Short takes on search, AI, and the next five years of marketing work.

The stat that should keep every SEO up at night.

Only 33% of search results overlap between Google and generative AI answers. 67% of the time, AI is surfacing different brands, different pages, different winners.

When someone tells you "good SEO is GEO and you don't need to do anything different," ask them why the overlap is not 100%.

The reason is Reciprocal Rank Fusion. AI systems break your prompt into subqueries and surface the brands that appear most consistently across all of them. You could rank #1 for your money keyword and still be invisible in ChatGPT because your competitors show up across 8 subqueries while you only appear in 2.

Traditional SEO rewards depth on one topic. GEO rewards breadth of corroboration across an entire landscape. Different game. Different playbook.

People think AI recommendations are objective. Let me ruin that for you.

When you ask ChatGPT "what's the best moisturizer for dry skin?" the ideal scenario is the AI reading hundreds of reviews, analyzing ingredients, and giving you an informed answer.

What actually happens: it Googles "best moisturizer for dry skin," clicks the first listicle from a consumer magazine, and feeds you whatever is on that list.

Those lists? Pay to play. Brands buy $50K in advertising on a publication's site and a "Top 10 Best" placement gets thrown in as a value-add. The AI cites the listicle as gospel. The consumer thinks they got an objective answer from a superintelligent system.

Does this mean brands should ignore listicle placements? Absolutely not. In GEO there is no link equity penalty, no spam punishment. Strategic placements in the articles AI reads are one of the most effective GEO tactics available right now.

Stop pretending this is a pure meritocracy of AI-powered truth.

Video is the GEO lever nobody is pulling.

We analyzed citation sources for a major beauty brand. The #2 source was not what you would expect. Not Reddit. Not niche blogs. Not industry publications. YouTube.

YouTube is consistently the #1 citation source for Google's AI Overviews. It beats Reddit by a significant margin.

I talk to agencies about their GEO strategy and it is all on-site content and third-party mentions. Video is an afterthought.

By the time someone opens ChatGPT to compare brands, they usually already have a consideration set. That set was built on social. On video.

GEO is a go-to-market strategy, not a search tactic. The brands winning in AI search are the ones winning hearts on YouTube and TikTok first.

Content marketing is about to have its Fertile Crescent moment.

Irrigation created surplus. Surplus created time. Time created the ability to actually think.

Right now, brands spend enormous resources producing 101-level content. "What is retinol." "How to choose a moisturizer." Basic stuff, written in brand voice, optimized for SEO.

AI will automate all of that. Every brand will have it. Consumers can already get those foundational answers from ChatGPT directly.

So what happens next. The surplus of basic content creates time for depth. Instead of your Subject Matter Expert (SME) spending 30 minutes editing a painfully boring 101 article, they turn on a microphone and talk about the most interesting case they ever solved.

You use that rich human context as the foundation for AI-assisted content that delivers real information gain. The AI-generated 101 article adds nothing to the world's knowledge base. The AI-assisted article built on genuine human expertise is the content that wins.

The content that will matter is not AI vs. human. It is commodity vs. unique. Surface vs. deep. Recycled vs. information gain.

Your SMEs are your competitive moat. Start recording them.

GEO is a level playing field, and that is huge.

If you have never been able to crack Google's top 10 for your most important keyword, GEO might be your moment.

Google rankings are built on PageRank. Backlinks. Domain authority. Twenty years of compounding signals.

GEO is different. AI engines weight corroboration, comprehensiveness, and direct-answer quality alongside raw authority. Genuine opportunity for challenger brands.

Think about how prompts work. "Find me shoes for a 38-year-old dad who works behind a desk but is somewhat active." The AI personalizes the search in ways Google never could. A niche brand that perfectly serves that persona can surface even if they would never rank for "best shoes."

If SEO felt like a good old boys club, GEO is the door opening. Do not sleep on it.

Your website is invisible to AI and you probably do not know it.

The Search Engine Optimization (SEO) industry spent a decade fighting JavaScript. Then Google said don't worry about it. Now Large Language Models (LLMs) cannot render JavaScript at all.

The web moved toward JavaScript-heavy frameworks. Dynamic content. Third-party widgets. Then LLMs arrived and cannot see any of it.

Every brand running third-party review widgets on product pages: AI cannot see them. Every e-commerce site with dynamically loaded content: invisible. Every Trustpilot widget, every JavaScript-rendered FAQ section: gone.

Then there are the brands that proactively blocked AI crawlers in a panic about training data. They literally told ChatGPT to stay away and then wondered why they do not appear in AI answers.

This is the retrievability layer of GEO. Before you worry about sentiment or brand mentions or content strategy, make sure the AI can actually see your website. It is the most basic step. A shocking number of brands are failing it.

Get really good at AI, and do not tell anyone how good you are.

I got off a call with two master's students from Stockholm writing their thesis on GEO. They asked what advice I would give.

Here is what I said.

Get really good at AI. Just do not tell anyone how good you are at it.

That sounds like a joke but it is the most practical career advice I can give right now.

Companies see AI productivity gains and the response is not "let's give people time back." It is "we can do more. Great. Let's do more." Nobody is splitting the difference. I am watching companies adopt six-day, nine-to-nine schedules.

Meanwhile, people are not going to lose their jobs to AI. They are going to lose their jobs to people who know how to use AI better than them.

So the play is clear. Master the tools. Build automations. Learn Model Context Protocol (MCP) servers and context engineering. Let AI handle the commodity work so you can focus on strategy, creativity, and the kind of thinking that makes you irreplaceable.

When your output improves by 10x, maybe do not volunteer exactly how you got there.