GEO advice we don't want to hear

The marketing industry has a new obsession, and it comes with a fresh set of acronyms, (GEO, AEO, etc.) and a growing ecosystem of tools, agencies, and frameworks promising to get your brand mentioned in ChatGPT.

I think most of the current conversation is getting it wrong. And the thing that actually matters, the thing that would genuinely help your brand show up in LLMs, is something marketers have known for decades but keep trying to shortcut. Build a brand that people talk about.

Before I get into why, let me address the elephant in the room.

I think SEO is still not dead

Every few years, SEO gets its obituary written. And by now, every few years we say “Every few years, SEO dies”. Social was supposed to kill it. Voice search was supposed to kill it. The amount of ads in the SERP, too. And now, LLMs are supposedly hammering the final nail in the coffin. But it isn’t that black and white.

As far as I’m concerned, as long as the default behavior on the majority of browsers is prompting a search query in a search engine, and it overwhelmingly still is, SEO exists as a channel. The plumbing of the internet hasn't changed. When someone opens Chrome, Safari, or Firefox, they get a search bar that sends them to Google. That's not something anyone will change lightly, and it's not going away tomorrow (maybe famous last words :) ).

Now, let's look at the actual data. Not to dismiss LLMs, but to put things in perspective.

Google still processes roughly 14 billion searches per day. LLM traffic is growing fast, but it remains a fraction of that with ChatGPT accounting for about 0.2% of total web referral traffic as of mid-2025, while the most generous estimates of its share of all digital queries land somewhere between 9 and 17%, depending on how broadly you define “query”.

But here's the part that most “SEO is dead” takes conveniently leave out. This isn't a zero-sum game. Google's search volume actually grew 21.64% in 2024 while ChatGPT was experiencing its explosive growth. And a Semrush study analyzing 260 billion rows of clickstream data found that when users started using ChatGPT for the first time, their Google Search activity didn't drop at all. People aren't replacing Google with ChatGPT. They're using both, for different things.

And both things can be true at once. LLMs are a meaningful and growing part of how people find information (which makes it a relevant channel for marketing) and Google is still the dominant channel by a wide margin (while still growing). The mistake isn't giving our attention to LLMs, but concluding that SEO is dying and that you need a new set of "GEO tactics" to replace it.

Linear projections into the future

Here’s a meme that I saw on Instagram the other day that I feel fits here quite well. You have a newborn. At birth, the baby weighs 3kg (6.6lbs for those across the pond). By their first birthday, they've tripled to 9kg (20lbs). If you projected linearly from there, you'd conclude your child is on track to weigh 100kg (220lbs) by age three.

Now, obviously, that's absurd. But that's exactly the logic being applied to LLM growth projections.

We saw this same pattern play out with Clubhouse, which hit 10 million users in early 2021 and was going to “replace podcasts”. We saw it with cryptocurrency adoption curves that were supposed to make fiat currency obsolete by now. Exponential early growth extrapolated into infinity, ignoring the very real forces of market saturation, user churn, novelty fatigue (read: Studio Ghibli-like images), and competitive response.

ChatGPT's app market share has already dropped from 69% in January 2025 to roughly 45% as of early 2026, according to Apptopia data. One in five AI users now uses multiple apps. Even within the AI chatbot space, we're watching the same fragmentation play out that we've seen in every technology category.

Distinguishing things that matter for SEO

Before going further, I want to make an important distinction that the GEO discourse keeps blurring. There are two separate things at play here and they require different frameworks:

1. AI Overviews on Google. - These are Google's AI-generated summaries that appear within the existing search experience. They have significantly affected the CTR (I believe we called that the great decoupling at some point). Users who saw AI summaries clicked traditional results about half as often as those who didn't. This is a Google problem, and it's an evolution of the zero-click search trend that's been building for years. It matters for SEO, but it's not "GEO." It's Google changing its own product.

2. LLMs within their own UX - ChatGPT, Claude, Gemini, Perplexity are standalone platforms where users have conversations with AI. Getting your brand mentioned here is a fundamentally different challenge from ranking on a search engine results page.

The GEO conversation often mashes these together because it's simpler to talk about them as one thing. But they work on completely different mechanics, and they shouldn’t be mixed up.

GEO isn't like SEO (not even close)

I deliberately avoid calling what happens in LLMs "ranking." Because a search engine results page is structured, hierarchical, and consistent. If you search for "best CRM software" on Google right now, and again in five minutes, you'll get essentially the same results. Position 1 is position 1. There's a clear pecking order.

LLM outputs are none of those things. Ask ChatGPT "what's the best CRM software" five times in a row and you'll get five different answers and perhaps different brands mentioned, in different orders, with different framing. The output is essentially random within a range. It varies by how you phrase the prompt, the conversation context, the model version, and what amounts to a digital coin flip.

The original GEO research paper from Princeton and Georgia Tech was based on experiments run on GPT-3.5 and GPT-4 in November 2023, one sunset model and the other one soon to be. The tactics it validated may or may not work on today's models, and there's no reliable way to verify because the underlying systems change with every update. I think that’s all normal, but we should be honest about how little we actually know.

If we want to compare appearing in LLMs to something familiar, it's not SEO. It's closer to word of mouth. You don't "rank" in a conversation between friends. You get mentioned (or you don't) based on whether you're relevant, respected, and top of mind. The same applies to an LLM putting together a response from its training data and retrieval sources.

Oversimplification of good conversion rates

But, we’ve seen ChatGPT converting better than Google. This is a massive oversimplification.

Yes, some studies have found higher conversion rates from LLM referral traffic with ChatGPT visitors converting at 15.9% versus Google organic at 1.76%. That sounds incredible until you look at the full picture. Because in others you’ll see that ChatGPT referral traffic accounted for 0.2% of total sessions, roughly 200 times less than Google organic. In conversion rate, organic search outperformed ChatGPT by 13%. In revenue per session, ChatGPT trailed every major channel except paid social.

