGPT and Google's AI answers decide my clients no longer exist.

I read the same roundup you probably did. A judge poking holes in Google's AI publisher arguments. Another spam update leaving a trail of confused site owners. The endless GEO debate. Studies showing AI tools cite wildly different sources for the same question. And the strange new fact that ChatGPT seems to treat Reddit like the town elder who knows everything.

Here is what I did not read anywhere. A good reason to believe the fundamentals changed.

The panic has a familiar smell

I have been in this industry long enough to recognize the cycle. A new interface appears. Everyone declares the old rules obsolete. A fresh acronym gets minted, conferences get booked, and a wave of consultants start selling you a map to territory that nobody has actually surveyed yet.

We did this with mobile. We did it with voice search, which was going to change everything and then quietly did not. We did it with the metaverse, remember that. Now we are doing it with AI answers, and the tell is always the same. The louder someone insists the entire playbook is void, the more likely they are selling a new one.

The louder someone insists the entire playbook is void, the more likely they are selling a new one.

I am not saying nothing changed. Something did. The way people find answers is shifting from a list of ten blue links to a synthesized paragraph that names a few sources and buries the rest. That matters. It matters a lot for how much traffic a page can expect and how visible a brand is at the moment of decision.

But changing the delivery of the answer is not the same as changing what makes you worth citing. A hammer is not a house. The tool moved. The thing being built did not.

You cannot optimize for a referee who keeps changing the rules

Here is the detail from this week's news that should have gotten more attention than the acronym did. AI tools cite different sources for the same query. Ask three of them the same question and you can get three different sets of names. Ask the same tool the same question next week and the list moves again.

Sit with that for a second. The whole pitch of GEO is that there is a formula to get named by the machine. But the machine has no stable opinion. It is a referee who calls a foul on Monday and waves off the identical play on Thursday.

So when someone hands you a checklist of tactics guaranteed to get you cited by ChatGPT, ask the obvious question. Compared to what? Measured against which model, on which day, under which version that quietly shipped overnight with no changelog?

You are being sold precision inside a system that has none. That is not strategy. That is superstition with a slide deck.

You are being sold precision inside a system that has none. That is not strategy. It is superstition with a slide deck.

I am not being cynical for sport. I want my clients recommended by these tools. We work on it. But I refuse to build a house on a floor that moves every time the vendor pushes an update, and I refuse to charge someone for the illusion that I can nail down something the platform itself has not decided.

The Reddit thing tells you what these systems actually reward

Now, the ChatGPT and Reddit behavior. This is the part everyone treated as a curiosity, and I think it is the most useful signal in the whole roundup.

Why would a language model lean so hard on Reddit? Because Reddit is full of real people arguing, correcting each other, and describing actual experiences in plain language. It is messy, it is opinionated, and it is unmistakably human. The model is not rewarding polish. It is rewarding evidence that a real person with real experience weighed in.

Read that back and tell me it sounds like a new discipline. It sounds like the oldest rule we have. Be the source that other people quote when the stakes are real.

For years I have told clients that the goal is not to rank. The goal is to become the reference. To be the company whose explanation of a problem is so clear that competitors accidentally borrow your framing. To be the name that comes up in the group chat, the forum thread, the Slack channel where a buyer asks who actually knows this stuff.

The AI answer is just a new place where that reputation gets cashed in. The machine is reading the room. If the room already respects you, you show up. If it does not, no amount of schema markup will fake it into existence.

The spam update fallout is the same lesson wearing a costume

Every spam update produces the same two camps. One group of site owners is baffled and furious because their traffic fell off a cliff. Another group barely notices. The gap between them is rarely a mystery once you look.

The sites that get hammered tend to have built for the algorithm instead of the reader. Thin pages spun up to catch a keyword. Content produced at volume because volume was the cheat code that worked last quarter. Then the rules tighten, the shortcut closes, and the whole structure collapses because it was never load bearing to begin with.

The sites that shrug it off usually share one boring trait. A real person would actually want to read the thing.

That is the connective tissue between the spam story and the AI story, and almost nobody framed it that way. Google tightening its spam net and ChatGPT leaning on Reddit are two versions of the same instinct. Both are trying, imperfectly, to separate genuine human usefulness from manufactured noise. The methods differ. The target is identical.

Google tightening its spam net and ChatGPT leaning on Reddit are two versions of the same instinct.

So if you spent the last year flooding the internet with mediocre AI written pages to win at search, you are about to lose twice. Once when the spam update prices you out, and again when the AI answers refuse to cite content that reads like it was assembled by the same kind of machine doing the citing.

The publisher fight is the real story, and it is not yours to solve

Then there is the judge questioning Google's AI arguments in the publisher dispute. This is the one thread in the whole roundup that genuinely could reshape the terrain, and it has almost nothing to do with your marketing tactics.

The unresolved question underneath all of this is brutal and simple. If AI answers are built on publishers' work but send those publishers less traffic in return, what happens to the people who make the answers worth having? Nobody has settled that. Not the courts, not the platforms, not the publishers watching their referral numbers slide.

I bring it up because it should change your posture, not your checklist. When the incentive structure of the open web is being actively fought over in a courtroom, the smart move is not to bet your entire business on one channel whose economics might be rewritten by a ruling you will never be consulted about.

Own something the platforms cannot revoke. Your email list. Your reputation. The relationships that make people seek you out by name rather than by query. These are the assets that survive whichever way the judge leans, and they are precisely the ones the GEO gurus never mention, because you cannot sell a monthly retainer against advice that sounds like patience.

So what would I actually do

Let me be useful instead of just skeptical, because skepticism without a better answer is just complaining.

I would stop treating GEO as a separate budget line and start treating it as the natural result of doing the real work well. If I had to put it plainly, here is where the effort actually pays:

• Write things a genuine expert would put their name on, on topics where you have real experience nobody can copy. That is what both the spam filters and the AI models are increasingly built to find.
• Get quoted by humans in the places humans trust. Forums, communities, real reviews, real conversations. If people cite you, the machines eventually will too, because they are reading over those people's shoulders.
• Build direct relationships you control, so that no ruling and no model update can quietly disconnect you from your own audience.

Notice that not one of those is a trick. Notice that all three would have been good advice five years ago and will still be good advice five years from now. That is the point. Durable strategy looks boring precisely because it does not need a new acronym every quarter to justify its existence.

The new game is the old game, played honestly

I understand why GEO caught on. It gives a frightening moment a name, and naming a fear makes it feel manageable. When your traffic charts wobble and a machine starts answering the questions your website used to answer, you want a lever to pull. GEO promises the lever.

But the news this week did not describe a new set of rules. It described a set of systems, some legal, some algorithmic, some conversational, all lurching in the same rough direction. Toward rewarding what is genuinely useful and human, and away from what is manufactured to game a formula. They are doing it clumsily. They contradict themselves. They cite Reddit on Tuesday and forget it on Wednesday.

The referees got worse. The game did not change.

Stop chasing the referee. Play so well that even a bad one has to acknowledge you.