When you're looking at a tiny number of visits from people who are already deep in a research process, of course the conversion rate looks good. These are some of the most engaged, most informed users on the internet. It's like saying “people who look at the menu order food more often than those who don’t”.

The conversion rate story is one metric taken out of context. It ignores scale, it ignores how messy attribution is across channels and it ignores that most purchase journeys. Especially when it comes to B2B.

What current GEO advice boils down to

Strip away the jargon and the GEO playbook is super short. Build authority, get mentioned by third-party sources, create clear, well-structured content, maintain consistent brand messaging across platforms and earn genuine coverage and reviews.

This should sound familiar because it’s essentially brand building. It's what good marketers have been doing since long before LLMs existed. The tactics that are supposed to help you with "GEO" are pretty much the same advice any solid marketing strategist would give you for building a respected brand.

That doesn't make the advice wrong, however I must say that it makes the rebranding somewhat unnecessary. And it means you probably don't need a separate "GEO strategy." You need a good brand strategy.

One of the more popular GEO "hacks" is flooding Reddit with brand mentions, since LLMs frequently cite Reddit discussions.

When you post on Reddit purely to insert your brand name into a relevant conversation, you're making a trade that most marketers haven't thought through. You're risking your brand's reputation among the actual humans reading those posts. And those people can spot fake recommendations instantly and will call you out in the comments. All this for the small chance that an LLM might pick up your brand mention.

You're trading real reputational damage for a “maybe” and I call that a pretty bad deal. Bonus points if you use LLMs to generate the comments on Reddit threads.

Reddit users are among the most skeptical, marketing-aware audiences on the internet. Getting caught planting fake recommendations there (and you will get caught) does more damage to your brand than any LLM mention could ever repair. The irony is that the authentic brand discussions that LLMs actually surface well are the organic ones, the ones where real users genuinely recommend your product because it solved their problem. And this is something you have to earn.

Now back to SEO

This doesn't mean SEO's role hasn't changed. It has, in important ways.

AI Overviews are compressing click-through rates. Zero-click searches continue to climb. The traffic you get from ranking #1 today is worth less than it was three years ago in raw click terms.

But what often gets lost in the GEO conversation is just how much SEO already does for brand building. It never was just at the bottom of the funnel, but at every stage of buyer awareness.

Think about the stages a buyer goes through: problem unaware, problem aware, solution aware, product aware. SEO touches all of them. A thought leadership piece helps a developer realize their CI pipeline is silently wasting hours. A deep technical guide makes your brand the trusted voice as they dig into the problem. A comparison page puts you in the consideration set. Your docs and quickstart guides earn their trust enough to adopt.

Every one of those touchpoints is a moment where a developer interacts with your brand. And all those interactions compound into the thing that actually matters for LLM visibility. Which is being a brand that developers know, trust, and recommend to each other. I honestly doubt it’s possible to skip to the end of that process.

But here's what many people are missing: brands are seeing rankings go up, traffic go down, and conversions hold steady or even increase. It’s because AI Overviews display company names. Users see your brand in the AI summary, open a new tab, and either search for you directly or enter your URL. The conversion still happens, but it’s a little bit less visible to us and doesn't show up as an SEO-driven visit in your analytics anymore.

SEO's job is evolving from “drive traffic through clicks” to ”make sure your brand is visible across all surfaces, including AI-generated ones”. In a way that’s what it always should’ve been.

So what should we do?

LLMs are trained on and pull from the internet's collective conversation about everything. If your brand is consistently mentioned in trustworthy contexts by real people who genuinely use and recommend your product, you will appear in LLM responses. Not because you optimized for it, but because you earned it.

If your brand isn't being mentioned, no amount of schema markup, prompt-friendly content, or Reddit posting will fix that. The model doesn't care about your structured data (hey, LLMs are really good at understanding unstructured data). It cares about whether the internet, real humans, real publications and real communities consider you relevant and trustworthy. So build something people want to talk about and the rest will follow. Which is to be differentiated from “build it and they will come” as it doesn’t just refer to the product.

And it's not just marketing, either

Treating LLM visibility as a marketing problem is perhaps the core of the problem. It's a product problem, a support problem, a documentation problem, a pricing problem. It's a whole-company problem.

If we take dev tools, think about what actually drives developer conversations online. It's not your blog posts. It's the developer experience itself.

Developers talk about tools that are genuinely helpful with clean API design, SDKs that work the way you'd expect, error messages that actually tell you what went wrong, onboarding that gets you to get going quickly (sometimes that’s an onboarding flow, but hey, sometimes that might be just “here’s the product”). If your product creates friction, no amount of content marketing will generate the organic praise that LLMs pick up on. On the other hand, if your DX is genuinely great, developers will write about it on their blogs, recommend it on Hacker News, answer Stack Overflow (okay maybe this one a little bit less these days) questions about it, and discuss it in general. That's the raw material LLMs are built on.

The list is endless, but the core idea is that the better the ecosystem around your company, the more likely you will be known. Is the company open-sourcing something, are your developers contributing to open-source, are they helpful publicly, do founders show up in discussions, etc. These organic interactions compound over time into an authentic presence.

The biggest lever you have for LLM visibility isn't sitting in marketing. It's in your product, your docs, your support, your developer experience and more. From the marketing perspective, our job is to champion it internally.

Earned media, genuine customer advocacy, and high-quality content are the closest thing to a GEO strategy that actually works, because they're not a GEO strategy at all. They're just good marketing. And the brands that will show up in LLMs are the same ones that have always won. The ones people genuinely respect, recommend, and talk about. There's no shortcut for that. There never was.